Interpreting Quantum Mechanics

in Terms of Random Discontinuous

Motion of Particles
by

Shan Gao

THESIS
submitted for the degree of
Doctor of Philosophy

Unit for History and Philosophy of Science Centre for Time, SOPHI
Faculty of Science Department of Philosophy

Supervisors: Dr. Dean Rickles

Prof. Huw Price

University of Sydney
Sydney, Australia
March 2012
 c Shan Gao 2011
All rights reserved

This thesis is copyright. Subject to statutory exception and to provisions of relevant collective
licensing agreements, no part of this publication may be reproduced, stored in a retrieval
system, or transmitted in any form or by any means, without the prior written permission
of the author.
 This thesis is dedicated to Erwin Schrödinger, who introduced the wave function,
discovered its equation named after him, and argued that quantum mechanics is in-
complete by his famous cat paradox.
 Abstract

This thesis is an attempt to reconstruct the conceptual foundations of quantum mechanics.

First, we argue that the wave function in quantum mechanics is a description of random
discontinuous motion of particles, and the modulus square of the wave function gives the
probability density of the particles being in certain locations in space. Next, we show that
the linear non-relativistic evolution of the wave function of an isolated system obeys the
free Schrödinger equation due to the requirements of spacetime translation invariance and
relativistic invariance. Thirdly, we argue that the random discontinuous motion of particles
may lead to a stochastic, nonlinear collapse evolution of the wave function. A discrete
model of energy-conserved wavefunction collapse is proposed and shown to be consistent
with existing experiments and our macroscopic experience. In addition, we also give a
critical analysis of the de Broglie-Bohm theory, the many-worlds interpretation and other
dynamical collapse theories, and briefly analyze the problem of the incompatibility between
quantum mechanics and special relativity.

4
 Acknowledgement

The idea of random discontinuous motion of particles came to my mind when I was a post-
graduate at the Institute of Electronics, Chinese Academy of Sciences in 1993. I am happy
that finally it has a more logical and satisfying formulation in this PhD thesis. During the
writing of the thesis, I have been supported financially by a University of Sydney Interna-
tional Scholarship and a Postgraduate Scholarship in Quantum Foundations provided by the
Unit for HPS and Centre for Time (SOPHI), as well as by two Lucy Firth Scholarships in
Philosophy provided by the Department of Philosophy.
I would like to express my deepest gratitude to my supervisors, Dean Rickles and Huw
Price, whose understanding, motivation and personal guidance was indispensable during my
PhD studies. I am deeply indebted to them for their continuous encouragement, stimulating
suggestions and healthy criticism which helped me greatly during my research and writing
of this thesis. Their professional approach and accuracy was particularly beneficial to my
study.
I wish to express my warm and sincere thanks to the head of the Unit for HPS, Ofer
Gal, the postgraduate coordinator, Hans Pols, and Debbie Castle. I am grateful for their
personal support and for their reassurance regarding technical conditions during my PhD
work. I would like to take this opportunity to thank all the other members of the Unit for
their help and encouragement throughout the years.
I would like to gratefully thank David Miller, Rod Sutherland, Hans Westman and Owen
Maroney for helpful discussions and useful suggestions. I owe a debt of gratitude to all the
PhD students here who I shared good times and had fun with.
My deepest gratitude goes to my parents, Qingfeng Gao and Lihua Zhao, who spared
no effort to provide the best possible environment for me to grow up in and their constant
support during my extended studies. They have always allowed me to have the freedom of
choice, and for that I am truly grateful. I am also thankful to my brother, Yuan Gao, and to
all my friends, as they have always been willing to share their thoughts with me and provide
alternative viewpoints which have helped me think in a more holistic fashion.
Finally, I am deeply indebted to my wife Huixia and my daughter Ruiqi for their un-
flagging love and support throughout my study; this dissertation would have been simply
impossible without them. Moreover, they have never let me forget the true values of life.

5
 Publications

1. On Diósi-Penrose criterion of gravity-induced quantum collapse, International Journal of

Theoretical Physics, 49, 849-853 (2010).

2. The wave function and quantum reality, in Proceedings of the International Conference on
Advances in Quantum Theory, A. Khrennikov, G. Jaeger, M. Schlosshauer, G. Weihs (eds),
AIP Conference Proceedings 1327, 334-338 (2011).

3. Meaning of the wave function, International Journal of Quantum Chemistry, 111, 4124-
4138 (2011).

4. Is gravity an entropic force? Entropy special issue “Black Hole Thermodynamics”, Jacob
D. Bekenstein (eds), 13, 936-948 (2011).

5. A quantum physical argument for panpsychism, Forthcoming in Journal of Consciousness

Studies, 2012.

6
 Contents

1 Introduction 10

2 Meaning of the Wave Function 13

2.1 Standard quantum mechanics and impulse measurements . . . . . . . . . . . 14
2.2 Weak measurements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.3 Protective measurements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.3.1 Measurements with natural protection . . . . . . . . . . . . . . . . . . 18
2.3.2 Measurements with artificial protection . . . . . . . . . . . . . . . . . 20
2.3.3 Further discussions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.4 On the mass and charge density of a quantum system . . . . . . . . . . . . . 22
2.4.1 A heuristic argument . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
2.4.2 The answer of protective measurement . . . . . . . . . . . . . . . . . . 23
2.4.3 A specific example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
2.5 The physical origin of mass and charge density . . . . . . . . . . . . . . . . . 26
2.5.1 The mass and charge density is effective . . . . . . . . . . . . . . . . . 27
2.5.2 The ergodic motion of a particle is discontinuous . . . . . . . . . . . . 28
2.5.3 An argument for random discontinuous motion . . . . . . . . . . . . . 30
2.6 The wave function as a description of random discontinuous motion of particles 31
2.6.1 An analysis of random discontinuous motion of particles . . . . . . . . 31
2.6.2 Interpreting the wave function . . . . . . . . . . . . . . . . . . . . . . 33

3 Schrödinger’s Equation and the Conservation Laws 36

3.1 Spacetime translation and its invariance . . . . . . . . . . . . . . . . . . . . . 37
3.2 Relativistic invariance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
3.3 Derivation of the free Schrödinger equation . . . . . . . . . . . . . . . . . . . 40
3.4 Further discussions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
3.5 On the conservation of energy and momentum . . . . . . . . . . . . . . . . . 43

4 A Suggested Solution to the Measurement Problem 47

4.1 The reality of wavefunction collapse . . . . . . . . . . . . . . . . . . . . . . . 48
4.1.1 Against the de Broglie-Bohm theory . . . . . . . . . . . . . . . . . . . 48
4.1.2 Against the many-worlds interpretation . . . . . . . . . . . . . . . . . 52
4.2 A conjecture on the origin of wavefunction collapse . . . . . . . . . . . . . . . 56
4.2.1 The chooser in discrete time . . . . . . . . . . . . . . . . . . . . . . . . 56
4.2.2 Energy conservation and the choices . . . . . . . . . . . . . . . . . . . 58
4.2.3 In search of a deeper basis . . . . . . . . . . . . . . . . . . . . . . . . . 59
4.3 A discrete model of energy-conserved wavefunction collapse . . . . . . . . . . 61
4.4 On the consistency of the model and experiments . . . . . . . . . . . . . . . . 67
4.4.1 Maintenance of coherence . . . . . . . . . . . . . . . . . . . . . . . . . 67
4.4.2 Rapid localization in measurement situations . . . . . . . . . . . . . . 68
4.4.3 Emergence of the classical world . . . . . . . . . . . . . . . . . . . . . 70
4.5 Critical comments on other dynamical collapse models . . . . . . . . . . . . . 72
4.5.1 Penrose’s gravity-induced wavefunction collapse model . . . . . . . . . 72
4.5.2 The CSL model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76

7
 CONTENTS

5 Random Discontinuous Motion and Special Relativity 79

5.1 The picture of motion distorted by the Lorentz transformation . . . . . . . . 79
5.1.1 Single particle picture . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
5.1.2 Picture of quantum entanglement . . . . . . . . . . . . . . . . . . . . . 81
5.1.3 Picture of wavefunction collapse . . . . . . . . . . . . . . . . . . . . . 82
5.2 On the absoluteness of simultaneity . . . . . . . . . . . . . . . . . . . . . . . . 84
5.3 Collapse dynamics and preferred Lorentz frame . . . . . . . . . . . . . . . . . 86

8
 I think I can safely say that nobody understands quantum mechanics... Do not keep
saying to yourself, if you can possible avoid it, “But how can it be like that?” because
you will get ‘down the drain’, into a blind alley from which nobody has escaped. Nobody
knows how it can be like that. — Richard Feynman, 1964
 1
Introduction

Quantum mechanics, according to its Schrödinger picture, is a non-relativistic theory about

the wave function and its evolution. There are two main problems in the conceptual foun-
dations of quantum mechanics. The first one concerns the physical meaning of the wave
function in the theory. It has been widely argued that the probability interpretation is not
wholly satisfactory because of resorting to the vague concept of measurement - though it is
still the standard interpretation in textbooks nowadays. On the other hand, the meaning of
the wave function is also in dispute in the alternatives to quantum mechanics such as the
de Broglie-Bohm theory and the many-worlds interpretation (de Broglie 1928; Bohm 1952;
Everett 1957; De Witt and Graham 1973). Exactly what does the wave function describe
then?
The second problem concerns the evolution of the wave function. It includes two parts.
One part concerns the linear Schrödinger evolution. Why does the linear non-relativistic
evolution of the wave function satisfy the Schrödinger equation? It seems that a satisfactory
derivation of the equation is still missing (cf. Nelson 1966). The other part concerns the
collapse of the wave function during a measurement, which is usually called the measurement
problem. The collapse postulate in quantum mechanics is ad hoc, and the theory does not
tell us how a definite measurement result emerges (Bell 1990). Although the alternatives to
quantum mechanics already give their respective solutions to this problem, it has been a hot
topic of debate which solution is right or in the right direction. In the final analysis, it is
still unknown whether the wavefunction collapse is real or not. Even if the wave function
does collapse under some circumstances, it remains unclear exactly why and how the wave
function collapses. The measurement problem has been widely acknowledged as one of the
hardest and most important problems in the foundations of quantum mechanics (see, e.g.
Wheeler and Zurek 1983).
In this thesis, we will try to solve these problems from a new angle. The key is to
realize that the problem of interpreting the wave function may be solved independent of
how to solve the measurement problem, and the solution to the first problem can then have
important implications for the solution to the second one. Although the meaning of the wave
function should be ranked as the first interpretative problem of quantum mechanics, it has
been treated as a marginal problem, especially compared with the measurement problem.
As noted above, there are already several alternatives to quantum mechanics which give
respective solutions to the measurement problem. However, these theories at their present
stages are unsatisfactory at least in one aspect; they have not succeeded in making sense of
the wave function. Different from them, our strategy is to first find what physical state the
wave function describes and then investigate the implications of the answer for the solutions
to other fundamental problems of quantum mechanics.
It seems quite reasonable that we had better know what the wave function is before we
want to figure out how it evolves, e.g. whether it collapses or not during a measurement.
However, these problems are generally connected to each other. In particular, in order to

10
know what physical state the wave function of a quantum system describes, we need to mea-
sure the system in the first place, while the measuring process and the measurement result are
necessarily determined by the evolution law for the wave function. Fortunately, it has been
realized that the conventional measurement that leads to the apparent collapse of the wave
function is only one kind of quantum measurement, and there also exists another kind of mea-
surement that avoids the collapse of the wave function, namely the protective measurement
(Aharonov and Vaidman 1993; Aharonov, Anandan and Vaidman 1993; Aharonov, Anandan
and Vaidman 1996). Protective measurement is a method to measure the expectation values
of observables on a single quantum system without disturbing its state appreciably, and its
mechanism is independent of the controversial process of wavefunction collapse and only
depends on the linear Schrödinger evolution and the Born rule, which are two established
parts of quantum mechanics. As a result, protective measurement can not only measure the
physical state of a quantum system and help to unveil the meaning of the wave function, but
also be used to examine the solutions to the measurement problem before experiments give
the last verdict. A full exposition of these ideas will be given in the subsequent chapters.
The plan of this thesis is as follows. In Chapter 2, we first investigate the physical
meaning of the wave function. According to protective measurement, the mass and charge
distributions of a quantum system as one part of its physical state can be measured as
expectation values of certain observables. It turns out that the mass and charge of a quantum
system are distributed throughout space, and the mass and charge density in each position
is proportional to the modulus square of the wave function of the system there. The key
to unveil the meaning of the wave function is to find the origin of the mass and charge
distributions. It is shown that the density is not real but effective; it is formed by the time
average of the ergodic motion of a localized particle with the total mass and charge of the
system. Moreover, it is argued that the ergodic motion is not continuous but discontinuous
and random. Based on this result, we suggest that the wave function represents the state
of random discontinuous motion of particles, and in particular, the modulus square of the
wave function (in position space) gives the probability density of the particles appearing in
certain positions in space.
In Chapter 3, we further analyze the linear evolution law for the wave function. It is shown
that the linear non-relativistic evolution of the wave function of an isolated system obeys the
free Schrödinger equation due to the requirements of spacetime translation invariance and
relativistic invariance. Though these requirements are already well known, an explicit and
complete derivation of the free Schrödinger equation using them seems still missing in the
literature. The new integrated analysis, which is consistent with the suggested interpretation
of the wave function, may be helpful for understanding the physical origin of the Schrödinger
equation. In addition, we also analyze the physical basis and meaning of the principle of
conservation of energy and momentum in quantum mechanics.
In Chapter 4, we investigate the implications of protective measurement and the sug-
gested interpretation of the wave function based on it for the solutions to the measurement
problem. To begin with, we argue that the two no-collapse quantum theories, namely the
de Broglie-Bohm theory and the many-worlds interpretation, are inconsistent with protec-
tive measurement and the picture of random discontinuous motion of particles. This result
strongly suggests that wavefunction collapse is a real physical process. Secondly, we ar-
gue that the random discontinuous motion of particles may provide an appropriate random
source to collapse the wave function. The key point is to realize that the instantaneous
state of a particle not only includes its wave function but also includes its random position,
momentum and energy that undergo the discontinuous motion, and these random variables
can have a stochastic influence on the evolution of the wave function and further lead to
the collapse of the wave function. Thirdly, we propose a discrete model of energy-conserved

11
wavefunction collapse. It is shown that the model is consistent with existing experiments
and our macroscopic experience. Lastly, we also give some critical comments on other dy-
namical collapse models, including Penrose’s gravity-induced collapse model and the CSL
(Continuous Spontaneous Localization) model.
In the last chapter, we briefly analyze the problem of the incompatibility between quan-
tum mechanics and special relativity in terms of random discontinuous motion of particles.
It is argued that a consistent description of random discontinuous motion of particles re-
quires absolute simultaneity, and this leads to the existence of a preferred Lorentz frame
when combined with the requirement of the constancy of speed of light. Moreover, it is
shown that the collapse dynamics may provide a method to detect the frame according to
the energy-conserved collapse model.

12
 2
What does the ψ-function mean now, that is, what does the
system described by it really look like in three dimensions?
— Erwin Schrödinger, 1927

Meaning of the Wave Function

The physical meaning of the wave function is an important interpretative problem of quantum
mechanics. Notwithstanding more than eighty years’ developments of the theory, however, it
is still a debated issue. Besides the standard probability interpretation in textbooks, there are
also various conflicting views on the wave function in the alternatives to quantum mechanics.
In this chapter, we will try to solve this fundamental interpretive problem through a new
analysis of protective measurement and the mass and charge density of a quantum system.
The meaning of the wave function is often analyzed in the context of conventional impulse
measurements, for which the coupling interaction between the measured system and the
measuring device is of short duration and strong. As a result, even though the wave function
of a quantum system is in general extended over space, an impulse position measurement will
inevitably collapse the wave function and can only detect the system in a random position
in space. Then it is unsurprising that the wave function is assumed to be related to the
probability of these random measurement results by the standard probability interpretation.
However, it has been known that there exists another kind of measurement that can avoid the
collapse of the wave function, namely the protective measurement (Aharonov and Vaidman
1993; Aharonov, Anandan and Vaidman 1993; Aharonov, Anandan and Vaidman 1996).
Protective measurement also uses a standard measuring procedure, but with a weak and
long duration coupling interaction and an appropriate procedure to protect the measured
wave function from collapsing. Its general method is to let the measured system be in a
nondegenerate eigenstate of the whole Hamiltonian using a suitable protective interaction
(in some situations the protection is provided by the measured system itself), and then make
the measurement adiabatically so that the state of the system neither changes nor becomes
entangled with the measuring device appreciably. In this way, such protective measurements
can measure the expectation values of observables on a single quantum system, and in
particular, the mass and charge density of a quantum system as one part of its physical state,
as well as its wave function, can be measured as expectation values of certain observables.
According to protective measurement, the mass and charge of a quantum system are
distributed throughout space, and the mass and charge density in each position is propor-
tional to the modulus square of the wave function of the system there. The key to unveil
the meaning of the wave function is to find the physical origin of the mass and charge
distributions. Historically, the charge density interpretation for electrons was originally sug-
gested by Schrödinger when he introduced the wave function and founded wave mechanics
(Schrödinger 1926). Although the existence of the charge density of an electron can provide a
classical explanation for some phenomena of radiation, its explanatory power is very limited.
In fact, Schrödinger clearly realized that the charge density cannot be classical because his
equation does not include the usual classical interaction between the densities. Presumably
since people thought that the charge density could not be measured and also lacked a consis-
tent physical picture, this initial interpretation of the wave function was soon rejected and
replaced by Born’s probability interpretation (Born 1926). Now protective measurement re-

13
 2.1. STANDARD QUANTUM MECHANICS AND IMPULSE MEASUREMENTS

endows the charge distribution of an electron with reality by a more convincing argument.
The question is then how to find a consistent physical explanation for it1 . Our following
analysis can be regarded as a further development of Schrödinger’s idea to some extent.
The twist is: that the charge distribution is not classical does not imply its non-existence;
rather, its existence points to a non-classical picture of quantum reality hiding behind the
mathematical wave function.
The charge distribution of a charged quantum system such as an electron has two possible
existent forms: it is either real or effective. The charge distribution is real means that it exists
throughout space at the same time, and the charge distribution is effective means that there
is only a localized particle with the total charge of the system at every instant, and its motion
forms the effective charge distribution. If the charge distribution is effective, then there will
exist no electrostatic self-interaction of the effective charge distribution, as there is only a
localized charged particle at every instant. By contrast, if the charge distribution is real,
then there will exist electrostatic self-interaction of the real charge distribution, as the charge
distribution exists throughout space at the same time. Since the superposition principle of
quantum mechanics prohibits the existence of electrostatic self-interaction, and especially,
the existence of the electrostatic self-interaction for the charge distribution of an electron
already contradicts experimental observations, the charge distribution of a quantum system
cannot be real but must be effective. This means that for a quantum system, at every instant
there is only a localized particle with the total mass and charge of the system, and during an
infinitesimal time interval at a given instant the time average of the ergodic motion of the
particle forms the effective mass and charge density in each position, which is proportional
to the modulus square of the wave function of the system there.
The next question is which sort of ergodic motion the particle undergoes. It can be argued
that the classical ergodic models, which assume continuous motion of particles, are incon-
sistent with quantum mechanics, and the effective mass and charge density of a quantum
system is formed by discontinuous motion of a localized particle. Moreover, the discontin-
uous motion is not deterministic but random. Based on this result, we suggest that the
wave function in quantum mechanics describes the state of random discontinuous motion of
particles, and at a deeper level, it represents the property of the particles that determines
their random discontinuous motion. In particular, the modulus square of the wave function
(in position space) determines the probability density of the particles appearing in every po-
sition in space. In the following, we will give a full exposition of this suggested interpretation
of the wave function.

2.1 Standard quantum mechanics and impulse measurements

The standard formulation of quantum mechanics, which was first developed by Dirac
(1930) and von Neumann (1955), is based on the following basic principles.
1. Physical states
The state of a physical system is represented by a normalized wave function or unit vector
|ψ(t)i in a Hilbert space2 . The Hilbert space is complete in the sense that every possible
physical state can be represented by a state vector in the space.
1
Note that the proponents of protective measurement did not give an analysis of the origin of the charge
distribution. According to them, this type of measurement implies that the wave function of a single quantum
system is ontological, i.e., that it is a real physical wave (Aharonov, Anandan and Vaidman 1993).
2
The Hilbert space is a compete vector space with scalar product. The state vector in a Hilbert space
contains proper vectors normalizable to unity as well as improper vectors normalizable only to the Dirac delta
functions. The exact nature of the Hilbert space depends on the system; for example, the state space for
position and momentum states is the space of square-integrable functions.

14
 2.1. STANDARD QUANTUM MECHANICS AND IMPULSE MEASUREMENTS

2. Physical properties
Every measurable property or observable of a physical system is represented by a Her-
mitian operator on the Hilbert space associated with the system. A physical system has a
determinate value for an observable if and only if it is in an eigenstate of the observable (this
is often called the eigenvalue-eigenstate link).
3. Composition rule
The Hilbert space associated with a composite system is the tensor product of the Hilbert
spaces associated with the systems of which it is composed. Similarly, the Hilbert space
associated with independent properties is the tensor product of the Hilbert spaces associated
with each property.
4. Evolution law
(1). Linear evolution
The state of a physical system |ψ(t)i obeys the linear Schrödinger equation i~ ∂|ψ(t)i ∂t =
H |ψ(t)i (when it is not measured), where ~ is Planck’s constant divided by 2π, H is the
Hamiltonian operator that depends on the energy properties of the system.
(2). Nonlinear collapse evolution P
If a physical system is in a state |ψi = i ci |ai i, where |ai i is the eigenstate of an
observable A with eigenvalue ai , then an (impulse) measurement of the observable A will
instantaneously, discontinuously, and randomly collapse the state into one of the eigenstates
|ai i with probability |ci |2 . This is usually called the collapse postulate, and the nonlinear
stochastic process is called the reduction of the state vector or the collapse of the wave
function.
The link between the mathematical formalism and experiments is provided by the Born
rule. It says that the probability of the above measurement of the observable A yielding the
result ai is |ci |2 .3 Note that the Born rule can be derived from the collapse postulate by
resorting to the eigenvalue-eigenstate link, but it does not necessarily depend on the postu-
late. Different from the controversial collapse postulate, the Born rule has been confirmed
by precise experiments and is an established part of quantum mechanics.
The conventional impulse measurements can be further formulated as follows. According
to the standard von Neumann procedure, measuring an observable A in a quantum state |ψi
involves an interaction Hamiltonian

HI = g(t)P A (2.1)
coupling the measured system to an appropriate measuring device, where P is the conjugate
momentum of the pointer R variable. The time-dependent coupling strength g(t) is a smooth
function normalized to dtg(t) = 1 during the interaction interval τ , and g(0) = g(τ ) = 0.
The initial state of the pointer is supposed to be a Gaussian wave packet of width w0 centered
at initial position 0, denoted by |φ(0)i.
For an impulse measurement, the interaction HI is of very short duration and so strong
that it dominates the rest of the Hamiltonian (i.e. the effect of the free Hamiltonians of
the measuring device and the measured system can be neglected). Then the state of the
combined system at the end of the interaction can be written as
i
|t = τ i = e− ~ P A |ψi |φ(0)i . (2.2)
By expanding |ψi in the eigenstates of A, |ai i, we obtain
3
For a continuous property such as position, P (x) = |hx|ψi|2 is the probability density at x, and P (x)dx
is the probability of obtaining a measurement result between x and x + dx.

15
 2.2. WEAK MEASUREMENTS

X i
|t = τ i = e− ~ P ai ci |ai i |φ(0)i , (2.3)
i
where ci are the expansion coefficients. The exponential term shifts the center of the pointer
by ai :
X
|t = τ i = ci |ai i |φ(ai )i . (2.4)
i
This is an entangled state, where the eigenstates of A with eigenvalues ai get correlated to
measuring device states in which the pointer is shifted by these values ai (but the width
of the pointer wavepacket is not changed). Then by the collapse postulate, the state will
instantaneously and randomly collapse into one of its branches |ai i |φ(ai )i with probability
|ci |2 . This means that the measurement result can only be one of the eigenvalues of measured
observable A, say ai , with a certain probability |ci |2 . The expectation value of A is then
obtained P as the statistical average of eigenvalues for an ensemble of identical systems, namely
2
hAi = i |ci | ai .

2.2 Weak measurements

The conventional impulse measurements are only one kind of quantum measurements, for
which the coupling between the measured system and the measuring device is very strong,
and the results are only the eigenvalues of measured observable. We can also obtain other
kinds of measurements by adjusting the coupling strength. An interesting example is weak
measurements (Aharonov, Albert and Vaidman 1988), for which the measurement result is
the expectation value of the measured observable. In this section, we will introduce the basic
principle of weak measurements.
A weak measurement is a standard measuring procedure with weakened coupling. As in
the conventional impulse measurement, the Hamiltonian of the interaction with the measur-
ing device is also given by Eq. (2.1) in a weak measurement. The weakness of the interaction
is achieved by preparing the initial state of the measuring device in such a way that the con-
jugate momentum of the pointer variable is localized around zero with small uncertainty,
and thus the interaction Hamiltonian (2.1) is small. As a simple example, let the initial state
of the pointer in position space be:
2 /2w 2
hx |φ(0)i = (w02 π)−1/4 e−x 0 . (2.5)
The corresponding initial probability distribution is
2 /w 2
Pi (x) = (w02 π)−1/2 e−x 0 . (2.6)

Expanding Pthe initial state of the system |ψi in the eigenstates |ai i of the measured observable
A, |ψi = i ci |ai i, then after the interaction (2.1) the state of the system and the measuring
device is:
2 2
X
|t = τ i = (w02 π)−1/4 ci |ai ie−(x−ai ) /2w0 . (2.7)
i
The probability distribution of the pointer variable corresponding to the final state (2.7) is:
2 /w 2
X
Pf (x) = (w02 π)−1/2 |ci |2 e−(x−ai ) 0 . (2.8)
i
In case of a conventional impulse measurement, this is a weighted sum of the initial
probability distribution localized around various eigenvalues ai . Therefore, the reading of

16
 2.3. PROTECTIVE MEASUREMENTS

the pointer variable in the end of the measurement always yields the value close to one of
the eigenvalues. By contrast, the limit of weak measurement corresponds to w0  ai for all
eigenvalues ai . Then we can perform the Taylor expansion of the sum (2.8) around x = 0
up to the first order and rewrite the final probability distribution of the pointer variable in
the following way:

|ci |2 ai )2 /w02
X P
Pf (x) ≈ (w02 π)−1/2 |ci |2 (1 − (x − ai )2 /w02 ) ≈ (w02 π)−1/2 e−(x− i (2.9)
i

This is the initial probability distribution shifted by the value i |ci |2 ai . It indicates that
P
the result of the weak measurement is the expectation value of the measured observable in
the measured state:
X
hAi ≡ hψ|A|ψi = |ci |2 ai . (2.10)
i

Certainly, since the width of the pointer wavepacket is much greater than the shift of
the center of the pointer, namely w0  hAi, the above weak measurement of a single system
is very imprecise4 . However, by performing the weak measurement√ on an ensemble of N
identical systems the precision can be improved by a factor N . This scheme of weak
measurement has been realized and proved useful in quantum optical experiments (see, e.g.
Hosten and Kwiat 2008).
Although weak measurements, like conventional impulse measurements, also need to
measure an ensemble of identical quantum systems, they are conceptually different. For
conventional impulse measurements, every identical system in the ensemble shifts the pointer
of the measuring device by one of the eigenvalues of the measured observable, and the
expectation value of the observable is then regarded as the property of the whole ensemble.
By contrast, for weak measurements, every identical system in the ensemble shifts the pointer
of the measuring device directly by the expectation value of the measured observable, and
thus the expectation value may be regarded as the property of individual systems.

2.3 Protective measurements

Protective measurements are improved methods based on weak measurements, and they
can measure the expectation values of observables on a single quantum system.
As we have seen above, although the measured state is not changed appreciably by a
weak measurement, the pointer of the measuring device hardly moves either. In particular,
the shift of the pointer due to the measurement is much smaller than its position uncertainty,
and thus little information can be obtained from individual measurements. A possible way to
remedy the weakness of weak measurements is to increase the time of the coupling between
the measured system and the measuring device. If the state is almost constant during the
measurement, the total shift of the pointer, which is proportional to the duration of the
interaction, will be large enough to be identified. However, under normal circumstances the
state of the system is not constant during the measurement, and the weak coupling also leads
to a small rate of change of the state. As a result, the reading of the measuring device will
correspond not to the state which the system had prior to the measurement, but to some
time average depending on the evolution of the state influenced by the measuring procedure.
4
In order to read the position of pointer, an impulse position measurement needs to be made after the
weak measurement, and this will lead to a partial collapse of the measured wave function. For a helpful
discussion see Miller (2010).

17
 2.3. PROTECTIVE MEASUREMENTS

Therefore, in order to be able to measure the state of a single system, we need, in

addition to the standard weak and long-duration measuring interaction, a procedure which
can protect the state from changing during the measuring interaction. A general method
is to let the measured system be in a nondegenerate eigenstate of the whole Hamiltonian
using a suitable protective interaction, and then make the measurement adiabatically so that
the state of the system neither collapses nor becomes entangled with the measuring device
appreciably. In this way, protective measurement can measure the expectation values of
observables on a single quantum system. In the following, we will introduce the principle of
protective measurement in more detail (Aharonov and Vaidman 1993; Aharonov, Anandan
and Vaidman 1993; Aharonov, Anandan and Vaidman 1996)5 .

2.3.1 Measurements with natural protection

As a typical example, we consider a quantum system in a discrete nondegenerate energy
eigenstate |En i. In this case, the system itself supplies the protection of the state due to
energy conservation and no artificial protection is needed.
The interaction Hamiltonian for a protective measurement of an observable A in this
state involves the same interaction Hamiltonian as the standard measuring procedure:

HI = g(t)P A, (2.11)
where P is the momentum conjugate to the pointer variable X of an appropriate measuring
device. Let the initial state of the pointer at t = 0 be |φ(x0 )i, which is a Gaussian wave
packet of eigenstates of X with width w0 , centered around the eigenvalue R x0 . The time-
dependent coupling strength g(t) is also a smooth function normalized to dtg(t) = 1. But
different from conventional impulse measurements, for which the interaction is very strong
and almost instantaneous, protective measurements make use of the opposite limit where
the interaction of the measuring device with the system is weak and adiabatic, and thus the
free Hamiltonians cannot be neglected. Let the Hamiltonian of the combined system be

H(t) = HS + HD + g(t)P A, (2.12)

where HS and HD are the Hamiltonians of the measured system and the measuring device,
respectively. The interaction lasts for a long time T , and g(t) is very small and constant for
the most part, and it goes to zero gradually before and after the interaction.
The state of the combined system after T is given by
i
RT
|t = T i = e− ~ 0 H(t)dt
|En i |φ(x0 )i . (2.13)
By ignoring the switching on and switching off processes6 , the full Hamiltonian (with g(t) =
1/T ) is time-independent and no time-ordering is needed. Then we obtain
i
|t = T i = e− ~ HT |En i |φ(x0 )i , (2.14)
5
The earlier objections to the validity and meaning of protective measurements have been answered
(Aharonov, Anandan and Vaidman 1996; Dass and Qureshi 1999). A unique exception is Uffink’s (1999)
objection. Although Vaidman (2009) regarded this objection as a misunderstanding, he gave no concrete
rebuttal. Recently we have argued in detail that Uffink’s objection is invalid due to several errors in his
arguments (Gao 2011d).
6
The change in the total Hamiltonian during these processes is smaller than P A/T , and thus the adia-
baticity of the interaction will not be violated and the approximate treatment given below is valid. For a
more strict analysis see Dass and Qureshi (1999).

18
 2.3. PROTECTIVE MEASUREMENTS

E
PA
where H = HS + HD + T . We further expand |φ(x0 )i in the eigenstate of HD , Ejd , and
write
i X E
|t = T i = e− ~ HT cj |En i Ejd , (2.15)
j

Let the exact eigenstates of H be |Ψk,m i and the corresponding eigenvalues be E(k, m), we
have
X X i
|t = T i = cj e− ~ E(k,m)T hΨk,m |En , Ejd i|Ψk,m i. (2.16)
j k,m

Since the interaction is very weak, the Hamiltonian H of Eq.(2.12) can be thought of as
H0 = HS + HD perturbed by PTA . Using the fact that PTA is a small perturbation and that
d , the perturbation theory gives
the eigenstates of H0 are of the form |Ek i Em

E
d
|Ψk,m i = |Ek i Em + O(1/T ),
d 1
E(k, m) = Ek + Em + hAik hP im + O(1/T 2 ). (2.17)
T
Note that it is a necessary condition for Eq.(2.17) to hold that |Ek i is a nondegenerate
eigenstate of HS . Substituting Eq.(2.17) in Eq.(2.16) and taking the large T limit yields
X − i (E T +E d T +hAi hP i ) E
|t = T i ≈ e ~ n j n j
cj |En i Ejd . (2.18)
j

For the special case when PE commutes with the free Hamiltonian of the device, i.e.,
[P, HD ] = 0, the eigenstates Ejd of HD are also the eigenstates of P , and thus the above
equation can be rewritten as
i i i
|t = T i ≈ e− ~ En T − ~ HD T − ~ hAin P |En i |φ(x0 )i . (2.19)
It can be seen that the third term in the exponent will shift the center of the pointer |φ(x0 )i
by an amount hAin :
i i
|t = T i ≈ e− ~ En T − ~ HD T |En i |φ(x0 + hAin )i. (2.20)
This shows that at the end of the interaction, the center of the pointer shifts by the expec-
tation value of the measured observable in the measured state. E
For the general case when [P, HD ] 6= 0, we can introduce an operator Y = j hP ij Ejd hEjd |
P

and rewrite Eq.(2.18) as

i i i
|t = T i ≈ e− ~ En T − ~ HD T − ~ hAin Y |En i |φ(x0 )i . (2.21)
Then by rechoosing the state of the device so that it is peaked around a value x00 of the
pointer variable X 0 conjugate to Y , i.e., [X 0 , Y ] = i~,7 we can obtain

i i i i i
|t = T i ≈ e− ~ En T − ~ HD T − ~ hAin Y |En i φ(x00 ) = e− ~ En T − ~ HD T |En i |φ(x00 + hAin )i. (2.22)
7
Note that it may not always be possible to physically realize the operator Y , and an operator canonically
conjugate to Y need not always exist either. For further discussions see Dass and Qureshi (1999).

19
 2.3. PROTECTIVE MEASUREMENTS

Thus the center of the pointer also shifts by hAin at the end of the interaction. This
demonstrates the generic possibility of the protective measurement of hAin .
It is worth noting that since the position variable of the pointer does not commute with
its free Hamiltonian, the pointer wave packet will spread during the long measuring time.
For example, the kinematic energy term P 2 /2M in the free Hamiltonian of the pointer will
spread the wave packet without shifting the center, and the width of the wave packet at
2 1
the end of interaction will be w(T ) = [ 21 (w02 + MT2 w2 )] 2 (Dass and Qureshi 1999). However,
0
the spreading of the pointer wave packet can be made as small as possible by increasing the
mass M of the pointer, and thus it will not interfere with resolving the shift of the center of
the pointer in principle8 .

2.3.2 Measurements with artificial protection

Protective measurements can not only measure the discrete nondegenerate energy eigenstates
of a single quantum system, which are naturally protected by energy conservation, but
also measure the general quantum states by adding an artificial protection procedure in
principle (Aharonov and Vaidman 1993). For this case, the measured state needs to be
known beforehand in order to arrange a proper protection.
For degenerate energy eigenstates, the simplest way is to add a potential (as part of
the measuring procedure) to change the energies of the other states and lift the degeneracy.
Then the measured state remains unchanged, but is now protected by energy conservation
like nondegenerate energy eigenstates. Although this protection does not change the state,
it does change the physical situation. This change can be brought to a minimum by adding
strong protection potential for a dense set of very short time intervals. Then most of the
time the system has not only the same state, but also the original potential.
The superposition of energy eigenstates can be measured by a similar procedure. One
can add a dense set of time-dependent potentials acting for very short periods of time such
that the state at all these times is the nondegenerate eigenstate of the Hamiltonian together
with the additional potential. Then most of the time the system also evolves under the
original Hamiltonian. A stronger protection is needed in order to measure all details of the
time-dependent state. One way is via the quantum Zeno effect. The frequent impulse mea-
surements can test and protect the time evolution of the quantum state. For measurement
of any desired accuracy of the state, there is a density of the impulse measurements which
can protect the state from being changed due to the measuring interaction. When the time
scale of intervals between consecutive protections is much smaller than the time scale of the
original state evolution, the system will evolve according to its original Hamiltonian most
of the time, and thus what’s measured is still the property of the system and not of the
protection procedure (Aharonov and Vaidman 1993).
Lastly, we note that the scheme of protective measurement can also be extended to a
many-particle system (Anandan 1993). If the system is in a product state, then this is
easily done by protectively measuring each state of the individual systems. But this is
8
As in conventional impulse measurements, there is also an issue of retrieving the information about the
center of the wave packet of the pointer (Dass and Qureshi 1999). One strategy is to consider adiabatic
coupling of a single quantum system to an ensemble of measuring devices and make impulse position mea-
surements on the ensemble of devices to determine the pointer position. For example, the ensemble of devices
could be a beam of atoms interacting adiabatically with the spin of the system. Although such an ensemble
approach inevitably carries with it uncertainty in the knowledge of the position of the device, the pointer
position, which is the average of the result of these position measurements, can be determined with arbitrary
accuracy. Another approach is to make repeated measurements (e.g. weak quantum nondemolition measure-
ments) on the single measuring device. This issue does not affect the principle of protective measurements.
In particular, retrieving the information about the position of the pointer only depends on the Born rule and
is irrelevant to whether the wave function collapses or not during a conventional impulse measurement.

20
 2.3. PROTECTIVE MEASUREMENTS

impossible when the system is in an entangled state because neither particle is then in a
unique state that can be protected. If a protective measurement is made only on one of the
particles, then this would also collapse the entangled state into one of the eigenstates of the
protecting Hamiltonian. The right method is by adding appropriate protection procedure
to the whole system so that the entangled state is a nondegenerate eigenstate of the total
Hamiltonian of the system together with the added potential. Then the entangled state can
be protectively measured. Note that the additional protection usually contains a nonlocal
interaction for separated particles. However, this measurement may be performed without
violating causality by having the entangled particles sufficiently close to each other so that
they have this protective interaction. Then when the particles are separated they would still
be in the same entangled state which has been protectively measured.

2.3.3 Further discussions

According to the standard view, the expectation values of observables are not the physi-
cal properties of a single system, but the statistical properties of an ensemble of identical
systems. This seems reasonable if there exist only conventional impulse measurements. An
impulse measurement can only obtain one of the eigenvalues of the measured observable,
and thus the expectation value can only be defined as a statistical average of the eigenvalues
for an ensemble of identical systems. However, as we have seen above, there exist other
kinds of quantum measurements, and in particular, protective measurements can measure
the expectation values of observables for a single system, using an adiabatic measuring pro-
cedure. Therefore, the expectation values of observables should be considered as the physical
properties of a single quantum system, not those of an ensemble (Aharonov, Anandan and
Vaidman 1996)9 .
It is worth pointing out that a realistic protective measurement (where the measuring
time T is finite) can never be performed on a single quantum system with absolute certainty
because of the tiny unavoidable entanglement in the final state10 . For example, we can
only obtain the exact expectation value hAi with a probability very close to one, and the
measurement may also result in apparent collapse and its result be the expectation value
hAi⊥ with a probability proportional to ∼ 1/T 2 , where ⊥ refers to a normalized state
in the subspace normal to the initial state as picked out by the first-order perturbation
theory(Dass and Qureshi 1999). Therefore, a small ensemble is still needed for a realistic
protective measurement, and the size of the ensemble is in inverse proportion to the duration
of measurement. However, the limitation of a realistic protective measurement does not
influence the above conclusion. The key point is that the effects of entanglement and apparent
collapse can be made arbitrarily small, and a protective measurement can measure the
expectation values of observables on a single quantum system with certainty in principle
(when the measuring time T approaches infinite). Thus the expectation values of observables
should be regarded as the physical properties of a quantum system.
In addition, we can also provide an argument against the standard view, independent of
the above analysis of protective measurement. First of all, although the expectation values
of observables can only be obtained by measuring an ensemble of identical systems in the
context of conventional impulse measurements, this fact does not necessarily entails that they
9
Anandan (1993) and Dickson (1995) gave some primary analyses of the implications of this result for
quantum realism. According to Anandan (1993), protective measurement refutes an argument of Einstein in
favor of the ensemble interpretation of quantum mechanics. Dickson’s (1995) analysis was more philosophical.
He argued that protective measurement provides a reply to scientific empiricism about quantum mechanics,
but it can neither refute that position nor confirm scientific realism, and the aim of his argument is to place
realism and empiricism on an even score in regards to quantum mechanics.
10
This point was discussed and stressed by Dass and Qureshi (1999).

21
 2.4. ON THE MASS AND CHARGE DENSITY OF A QUANTUM SYSTEM

can only be the statistical properties of the ensemble. Next, if each system in the ensemble is
indeed identical as the standard view holds (this means that the quantum state is a complete
description of a single system), then obviously the expectation values of observables will be
also the properties of each individual system in the ensemble. Thirdly, even if the quantum
state is not a complete description of a single system and additional variables are needed as
in the de Broglie-Bohm theory (de Broglie 1928; Bohm 1952), the quantum state of each
system in an ensemble of identical systems is still the same, and thus the expectation values
of observables, which are calculated in terms of the quantum state, are also the same for
every system in the ensemble. As a result, the expectation values of observables can still be
regarded as the properties of individual systems.
Lastly, we stress that the expectation values of observables are instantaneous properties
of a quantum system (Aharonov, Anandan and Vaidman 1996). Although the measured
state may be unchanged during a protective measurement and the duration of measurement
may be very long, for an arbitrarily short period of time the measuring device always shifts
by an amount proportional to the expectation value of the measured observable in the
state. Therefore, the expectation values of observables are not time-averaged properties
of a quantum system during a finite period of time, but instantaneous properties of the
system during an infinitesimal period of time or at a precise instant11 .

2.4 On the mass and charge density of a quantum system

The fundamental assumption is that the space density of electricity is given by the square
of the wavefunction.12 — Erwin Schrödinger, 1926

According to protective measurement, the expectation values of observables are prop-

erties of a single quantum system. Typical examples of such properties are the mass and
charge density of a quantum system. In this section, we will present a detailed analysis of
this property, as it may have important implications for the physical meaning of the wave
function.

2.4.1 A heuristic argument

The mass and charge of a classical system always localize in a definite position in space at
each moment. For a charged quantum system described by the wave function ψ(x, t), how
do its mass and charge distribute in space then? We can measure the total mass and charge
of the quantum system by the gravitational and electromagnetic interactions and find them
in certain region of space. Thus it seems that the mass and charge of a quantum system
must also exist in space with a certain distribution. Before we discuss the answer given by
protective measurement, we will first give a heuristic argument.
The Schrödinger equation of a charged quantum system under an external electromag-
netic potential may provide a clue to the answer. The equation is

∂ψ(x, t) ~2 iQ 2
i~ = [− (∇ − A) + Qϕ]ψ(x, t), (2.23)
∂t 2m ~
where m and Q are the mass and charge of the system, respectively, ϕ and A are the electro-
magnetic potential. The electrostatic interaction term Qϕψ(x, t) in the equation indicates
11
Our later analysis of the mass and charge density of a quantum system will further show that the
expectation values of observables are the instantaneous properties of a quantum system during an infinitesimal
time interval, like the standard velocities in classical mechanics.
12
Quoted in Moore (1994), p.148.

22
 2.4. ON THE MASS AND CHARGE DENSITY OF A QUANTUM SYSTEM

that the interaction exists in all regions where the wave function of the system, ψ(x, t), is
nonzero, and thus it seems to suggest that the charge of the system also distributes through-
out these regions. If the charge does not distribute in some regions where the wave function
is nonzero, then
R ∞ there will2 not exist an electrostatic interaction there. Furthermore, since
3
the integral −∞ Q|ψ(x, t)| d x is the total charge of the system, the charge density in space,
if indeed exists, will be Q|ψ(x, t)|2 . Similarly, the mass density can be obtained from the
Schrödinger equation of a quantum system under an external gravitational potential:

∂ψ(x, t) ~2 2
i~ = [− ∇ + mVG ]ψ(x, t). (2.24)
∂t 2m
The gravitational interaction term mVG ψ(x, t) in the equation also suggests that the (passive
gravitational) mass of the quantum system distributes throughout the whole region where
its wave function ψ(x, t) is nonzero, and the mass density in space is m|ψ(x, t)|2 .

2.4.2 The answer of protective measurement

In the following, we will show that protective measurement provides a more convincing argu-
ment for the existence of mass and charge density. The mass and charge density of a single
quantum system, as well as its wave function, can be measured by protective measurement
as expectation values of certain observables (Aharonov and Vaidman 1993). For example,
a protective measurement of the flux of the electric field of a charged quantum system out
of a certain region will yield the expectation value of its charge inside this region, namely
the integral of its charge density over this region. Similarly, we can also measure the mass
density of a quantum system by a protective measurement of the flux of its gravitational
field in principle (Anandan 1993).
Consider a quantum system in a discrete nondegenerate energy eigenstate ψ(x). We take
the measured observable An to be (normalized) projection operators on small spatial regions
Vn having volume vn :
(
1
, if x ∈ Vn ,
An = vn (2.25)
0, if x 6∈ Vn .
The protective measurement of An then yields
Z
1
hAn i = |ψ(x)|2 dv = |ψn |2 , (2.26)
v n Vn
where |ψn |2 is the average of the density ρ(x) = |ψ(x)|2 over the small region Vn . Then when
vn → 0 and after performing measurements in sufficiently many regions Vn we can measure
ρ(x) everywhere in space.
When the observable An and the corresponding interaction Hamiltonian are physically
realized by the electromagnetic or gravitational interaction between the measured system
and the measuring device, what the above protective measurement measures is in fact the
charge or mass density of the quantum system13 , and its result indicates that the mass and
charge density of the system in each position x is proportional to the modulus square of
its wave function there, namely the density ρ(x). In the following, we will give a concrete
example to illustrate this important result (see also Aharonov, Anandan and Vaidman 1993).
13
This important point was also stressed by Aharonov, Anandan and Vaidman (1993).

23
 2.4. ON THE MASS AND CHARGE DENSITY OF A QUANTUM SYSTEM

2.4.3 A specific example

Consider the spatial wave function of a single quantum system with negative charge Q (e.g.
Q = −e)

ψ(x, t) = aψ1 (x, t) + bψ2 (x, t), (2.27)

where ψ1 (x, t) and ψ2 (x, t) are two normalized wave functions respectively localized in their
ground states in two small identical boxes 1 and 2, and |a|2 + |b|2 = 1. An electron, which
initial state is a Gaussian wave packet narrow in both position and momentum, is shot along
a straight line near box 1 and perpendicular to the line of separation between the boxes.
The electron is detected on a screen after passing by box 1. Suppose the separation between
the boxes is large enough so that a charge Q in box 2 has no observable influence on the
electron. Then if the system were in box 2, namely |a|2 = 0, the trajectory of the electron
wave packet would be a straight line as indicated by position “0” in Fig.1. By contrast, if the
system were in box 1, namely |a|2 = 1, the trajectory of the electron wave packet would be
deviated by the electric field of the system by a maximum amount as indicated by position
“1” in Fig.1.
We first suppose that ψ(x, t) is unprotected, then the wave function of the combined
system after interaction will be

ψ(x, x0 , t) = aϕ1 (x0 , t)ψ1 (x, t) + bϕ2 (x0 , t)ψ2 (x, t), (2.28)
where ϕ1 (x0 , t) and ϕ2 (x0 , t) are the wave functions of the electron influenced by the electric
fields of the system in box 1 and box 2, respectively, the trajectory of ϕ1 (x0 , t) is deviated
by a maximum amount, and the trajectory of ϕ2 (x0 , t) is not deviated and still a straight
line. When the electron is detected on the screen, the above wave function will collapse to
ϕ1 (x0 , t)ψ1 (x, t) or ϕ2 (x0 , t)ψ2 (x, t). As a result, the detected position of the electron will be
either “1” or “0” in Fig.1, indicating that the system is in box 1 or 2 after the detection.
This is a conventional impulse measurement of the projection operator on the spatial region
of box 1, denoted by A1 . A1 has two eigenstates corresponding to the system being in box
1 and 2, respectively, and the corresponding eigenvalues are 1 and 0, respectively. Since the
measurement is accomplished through the electrostatic interaction between two charges, the
measured observable A1 , when multiplied by the charge Q, is actually the observable for
the charge of the system in box 1, and its eigenvalues are Q and 0, corresponding to the
charge Q being in boxes 1 and 2, respectively. Such a measurement cannot tell us the charge
distribution of the system in each box before the measurement.

24
 2.4. ON THE MASS AND CHARGE DENSITY OF A QUANTUM SYSTEM

Fig.1 Scheme of a protective measurement of the charge density of a quantum system

Now let’s make a protective measurement of A1 . Since ψ(x, t) is degenerate with its
0
orthogonal state ψ (x, t) = b∗ ψ1 (x, t) − a∗ ψ2 (x, t), we need an artificial protection procedure
to remove the degeneracy, e.g. joining the two boxes with a long tube whose diameter is
small compared to the size of the box14 . By this protection ψ(x, t) will be a nondegenerate
energy eigenstate. The adiabaticity condition and the weakly interacting condition, which
are required for a protective measurement, can be further satisfied when assuming that (1)
the measuring time of the electron is long compared to ~/∆E, where ∆E is the smallest of
the energy differences between ψ(x, t) and the other energy eigenstates, and (2) at all times
the potential energy of interaction between the electron and the system is small compared
to ∆E. Then the measurement of A1 by means of the electron trajectory is a protective
measurement, and the trajectory of the electron is only influenced by the expectation value
of the charge of the system in box 1. In particular, when the size of box 1 can be omitted
compared with the separation between it and the electron wave packet, the wave function of
the electron will obey the following Schrödinger equation:

∂ψ(~r, t) ~2 2 e · |a|2 Q
i~ =− ∇ ψ(~r, t) − k ψ(~r, t), (2.29)
∂t 2me |~r − r~1 |
where me is the mass of electron, k is the Coulomb constant, r~1 is the position of the center
of box 1, and |a|2 Q is the expectation value of the charge Q in box 1. Correspondingly, the
trajectory of the center of the electron wave packet, r~c (t), will satisfy the following equation
by Ehrenfest’s theorem:

d2 r~c e · |a|2 Q
me = −k . (2.30)
dt2 |~
rc − r~1 |(~
rc − r~1 )
Then the electron wave packet will reach the position “|a|2 ” between “0” and “1” on the
screen as denoted in Fig.1. This shows that the result of the protective measurement is the
14
It is worth stressing that the added protection procedure depends on the measured state, and different
states need different protection procedures in general.

25
 2.5. THE PHYSICAL ORIGIN OF MASS AND CHARGE DENSITY

expectation value of the projection operator A1 , namely the integral of the density |ψ(x)|2
in the region of box 1. When multiplied by Q, it is the expectation value of the charge Q in
the state ψ1 (x, t) in box 1, namely the integral of the charge density Q|ψ(x)|2 in the region
of box 1. In fact, as Eq. (2.29) and Eq. (2.30) clearly indicate, this is what the protective
measurement really measures.
As we have argued in the last section, the result of a protective measurement reflects the
objective property or physical state of the measured system. Thus the result of the above
protective measurement, namely the expectation value of the charge Q in the state ψ1 (x, t),
|a|2 Q, will reflect the actual charge distribution of the system in box 1. In other words,
the result indicates that there exists a charge |a|2 Q in box 1. In the following, we will give
another two arguments for this conclusion.
First of all, let’s analyze the result of the protective measurement. Suppose we can
continuously change the measured state from |a|2 = 0 to |a|2 = 1 (and adjust the protective
interaction correspondingly). When |a|2 = 0, the single electron will reach the position
“0” of the screen one by one, and it is incontrovertible that no charge is in box 1. When
|a|2 = 1, the single electron will reach the position “1” of the screen one by one, and it is
also incontrovertible that there is a charge Q in box 1. Then when |a|2 assumes a numerical
value between 0 and 1 and the single electron reaches the position “|a|2 ” between “0” and
“1” on the screen one by one, the results should similarly indicate that there is a charge
|a|2 Q in the box by continuity. The point is that the definite deviation of the trajectory of
the electron will reflect that there exists a definite amount of charge in box 1.15 Next, let’s
analyze the equation that determines the result of the protective measurement, namely Eq.
(2.30). It gives a more direct support for the existence of a charge |a|2 Q in box 1. The r.h.s
of Eq. (2.30) is the formula of the electric force between two charges located in different
spatial regions. It is incontrovertible that e is the charge of the electron, and it exists in the
position ~r. Then |a|2 Q should be the other charge that exists in the position r~1 . In other
words, there exists a charge |a|2 Q in box 1.
In conclusion, protective measurement shows that a quantum system with mass m and
charge Q, which is described by the wave function ψ(x, t), has mass density m|ψ(x, t)|2 and
charge density Q|ψ(x, t)|2 in space, respectively16 . Note that this conclusion is only based
on two established parts of quantum mechanics, namely the linear Schrödinger evolution
and the Born rule. In the above example, the linear Schrödinger evolution determines the
deviation of the electron wave packet, and the Born rule is needed to obtain the information
about the center of the electron wave packet detected on the screen.

2.5 The physical origin of mass and charge density

We have argued that the mass and charge of a quantum system are distributed throughout
space, and the mass and charge density in each position is proportional to the modulus
square of the wave function of the system there. In this section, we will further investigate
the physical origin of the mass and charge density. As we will see, the answer may provide
15
Any physical measurement is necessarily based on some interaction between the measured system and the
measuring system. One basic form of interaction is the electrostatic interaction between two electric charges
as in our example, and the existence of this interaction during a measurement, which is indicated by the
deviation of the trajectory of the charged measuring system such as an electron, means that the measured
system also has the charge responsible for the interaction. If one denies this point, then it seems that one
cannot obtain any information about the measured system by the measurement. Note that the arguments
against the naive realism about operators and the eigenvalue realism in the quantum context are irrelevant
here (Daumer et al 1997; Valentini 2010).
16
Strictly speaking, the mass density is m|ψ(x)|2 + ψ ∗ Hψ/c2 in the non-relativistic domain, but the second
term is very small compared with the first term and can be omitted.

26
 2.5. THE PHYSICAL ORIGIN OF MASS AND CHARGE DENSITY

an important clue to the physical meaning of the wave function.

2.5.1 The mass and charge density is effective

As noted earlier, the expectation values of observables are the properties of a quantum
system either at a precise instant or during an infinitesimal time interval. Correspondingly,
the mass and charge distribution of a quantum system, which can be protectively measured
as the expectation values of certain observables, has two possible existent forms: it is either
real or effective. The distribution is real means that it exists throughout space at the same
time. The distribution is effective means that there is only a localized particle with the total
mass and charge of the system at every instant, and the time average of its motion during
an infinitesimal time interval forms the effective distribution. Moreover, since the integral of
the formed mass and charge density in any region is required to be equal to the expectation
value of the total mass and charge in the region, the motion of the particle must be ergodic.
In the following, we will determine the existent form of the mass and charge distribution of
a quantum system.
If the mass and charge distribution is effective, then there will exist no gravitational
and electrostatic self-interactions of the effective distribution, as there is only a localized
particle at every instant. This is consistent with the superposition principle of quantum
mechanics and the derivation of the mass and charge distribution based on protective mea-
surement. By contrast, if the mass and charge distribution is real, then there will exist
gravitational and electrostatic self-interactions of the real distribution, as the distribution
exists throughout space at the same time17 . The existence of the gravitational and electro-
static self-interactions is inconsistent with the superposition principle of quantum mechanics
(at least for microscopic systems such as electrons). Moreover, as we will show below, the
existence of the electrostatic self-interaction for the charge distribution of an electron also
contradicts experimental observations.
Interestingly, the Schrödinger-Newton equation, which was proposed by Diosi (1984) and
Penrose (1998), just describes the gravitational self-interaction of the mass density. The
equation for a single quantum system can be written as

~2 2 |ψ(x0 , t)|2 3 0
Z
∂ψ(x, t)
i~ =− ∇ ψ(x, t) − Gm2 d x ψ(x, t) + V ψ(x, t), (2.31)
∂t 2m |x − x0 |

where m is the mass of the quantum system, V is an external potential, G is Newton’s

gravitational constant18 . If there is also an electrostatic self-interaction, then the equation
for a free quantum system with mass m and charge Q will be

~2 2 |ψ(x0 , t)|2 3 0
Z
∂ψ(x, t) 2 2
i~ =− ∇ ψ(x, t) + (kQ − Gm ) d x ψ(x, t). (2.32)
∂t 2m |x − x0 |
Note that the gravitational self-interaction is attractive, while the electrostatic self-interaction
is repulsive. It has been shown that the measure of the potential strength of the gravitational
2
self-interaction is ε2 = ( 4Gm 2
~c ) for a free system with mass m (Salzman 2005). This quantity
17
That a real charge distribution has electrostatic self-interaction has been confirmed not only in the classical
domain but also in the quantum domain for many-body systems. For example, two charged quantum systems
such as two electrons have electrostatic interaction, and thus a real charge distribution containing these two
charges has electrostatic self-interaction. Thus it is reasonable to expect that this assumption also holds true
for individual quantum systems.
18
It has been argued that the existence of a gravitational self-interaction term in the Schrödinger-Newton
equation does not have a consistent Born rule interpretation (Adler 2007). The reason is that the probability
of simultaneously finding a particle in different positions is zero.

27
 2.5. THE PHYSICAL ORIGIN OF MASS AND CHARGE DENSITY

represents the strength of the influence of the self-interaction on the normal evolution of the
wave function; when ε2 ≈ 1 the influence is significant. Similarly, for a free charged system
with charge Q, the measure of the potential strength of the electrostatic self-interaction is
2
ε2 = ( 4kQ 2
~c ) . As a typical example, for a free electron the potential strength of the electro-
2
static self-interaction will be ε2 = ( 4ke 2 −3
~c ) ≈ 1 × 10 . This indicates that the electrostatic
self-interaction will have a remarkable influence on the evolution of the wave function of a
free electron19 . If such an interaction indeed exists, it should have been detected by precise
interference experiments on electrons. As another example, consider the electron in the hy-
drogen atom. Since the potential of the electrostatic self-interaction is of the same order as
the Coulomb potential produced by the nucleus, the energy levels of hydrogen atoms will
be remarkably different from those predicted by quantum mechanics and confirmed by ex-
periments. Therefore, the electrostatic self-interaction cannot exist for a charged quantum
system such as an electron.
To sum up, the superposition principle of quantum mechanics requires that the mass and
charge distribution of a quantum system such as an electron is not real but effective; at every
instant there is only a localized particle with the total mass and charge of the system, while
during an infinitesimal time interval the time average of the ergodic motion of the particle
forms the effective mass and charge distribution20 , and the mass and charge density in each
position is proportional to the modulus square of the wave function of the system there.

2.5.2 The ergodic motion of a particle is discontinuous

Which sort of ergodic motion then? If the ergodic motion of a particle is continuous, then
it can only form the effective mass and charge density during a finite time interval. But
the effective mass and charge density is required to be formed by the ergodic motion of the
particle during an infinitesimal time interval (not during a finite time interval) at a given
instant. Thus it seems that the ergodic motion of the particle cannot be continuous. This
is at least what the existing quantum mechanics says. However, there may exist a possible
loophole here. Although the classical ergodic models that assume continuous motion are
inconsistent with quantum mechanics due to the existence of a finite ergodic time, they may
be not completely precluded by experiments if only the ergodic time is extremely short. After
all quantum mechanics is only an approximation of a more fundamental theory of quantum
gravity, in which there may exist a minimum time scale such as the Planck time. Therefore,
we need to investigate the classical ergodic models more thoroughly.
Consider an electron in a one-dimensional box in the first excited state ψ(x) (Aharonov
and Vaidman 1993). Its wave function has a node at the center of the box, where its charge
density is zero. Assume the electron performs a very fast continuous motion in the box, and
during a very short time interval its motion generates an effective charge distribution. Let’s
see whether this distribution can assume the same form as e|ψ(x)|2 , which is required by
protective measurement21 . Since the effective charge density is proportional to the amount
19
By contrast, the potential strength of the gravitational self-interaction for a free electron is ε2 =
4Gm2
( ~c e )2 ≈ 4×10−89 . This kind of gravitational self-interaction is too weak to be detected presently (Salzman
and Carlip 2006).
20
At a particular time the charge density is either zero (if the electron is not there) or singular (if the
electron is inside the infinitesimally small region including the space point in question).
21
Note that in Nelson’s stochastic mechanics, the electron, which is assumed to undergo a Brownian motion,
moves only within a region bounded by the nodes (Nelson 1966). This ensures that the theory can be
equivalent to quantum mechanics in a limited sense. Obviously this sort of motion is not ergodic and cannot
generate the required charge distribution. This conclusion also holds true for the motion of particles in some
variants of stochastic mechanics (Bell 1986b; Vink 1993; Barrett, Leifer and Tumulka 2005), as well as in the
de Broglie-Bohm theory (de Broglie 1928; Bohm 1952).

28
 2.5. THE PHYSICAL ORIGIN OF MASS AND CHARGE DENSITY

of time the electron spends in a given position, the electron must be in the left half of the
box half of the time and in the right half of the box half of the time. But it can spend no
time at the center of the box where the effective charge density is zero; in other words, it
must move at infinite velocity at the center. Certainly, the appearance of velocities faster
than light or even infinite velocities may be not a fatal problem, as our discussion is entirely
in the context of non-relativistic quantum mechanics, and especially the infinite potential
in the example is also an ideal situation. However, it seems difficult to explain why the
electron speeds up at the node and where the infinite energy required for the acceleration
comes from. Moreover, the sudden acceleration of the electron near the node may also
result in large radiation (Aharonov, Anandan and Vaidman 1993), which is inconsistent
with the predictions of quantum mechanics. Again, it seems very difficult to explain why
the accelerating electron does not radiate here.
Let’s further consider an electron in a superposition of two energy eigenstates in two
boxes ψ1 (x) + ψ2 (x). In this example, even if one assumes that the electron can move with
infinite velocity (e.g. at the nodes), it cannot continuously move from one box to another
due to the restriction of box walls. Therefore, any sort of continuous motion cannot generate
the effective charge distribution e|ψ1 (x) + ψ2 (x)|2 . One may still object that this is merely
an artifact of the idealization of infinite potential. However, even in this ideal situation, the
model should also be able to generate the effective charge distribution by means of some sort
of ergodic motion of the electron; otherwise it will be inconsistent with quantum mechanics.
On the other hand, it is very common in quantum optics experiments that a single-photon
wave packet is split into two branches moving along two well separated paths in space. The
wave function of the photon disappears outside the two paths for all practical purposes.
Moreover, the experimental results are not influenced by the environment and experimental
setup between the two paths of the photon. Thus it is very difficult to imagine that the
photon performs a continuous ergodic motion back and forth in the space between its two
paths.
In view of these serious drawbacks of the classical ergodic models and their inconsis-
tency with quantum mechanics, we conclude that the ergodic motion of particles cannot be
continuous. If the motion of a particle is discontinuous, then the particle can readily move
throughout all regions where the wave function is nonzero during an arbitrarily short time
interval at a given instant. Furthermore, if the probability density of the particle appearing
in each position is proportional to the modulus square of its wave function there at every
instant, the discontinuous motion can also generate the right mass and charge distribution.
This will solve the above problems plagued by the classical ergodic models. The discon-
tinuous ergodic motion requires no existence of a finite ergodic time. Moreover, a particle
undergoing discontinuous motion can also move from one region to another spatially sepa-
rated region, no matter whether there is an infinite potential wall between them, and such
discontinuous motion is not influenced by the environment and experimental setup between
these regions either. Besides, discontinuous motion can also solve the problems of infinite
velocity and accelerating radiation. The reason is that no classical velocity and accelera-
tion can be defined for discontinuous motion, and energy and momentum will require new
definitions and understandings as in quantum mechanics.
In summary, we have argued that the mass and charge distribution of a quantum system,
which can be measured by protective measurement, is not real but effective. Moreover, the
effective mass and charge distribution is formed by the discontinuous motion of a localized
particle, and the probability density of the particle appearing in each position is proportional
to the modulus square of its wave function there.

29
 2.5. THE PHYSICAL ORIGIN OF MASS AND CHARGE DENSITY

2.5.3 An argument for random discontinuous motion

Although the above analysis demonstrates that the ergodic motion of a particle is discon-
tinuous, it doesn’t say that the discontinuous motion must be random. In particular, the
randomness of the result of a quantum measurement may be only apparent. In order to
know whether the motion of particles is random or not, we need to analyze the cause of
motion. For example, if motion has no deterministic cause, then it will be random, only
determined by a probabilistic cause. This may also be the right way to find how particles
move. Since motion involves change in position, if we can find the cause or instantaneous
condition determining the change22 , we will be able to find how particles move in reality.
Let’s consider the simplest states of motion of a free particle, for which the instantaneous
condition determining the change of its position is a constant during the motion. In logic
the instantaneous condition can only be deterministic or indeterministic. That the instanta-
neous condition is deterministic means that it leads to a deterministic change of the position
of the particle at a given instant. That the instantaneous condition is indeterministic means
that it only determines the probability of the particle appearing in each position in space at
a given instant. If the instantaneous condition is deterministic, then the simplest states of
motion of the free particle will have two possible forms. The first one is continuous motion
with constant velocity, and the equation of motion of the particle is x(t + dt) = x(t) + vdt,
where the deterministic instantaneous condition v is a constant23 . The second one is dis-
continuous motion with infinite average velocity; the particle performs a finite jump along
a fixed direction at every instant, where the jump distance is a constant, determined by the
constant instantaneous condition24 . On the other hand, if the instantaneous condition is
indeterministic, then the simplest states of motion of the free particle will be random discon-
tinuous motion with even position probability distribution. At each instant the probability
density of the particle appearing in every position is the same.
In order to know whether the instantaneous condition is deterministic or not, we need
to determine which sort of simplest states of motion are the solutions of the equation of
free motion in quantum mechanics (i.e. the free Schrödinger equation)25 . According to the
analysis in the last subsection, the momentum eigenstates of a free particle, which are the
solutions of the free Schrödinger equation, describe the ergodic motion of the particle with
even position probability distribution in space. Therefore, the simplest states of motion
with a constant probabilistic instantaneous condition are the solutions of the equation of
free motion, while the simplest states of motion with a constant deterministic instantaneous
condition are not.
When assuming that (1) the simplest states of motion of a free particle are the solutions
of the equation of free motion; and (2) the instantaneous condition determining the position
change of a particle is always deterministic or indeterministic for any state of motion, the
above result then implies that motion, no matter whether it is free or forced, has no deter-
ministic cause, and thus it is random and discontinuous, only determined by a probabilistic
cause. The argument may be improved by further analyzing these two seemingly reasonable
assumptions, but we will leave this for future work.
22
The word “cause” used here only denotes a certain instantaneous condition determining the change of
position, which may appear in the laws of motion. Our analysis is irrelevant to whether the condition has
causal power or not.
23
This deterministic instantaneous condition has been often called intrinsic velocity (Tooley 1988).
24
In discrete space and time, the motion will be a discrete jump across space along a fixed direction at each
time unit, and thus it will become continuous motion with constant velocity in the continuous limit.
25
In the next chapter, we will derive this equation of free motion from fundamental physical principles.
This will make the argument given here more complete. Moreover, the derivation itself may also provide an
argument for discontinuous motion that does not resort to direct experience, as the equation of free motion
does not permit the persisting existence of the local state of continuous motion. For details see Section 3.4.

30
 2.6. THE WAVE FUNCTION AS A DESCRIPTION OF RANDOM
DISCONTINUOUS MOTION OF PARTICLES

2.6 The wave function as a description of random discontinuous

motion of particles
The wavefunction gives not the density of stuff, but gives rather (on squaring its modulus)
the density of probability. Probability of what, exactly? Not of the electron being there,
but of the electron being found there, if its position is ‘measured’. Why this aversion
to ‘being’ and insistence on ‘finding’ ? The founding fathers were unable to form a clear
picture of things on the remote atomic scale. — John Bell, 1990

In classical mechanics, we have a clear physical picture of motion. It is well understood

that the trajectory function x(t) in classical mechanics describes the continuous motion of a
particle. In quantum mechanics, the trajectory function x(t) is replaced by a wave function
ψ(x, t). If the particle ontology is still viable in the quantum domain, then it seems natural
that the wave function should describe some sort of more fundamental motion of particles,
of which continuous motion is only an approximation in the classical domain, as quantum
mechanics is a more fundamental theory of the physical world, of which classical mechanics
is an approximation. The analysis in the last section provides a strong support for this
conjecture. It shows that a quantum system such as an electron is a localized particle
that undergoes random discontinuous motion, and the probability density of the particle
appearing in each position is proportional to the modulus square of its wave function there.
As a result, the wave function in quantum mechanics can be regarded as a description of the
more fundamental motion of particles, which is essentially discontinuous and random. In
this section, we will give a more detailed analysis of random discontinuous motion and the
meaning of the wave function (Gao 1993, 1999, 2000, 2003, 2006b, 2008, 2011a, 2011b).

2.6.1 An analysis of random discontinuous motion of particles

Let’s first make clearer what we mean when we say a quantum system such as an electron
is a particle. The picture of particles appears from our analysis of the mass and charge
density of a quantum system. As we have shown in the last section, the mass and charge
density of an electron, which is measurable by protective measurement and proportional to
the modulus square of its wave function, is not real but effective; it is formed by the ergodic
motion of a localized particle with the total mass and charge of the electron. If the mass and
charge density is real, i.e., if the mass and charge distributions at different locations exist
at the same time, then there will exist gravitational and electrostatic interactions between
the distributions, the existence of which not only contradicts experiments but also violates
the superposition principle of quantum mechanics. It is this analysis that reveals the basic
existent form of a quantum system such as an electron in space and time. An electron is
a particle26 . Here the concept of particle is used in its usual sense. A particle is a small
localized object with mass and charge, and it is only in one position in space at an instant.
However, as we have argued above, the motion of an electron described by its wave function
is not continuous but discontinuous and random in nature. We may say that an electron is a
quantum particle in the sense that its motion is not continuous motion described by classical
mechanics, but random discontinuous motion described by quantum mechanics.
Next, let’s analyze the random discontinuous motion of particles. From a logical point of
view, for the random discontinuous motion of a particle, the particle must have an instanta-
neous property (as a probabilistic instantaneous condition) that determines the probability
density of it appearing in every position in space; otherwise the particle would not “know”
how frequently it should appear in each position in space. This property is usually called
26
However, the analysis cannot tell us the precise size and possible structure of an electron.

31
 2.6. THE WAVE FUNCTION AS A DESCRIPTION OF RANDOM
DISCONTINUOUS MOTION OF PARTICLES

indeterministic disposition or propensity in the literature27 , and it can be represented by

%(x,
R +∞t), which satisfies the nonnegative condition %(x, t) > 0 and the normalization relation
−∞ %(x, t)dx = 1. As a result, the position of the particle at every instant is random, and
its trajectory formed by the random position series is also discontinuous at every instant28 .
Unlike the deterministic continuous motion, the trajectory function x(t) no longer pro-
vides a useful description for random discontinuous motion. In the following, we will give a
strict description of random discontinuous motion of particles based on measure theory. For
simplicity but without losing generality, we will mainly analyze the one-dimensional motion
that corresponds to the point set in two-dimensional space and time. The results can be
readily extended to the three-dimensional situation.

Fig.2 The description of random discontinuous motion of a single particle

We first analyze the random discontinuous motion of a single particle. Consider the
state of motion of the particle in finite intervals ∆t and ∆x near a space-time point (ti ,xj )
as shown in Fig. 2. The positions of the particle form a random, discontinuous trajectory
in this square region29 . We study the projection of this trajectory in the t-axis, which is
a dense instant set in the time interval ∆t. Let W be the discontinuous trajectory of the
particle and Q be the square region [xj , xj + ∆x] × [ti , ti + ∆t]. The dense instant set can
be denoted by πt (W ∩ Q) ∈ <, where πt is the projection on the t-axis. According to the
measure theory, we can define the Lebesgue measure:
Z
M∆x,∆t (xj , ti ) = dt. (2.33)
πt (W ∩Q)∈<

Since the sum of the measures of all such dense instant sets in the time interval ∆t is equal
to the length of the continuous time interval ∆t, we have:
27
Note that the propensity here denotes single case propensity. For long run propensity theories fail to
explain objective single-case probabilities. According to these theories, it makes no sense to speak of the
propensity of a single isolated event in the absence of a sequence that contains it. For a helpful analysis
of the single-case propensity interpretation of probability in GRW theory see Frigg and Hoefer (2007). In
addition, it is worth stressing that the propensities possessed by particles relate to their objective motion, not
to the measurements on them. By contrast, according to the existing propensity interpretations of quantum
mechanics, the propensities a quantum system has relate only to measurements; a quantum system possesses
the propensity to exhibit a particular value of an observable if the observable is measured on the system (see
Suárez 2004, 2007 for a comprehensive analysis). Like the Copenhagen interpretation of quantum mechanics,
these interpretations cannot be wholly satisfactory because of resorting to the vague concept of measurement.
28
However, there is an exception. When the probability density function is a special δ-function such as
δ(x−x(t)), where x(t) is a continuous function of t, the motion of the particle is deterministic and continuous.
In addition, even for a general probability density function it is still possible that the random position series
forms a continuous trajectory, though the happening probability is zero.
29
Recall that a trajectory function x(t) is essentially discontinuous if it is not continuous at every instant
t. A trajectory function x(t) is continuous if and only if for every t and every real number ε > 0, there exists
a real number δ > 0 such that whenever a point t0 has distance less than δ to t, the point x(t0 ) has distance
less than ε to x(t).

32
 2.6. THE WAVE FUNCTION AS A DESCRIPTION OF RANDOM
DISCONTINUOUS MOTION OF PARTICLES

X
M∆x,∆t (xj , ti ) = ∆t. (2.34)
j

Then we can define the measure density as follows:

ρ(x, t) = lim M∆x,∆t (x, t)/(∆x · ∆t). (2.35)

∆x,∆t→0

This quantity provides a strict description of the position distribution of the particle or the
relative frequency of the particle appearing in an infinitesimal space interval dx near position
x during an infinitesimal interval dt near instant t, and it satisfies the normalization relation
R +∞
−∞ ρ(x, t)dx = 1 by Eq. (2.34). Note that the existence of the limit relies on the continuity
of the evolution of %(x, t), the property of the particle that determines the probability density
of it appearing in every position in space. In fact, ρ(x, t) is determined by %(x, t), and there
exists the relation ρ(x, t) = %(x, t). We call ρ(x, t) position measure density or position
density in brief.
Since the position density ρ(x, t) changes with time in general, we may further define the
position flux density j(x, t) through the relation j(x, t) = ρ(x, t)v(x, t), where v(x, t) is the
velocity of the local position density. It describes the change rate of the position density.
Due to the conservation of measure, ρ(x, t) and j(x, t) satisfy the continuity equation:

∂ρ(x, t) ∂j(x, t)
+ = 0. (2.36)
∂t ∂x
The position density ρ(x, t) and position flux density j(x, t) provide a complete description
of the state of random discontinuous motion of a single particle30 .
The description of the motion of a single particle can be extended to the motion of many
particles. For the random discontinuous motion of N particles, we can define joint position
density ρ(x1 , x2 , ...xN , t) and joint position flux density j(x1 , x2 , ...xN , t) = ρ(x1 , x2 , ...xN , t)
v(x1 , x2 , ...xN , t). They also satisfy the continuity equation:
N
∂ρ(x1 , x2 , ...xN , t) X ∂j(x1 , x2 , ...xN , t)
+ = 0. (2.37)
∂t ∂xi
i=1

When these N particles are independent, the joint position density can be reduced to
the
QN direct product of the position density for each particle, namely ρ(x1 , x2 , ...xN , t) =
i=1 ρ(xi , t). Note that the joint position density ρ(x1 , x2 , ...xN , t) and joint position flux
density j(x1 , x2 , ...xN , t) are not defined in the real three-dimensional space, but defined in
the 3N-dimensional configuration space.

2.6.2 Interpreting the wave function

Although the motion of particles is essentially discontinuous and random, the discontinuity
and randomness of motion are absorbed into the state of motion, which is defined during
an infinitesimal time interval and represented by the position density ρ(x, t) and position
flux density j(x, t). Therefore, the evolution of the state of random discontinuous motion of
particles may obey a deterministic continuous equation. By assuming that the nonrelativistic
30
It is also possible that the position density ρ(x, t) alone provides a complete description of the state of
random discontinuous motion of a particle. Which one is right depends on the laws of motion. As we will see
later, quantum mechanics requires that a complete description of the state of random discontinuous motion
of particles includes both the position density and the position flux density.

33
 2.6. THE WAVE FUNCTION AS A DESCRIPTION OF RANDOM
DISCONTINUOUS MOTION OF PARTICLES

equation of random discontinuous motion is the Schrödinger equation in quantum mechanics,

both ρ(x, t) and j(x, t) can be expressed by the wave function in a unique way31 :

ρ(x, t) = |ψ(x, t)|2 , (2.38)

~ ∂ψ(x, t) ∂ψ ∗ (x, t)
j(x, t) = [ψ ∗ (x, t) − ψ(x, t) ]. (2.39)
2mi ∂x ∂x
Correspondingly, the wave function ψ(x, t) can be uniquely expressed by ρ(x, t) and j(x, t)
(except for a constant phase factor):
j(x0 ,t)
im x dx0 /~
p R
ψ(x, t) = ρ(x, t)e −∞ ρ(x0 ,t) . (2.40)
In this way, the wave function ψ(x, t) also provides a complete description of the state of
random discontinuous motion of particles. For the motion of many particles, the joint posi-
tion density and joint position flux density are defined in the 3N-dimensional configuration
space, and thus the many-particle wave function, which is composed of these two quantities,
is also defined in the 3N-dimensional configuration space.
Interestingly, we can reverse the above logic in some sense, namely by assuming the wave
function is a complete objective description for the motion of particles, we can also reach
the random discontinuous motion of particles, independent of our previous analysis. If the
wave function ψ(x, t) is a complete description of the state of motion for a single particle,
then the quantity |ψ(x, t)|2 dx will not only give the probability of the particle being found
in an infinitesimal space interval dx near position x at instant t (as required by quantum
mechanics), but also give the objective probability of the particle being there at the instant.
This accords with the common-sense assumption that the probability distribution of the
measurement results of a property is the same as the objective distribution of the values
of the property in the measured state. Then at instant t the particle will be in a random
position where the probability density |ψ(x, t)|2 is nonzero, and during an infinitesimal time
interval near instant t it will move throughout the whole region where the wave function
ψ(x, t) spreads. Moreover, its position density in each position is equal to the probability
density there. Obviously this kind of motion is random and discontinuous.
One important point needs to be pointed out here. Since the wave function in quantum
mechanics is defined at an instant, not during an infinitesimal time interval, it should be
regarded not simply as a description of the state of random discontinuous motion of particles,
but more suitably as a description of the intrinsic property of the particles that determines
their random discontinuous motion at a deeper level32 . In particular, the modulus square
of the wave function represents the property that determines the probability density of the
particles appearing in certain positions in space at a given instant (this means %(x, t) ≡
|ψ(x, t)|2 ). By contrast, the position density and position flux density, which are defined
during an infinitesimal time interval at a given instant, are only a description of the state
of the resulting random discontinuous motion of particles, and they are determined by the
wave function. In this sense, we may say that the motion of particles is “guided” by their
wave function in a probabilistic way.
We have been discussed random discontinuous motion of particles in real space. The
picture of random discontinuous motion may exist not only for position but also for other
31
Note that the relation between j(x, t) and ψ(x, t) depends on the concrete evolution under an external
potential such as electromagnetic vector potential. By contrast, the relation ρ(x, t) = |ψ(x, t)|2 holds true
universally, independent of the concrete evolution.
32
For a many-particle system in an entangled state, this property is possessed by the whole system. See
Chapter 5 for a detailed analysis of the physical picture of quantum entanglement.

34
 2.6. THE WAVE FUNCTION AS A DESCRIPTION OF RANDOM
DISCONTINUOUS MOTION OF PARTICLES

dynamical variables such as momentum and energy, and thus the interpretation of the wave
function in position space may also apply to the wave function in momentum space etc. This
means that if a particle is in a superposition of the eigenstates of a variable, it will undergo
random discontinuous motion among the corresponding eigenvalues of this variable33 . How-
ever, we note that spin is a distinct property. Since the spin of a free particle is always
definite along one direction, the spin of the particle does not undergo random discontinuous
motion, though a spin eigenstate along one direction can always be decomposed into two
different spin eigenstates along another direction. But if the spin state of a particle is en-
tangled with its spatial state due to interaction and the branches of the entangled state are
well separated in space, the particle in different branches will have different spin, and it will
also undergo random discontinuous motion between these different spin states. This is the
situation that usually happens during a spin measurement.

33
Since the values of two noncommutative variables (e.g. position and momentum) at every instant are
random and mutually independent, the objective value distribution of every variable can be equal to the
modulus square of its respective wave function. Besides, it is worth stressing that for random discontinuous
motion the property of a particle in a superposition of the eigenstates of the property is indeterminate in the
sense of usual hidden variables, though it does have a definite value at every instant. This makes the theorems
that restrict hidden variables such as the Kochen-Specker theorem (Kochen and Specker 1967) irrelevant.

35
 3
The motion of particles follows probability law but the proba-
bility itself propagates according to the law of causality.
— Max Born, 1926

Schrödinger’s Equation and the Conservation Laws

After investigating the physical meaning of the wave function, we will further analyze the
linear evolution law for the wave function in this chapter. It is demonstrated that the
linear non-relativistic evolution of the wave function of an isolated system obeys the free
Schrödinger equation due to the requirements of spacetime translation invariance and rel-
ativistic invariance. In addition, we also investigate the meaning and implications of the
conservation laws in quantum mechanics.
Many quantum mechanics textbooks provide a heuristic “derivation” of the Schrödinger
equation. It begins with the assumption that the state of a free quantum system has the form
of a plane wave ei(kx−ωt) . When combining with the de Broglie relations for momentum and
energy p = ~k and E = ~ω, this state becomes ei(px−Et)/~ . Then it uses the nonrelativistic
energy-momentum relation E = p2 /2m to obtain the free particle Schrödinger equation.
Lastly, this equation is generalized to include an external potential, and the end result is the
Schrödinger equation.
In the following sections, we will show that the heuristic “derivation” of the free Schrödinger
equation can be turned into a real derivation by resorting to spacetime translation invariance
and relativistic invariance1 . Spacetime translation gives the definitions of momentum and
energy, and spacetime translation invariance entails that the state of a free quantum system
with definite momentum and energy assumes the plane wave form ei(px−Et)/~ . Moreover, the
relativistic invariance of the free states further determines the relativistic energy-momentum
relation, whose nonrelativistic approximation is E = p2 /2m. Though the requirements
of these invariances are already well known, an explicit and complete derivation of the free
Schrödinger equation using them seems still missing in the literature and textbooks. The new
integrated analysis may not only help to understand the physical origin of the Schrödinger
equation, but also help to understand momentum and energy and their conservation for
random discontinuous motion of particles.
1
There have been some attempts to derive the Schrödinger equation from Newtonian mechanics, one typical
example of which is Nelson’s stochastic mechanics (Nelson 1966). However, it has been argued that Nelson’s
derivation is problematic, and in particular, stochastic mechanics is inconsistent with quantum mechanics
(Glabert, Hänggi and Talkner 1979; Wallstrom 1994). Glabert, Hänggi and Talkner (1979) argued that the
Schrödinger equation is not equivalent to a Markovian process, and the various correlation functions used in
quantum mechanics do not have the properties of the correlations of a classical stochastic process. Wallstrom
(1994) further showed that one must add by hand a quantization condition, as in the old quantum theory, in
order to recover the Schrödinger equation, and thus the Schrödinger equation and the Madelung hydrodynamic
equations are not equivalent. In addition, Nelson (2005) also showed that there is an empirical difference
between the predictions of quantum mechanics and his stochastic mechanics when considering quantum
entanglement and nonlocality. For example, for two widely-separated but entangled harmonic oscillators,
the two theories predict totally different statistics; stochastic mechanics predicts that measurements of the
position of the first one at time T (oscillation period) and the position of the second one at time 0 do not
interfere with each other, while quantum mechanics predicts that there exists a strong correlation between
them.

36
 3.1. SPACETIME TRANSLATION AND ITS INVARIANCE

3.1 Spacetime translation and its invariance

In this section, we will show that the free states of motion for a quantum system can
be basically determined by spacetime translation invariance. The spacetime translation
invariance of natural laws reflects the homogeneity of space and time. The homogeneity of
space ensures that the same experiment performed at two different places gives the same
result, and the homogeneity in time ensures that the same experiment repeated at two
different times gives the same result. There are in general two different pictures of translation:
active transformation and passive transformation. The active transformation corresponds
to displacing the studied system, and the passive transformation corresponds to moving the
coordinate system. Physically, the equivalence of the active and passive pictures is due to
the fact that moving the system one way is equivalent to moving the coordinate system the
other way by an equal amount (see also Shankar 1994). In the following, we will mainly
analyze spacetime translations in terms of active transformations.
A space translation operator can be defined as

T (a)ψ(x, t) = ψ(x − a, t). (3.1)

It means translating rigidly the state of a system, ψ(x, t), by an amount
R∞ a in the positive
x direction. The operator preserves the norm of the state because −∞ ψ ∗ (x, t)ψ(x, t)dx =
R∞ ∗ †
−∞ ψ (x − a, t)ψ(x − a, t)dx. This implies that T (a) is unitary, satisfying T (a)T (a) = I.
As a unitary operator, T (a) can be further expressed as

T (a) = e−iaP , (3.2)

where P is called the generator of space translation. It can be seen that P is Hermitian, and
its eigenvalues are real. By expanding ψ(x − a, t) in order of a, we can further get

∂
P = −i
. (3.3)
∂x
Similarly, a time translation operator can be defined as

U (t)ψ(x, 0) = ψ(x, t). (3.4)

Let the evolution equation of state be of the following form:

∂ψ(x, t)
i = Hψ(x, t). (3.5)
∂t
where H is a to-be-determined operator that depends on the properties of the system. In
the following analysis of this section, we assume H is independent of the evolved state,
namely the evolution is linear2 . Then the time translation operator U (t) can be expressed
as U (t) = e−itH , and H is the generator of time translation. Note that we cannot determine
whether U (t) is unitary and H is Hermitian here.
Let’s now analyze the implications of spacetime translation invariance for the laws of
motion of a free system or an isolated system. First, time translational invariance requires
that H has no time dependence, namely dH/dt = 0. This can be demonstrated as follows
(Shankar 1994, p.295). Suppose an isolated system is in state ψ0 at time t1 and evolves for
an infinitesimal time δt. The state of the system at time t1 + δt, to first order in δt, will be

ψ(x, t1 + δt) = [I − iδtH(t1 )]ψ0 . (3.6)

2
This is an important presupposition in our derivation. We will consider the possible cases where H is
nonlinear in the next section.

37
 3.1. SPACETIME TRANSLATION AND ITS INVARIANCE

If the evolution is repeated at time t2 , beginning with the same initial state, the state at
t2 + δt will be

ψ(x, t2 + δt) = [I − iδtH(t2 )]ψ0 . (3.7)

Time translational invariance requires the outcome state should be the same:

ψ(x, t2 + δt) − ψ(x, t1 + δt) = iδt[H(t1 ) − H(t2 )]ψ0 = 0. (3.8)

Since the initial state ψ0 is arbitrary, it follows that H(t1 ) = H(t2 ). Moreover, since t1 and
t2 are also arbitrary, it follows that H is time-independent, namely dH/dt = 0. It can be
seen that this result relies on the linearity of evolution. If H depends on the state, then
obviously we cannot obtain dH/dt = 0 because the state is time-dependent, though we still
have H(t1 , ψ0 ) = H(t2 , ψ0 ), which means that the state-dependent H also satisfies time
translational invariance.
Secondly, space translational invariance requires [T (a), U (t)] = 0, which further leads to
[P, H] = 0. This can be demonstrated as follows (Shankar 1994, p.293). Suppose at t = 0
two observers A and B prepare identical isolated systems at x = 0 and x = a, respectively.
Let ψ(x, 0) be the state of the system prepared by A. Then T (a)ψ(x, 0) is the state of
the system prepared by B, which is obtained by translating (without distortion) the state
ψ(x, 0) by an amount a to the right. The two systems look identical to the observers who
prepared them. After time t, the states evolve into U (t)ψ(x, 0) and U (t)T (a)ψ(x, 0). Since
the time evolution of each identical system at different places should appear the same to
the local observers, the above two systems, which differed only by a spatial translation at
t = 0, should differ only by the same spatial translation at future times. Thus the state
U (t)T (a)ψ(x, 0) should be the translated version of A’s system at time t, namely we have
U (t)T (a)ψ(x, 0) = T (a)U (t)ψ(x, 0). This relation holds true for any initial state ψ(x, 0),
and thus we have [T (a), U (t)] = 0, which says that space translation operator and time
translation operator are commutative. Again, we note that the linearity of evolution is
an important presupposition of this result. If U (t) depends on the state, then the space
translational invariance will only lead to U (t, T ψ)T (a)ψ(x, 0) = T (a)U (t, ψ)ψ(x, 0), from
which we cannot obtain [T (a), U (t)] = 0.
When dH/dt = 0, the solutions of the evolution equation Eq.(3.5) assume the following
form

ψ(x, t) = ϕE (x)e−iEt , (3.9)

where E is a constant, and ϕE (x) is the eigenstate of H and satisfies the time-independent
equation:

HϕE (x) = EϕE (x). (3.10)

The commutative relation [P, H] = 0 further implies that P and H have common eigenstates.
∂
This means that ϕE (x) is also the eigenstate of P . Since the eigenstate of P ≡ −i ∂x is eipx
(except a normalization factor), where p is a real eigenvalue, the solutions of the evolution
equation Eq.(3.5) for an isolated system will be ei(px−Et) . In quantum mechanics, P and
H, the generators of space translation and time translation, are also called momentum
operator and energy operator, respectively. Correspondingly, ei(px−Et) is the eigenstate of
both momentum and energy, and p and E are the corresponding momentum and energy
eigenvalues, respectively. In other words, the state ei(px−Et) describes an isolated system
(e.g. a free electron) with definite momentum p and energy E.

38
 3.2. RELATIVISTIC INVARIANCE

3.2 Relativistic invariance

The relation between momentum p and energy E can be determined by the relativistic
invariance of the momentum eigenstate ei(px−Et) , and it turns out to be E 2 = p2 c2 + m2 c4 ,
where m is the mass of the system, and c is the speed of light3 . In the nonrelativistic domain,
the energy-momentum relation reduces to E = p2 /2m.
Consider two inertial frames S0 and S with coordinates x0 , t0 and x, t. S0 is moving with
velocity v relative to S. Then x, t and x0 , t0 satisfy the Lorentz transformations:
x − vt
x0 = p , (3.11)
1 − v 2 /c2
t − xv/c2
t0 = p . (3.12)
1 − v 2 /c2
Suppose the state of a free particle is ψ = ei(p0 x0 −E0 t0 ) , an eigenstate of P , in S0 , where
p0 , E0 is the momentum and energy of the particle in S0 , respectively. When described in S
by coordinates x, t, the state is
t−xv/c2 p0 +E0 v/c 2 E +p v
x−vt
i(p0 √ −E0 √ ) i( √ x− √ 0 20 2 t)
ψ=e 1−v 2 /c2 1−v 2 /c2 =e 2 2
1−v /c 1−v /c . (3.13)
This means that in frame S the state is still the eigenstate of P , and the corresponding
momentum p and energy E are4

p0 + E0 v/c2
p= p , (3.14)
1 − v 2 /c2
E0 + p0 v
E=p . (3.15)
1 − v 2 /c2
We further suppose that the particle is at rest in frame S0 . Then the velocity of the
particle is v in frame S.5 Considering that the velocity of a particle in the momentum
eigenstate ei(px−Et) or a wavepacket superposed by these eigenstates is defined as the group
velocity of the wavepacket, namely

dE
u= , (3.16)
dp
we have

dE0 /dp0 = 0, (3.17)

3
Different from the derivation given here, most existing derivations of the energy-momentum relation
are based on somewhat complex analysis of an elastic collision process. Moreover, they resort to either
Newtonian limit (e.g. p = mv) or less fundamental relation (e.g. p = Eu/c2 ) or even mathematical intuition
(e.g. four-vectors) (see Sonego and Pin 2005 and references therein).
4
Alternatively we can obtain the transformations of momentum and energy by directly requiring the
relativistic invariance of the momentum eigenstate ei(px−Et) , which leads to the relation px−Et = p0 x0 −E0 t0 .
Note that any superposition of momentum eigenstates is also invariant under the coordinates transformation.
The reason is that it is a scalar that describes the physical state of a quantum system (except an absolute
phase), and when observed in different reference frames it should be the same. This also means that the
state evolution equation must be relativistically invariant. However, if the relativistically invariant equation
is replaced by the nonrelativistic approximation such as the Schrödinger equation, the state will no longer
satisfy the relativistic invariance.
5
We can also obtain this result from the definition Eq. (3.16) by using the transformations of momentum
and energy Eq.(3.14) and Eq.(3.15).

39
 3.3. DERIVATION OF THE FREE SCHRÖDINGER EQUATION

dE/dp = v. (3.18)
Eq.(3.17) means that E0 and p0 are independent. Moreover, since the particle is at rest in
S0 , E0 and p0 do not depend on v. By differentiating both sides of Eq.(3.14) and Eq.(3.15)
relative to v we obtain

dp v p0 + E0 v/c2 E0 /c2
= 2 + 1 , (3.19)
dv c (1 − v 2 /c2 ) 23 (1 − v 2 /c2 ) 2
dE v E0 + p0 v p0
= 2 3 + 1 . (3.20)
dv c (1 − v 2 /c2 ) 2 (1 − v 2 /c2 ) 2
Dividing Eq.(3.20) by Eq.(3.19) and using Eq.(3.18) we obtain
p0
p = 0. (3.21)
1 − v 2 /c2
This means that p0 = 0. Inputting this important result into Eq.(3.15) and Eq.(3.14), we
immediately obtain

E0
E=p , (3.22)
1 − v 2 /c2
E0 v/c2
p= p . (3.23)
1 − v 2 /c2
Then the energy-momentum relation is:

E 2 = p2 c2 + E02 , (3.24)
where E0 is the energy of the particle at rest, called rest energy of the particle, and p and E
is the momentum and energy of the particle with velocity v. By defining m = E0 /c2 as the
(rest) mass of the particle6 , we can further obtain the familiar energy-momentum relation

E 2 = p2 c2 + m2 c4 . (3.25)
In the nonrelativistic domain, this energy-momentum relation reduces to E = p2 /2m.

3.3 Derivation of the free Schrödinger equation

The relation between energy E and momentum p for momentum eigenstates in the non-
relativistic domain implies that the operator relation is H = P 2 /2m for an isolated system,
where H is the free Hamiltonian of the system. Note that since the value of E is real by
Eq.(3.24), H is Hermitian and U (t) is unitary for free evolution. By inputting this opera-
tor relation into the evolution equation Eq.(3.5), we can obtain the free evolution equation,
which assumes the same form as the free particle Schrödinger equation7 :

∂ψ(x, t) 1 ∂ 2 ψ(x, t)
i =− . (3.26)
∂t 2m ∂x2
6
According to the analysis given here, it seems that we can in principle avoid talking about mass in modern
physics from a more fundamental point of view (cf. Okun 2009).
7
This also means that the Klein-Gordon equation can be derived in the relativistic domain when assuming
that the wave function is a number function.

40
 3.4. FURTHER DISCUSSIONS

It is worth noting that, unlike the free particle Schrödinger equation, the reduced Planck
constant ~ with dimension of action is missing in this equation. However, this is not a
problem. The reason is that the dimension of ~ can be absorbed into the dimension of the
mass m. For example, we can stipulate the dimensional relations as p = 1/L, E = 1/T and
m = T /L2 , where L and T represents the dimensions of space and time, respectively (see
Duff, Okun and Veneziano 2002 for more discussions). Moreover, the value of ~ can be set
to the unit of number 1 in principle. Thus the above equation is essentially the free particle
Schrödinger equation in quantum mechanics.
By using the definition of classical potential and requiring an appropriate expectation
value correspondence, dhP i/dt = −h∂V /∂xi, we can further obtain the Schrödinger equation
under an external potential V (x, t)8 :

∂ψ(x, t) 1 ∂ 2 ψ(x, t)
i =− + V (x, t)ψ(x, t). (3.27)
∂t 2m ∂x2
∂
The general form of a classical potential may be V (x, ∂x , t), and its concrete form is deter-
mined by the non-relativistic approximation of the quantum interactions involved, which are
described by the relativistic quantum field theory. Since the potential V (x, t) is real-valued,
the Hamiltonian H = P 2 /2m + V (x, t) is Hermitian, and as a result, the time translation
operator or evolution operator U (t) is also unitary.

3.4 Further discussions

We have derived the free Schrödinger equation in quantum mechanics based on spacetime
translation invariance and relativistic invariance. The derivation may not only make the
equation more logical and understandable, but also shed some new light on the physical
meaning of the wave function in the equation.
The free Schrödinger equation is usually “derived” in textbooks by analogy and cor-
respondence with classical physics. There are at least two mysteries in such a heuristic
“derivation”. First, even if the behavior of microscopic particles is like wave and thus a
wave function is needed to describe them, it is unclear why the wave function must assume
a complex form. Indeed, when Schrödinger originally invented his equation, he was very
puzzled by the inevitable appearance of the imaginary unit “i ” in the equation. Next, one
doesn’t know why there are the de Broglie relations for momentum and energy and why the
non-relativistic energy-momentum relation must be E = p2 /2m. Usually one can only resort
to experience and classical physics to answer these questions. This is unsatisfactory in logic
as quantum mechanics is a more fundamental theory, of which classical mechanics is only an
approximation.
As we have demonstrated above, the key to unveil these mysteries is to analyze the origin
of momentum and energy. According to the modern understanding, spacetime translation
gives the definitions of momentum and energy. The momentum operator P is defined as the
generator of space translation, and it is Hermitian and its eigenvalues are real. Moreover, the
∂
form of momentum operator can be uniquely determined by its definition. It is P = −i ∂x ,
ipx
and its eigenstate is e , where p is a real eigenvalue. Similarly, the energy operator H is
defined as the generator of time translation. But its form depends on the concrete situation.
Fortunately, for an isolated system the form of the energy operator, which determines the
evolution of the state of the system, can be fixed by the requirements of spacetime translation
invariance and relativistic invariance (when assuming the evolution is linear). Concretely
8
In order to derive the complete Schrödinger equation in a fundamental way, we need a fundamental
theory of interactions such as quantum field theory.

41
 3.4. FURTHER DISCUSSIONS

speaking, time translational invariance requires that dH/dt = 0, and the solution of the
evolution equation i ∂ψ(x,t)
∂t = Hψ(x, t) must assume the form ψ(x, t) = ϕE (x)e−iEt . Space
translational invariance requires [P, H] = 0, and this further determines that ϕE (x) is the
eigenstate of P , namely ϕE (x) = eipx . Thus spacetime translation invariance entails that the
state of an isolated system with definite momentum and energy assumes the plane wave form
ei(px−Et) . Furthermore, the relation between p and E or the energy-momentum relation can
be determined by the relativistic invariance of the momentum eigenstate ei(px−Et) , and its
non-relativistic approximation is just E = p2 /2m. Then we can obtain the form of the energy
operator for an isolated system, H = P 2 /2m, and the free Schrödinger equation, Eq.(3.26).
To sum up, this analysis may answer why the wave function must assume a complex form
in general and why there are the de Broglie relations and why the non-relativistic energy-
momentum relation is what it is.
So far so good. But how does the wave function ψ(x, t) in the thus-derived free Schrödinger
equation relate to the actual physical state of the system? Without answering this question
the above analysis seems vacuous in physics. This leads us to the problem of interpreting
the wave function. According to the standard probability interpretation, the wave function
in quantum mechanics is a probability amplitude, and its modulus square gives the proba-
bility density of finding a particle in certain locations. Notwithstanding the success of the
standard interpretation, our derivation of the free Schrödinger equation seems to suggest
that the wave function ψ(x, t) is a description of the objective physical state of a quantum
system, rather than the probability amplitude relating only to measurement outcomes. In
our derivation we never refer to the measurement of the isolated system at all. Moreover, the
derivation seems to further suggest that the wave function ψ(x, t) is a complete description
of the physical state of the system. As we have argued in the last chapter, ψ(x, t) can be re-
garded as an objective description of the state of random discontinuous motion of a particle,
and |ψ(x, t)|2 dx gives the objective probability of the particle being in an infinitesimal space
interval dx near position x at instant t. This objective interpretation of the wave function
is quite consistent with the above derivation of the free Schrödinger equation.
In addition, the derivation may provide another argument for the non-existence of con-
tinuous motion of particles from the aspect of the laws of motion. The continuous motion
of a particle can be regarded as a very special form of discontinuous motion, for which the
position density of the particle is ρ(x, t) = δ 2 (x − x(t)), where x(t) is the continuous trajec-
tory of the particle. However, such states are not solutions of the equation of free motion,
namely the free Schrödinger equation, though they satisfy the continuity equation. Accord-
ing to the free Schrödinger equation, an initial local state like δ(x − x0 ) cannot sustain its
locality during the evolution, and it will immediately spread throughout the whole space.
Thus the equation of free motion, which is derived based on the fundamental requirements of
spacetime translation invariance and relativistic invariance, does not describe the continuous
motion of particles. If the equation of free motion still describes the motion of particles as
we have argued in the last chapter, then the motion of particles cannot be continuous but
must be essentially discontinuous. Note that our derivation of the free Schrödinger equation
does not depend on the picture of discontinuous motion, and thus this argument for the
non-existence of continuous motion of particles is not a vicious circle.
As stressed earlier, our derivation of the free Schrödinger equation relies on the presuppo-
sition that the Hamiltonian H is independent of the evolved state, i.e., that the evolution is
linear. It can be reasonably assumed that the linear evolution and nonlinear evolution both
exist, and moreover, they satisfy spacetime translation invariance respectively because their
effects cannot counteract each other in general. Then our derivation only shows that the
linear part of free evolution, if satisfying spacetime translation invariance and relativistic in-
variance, must assume the same form as the free Schrödinger equation in the non-relativistic

42
 3.5. ON THE CONSERVATION OF ENERGY AND MOMENTUM

domain. Obviously, our derivation cannot exclude the existence of nonlinear quantum evolu-
tion. Moreover, since a general nonlinear evolution can readily satisfy spacetime translation
invariance, the invariance requirement can no longer determine the concrete form of possible
nonlinear evolution.

3.5 On the conservation of energy and momentum

The conservation of energy and momentum is one of the most important principles in
modern physics. In this section, we will analyze the basis and physical meaning of this
principle, especially its relationship with the linearity of quantum dynamics.
As we have noted in the above derivation of the free Schrödinger equation, the origin
of momentum and energy is closely related to spacetime translation; the momentum oper-
ator P and energy operator H are defined as the generators of space translation and time
translation, respectively. Moreover, it is well known that the conservation of energy and
momentum results from spacetime translation invariance. The usual derivation is as follows.
The evolution law for an isolated system satisfies spacetime translation invariance due to
the homogeneity of space and time. Time translational invariance requires that H has no
time dependence, namely dH/dt = 0, and space translational invariance requires that the
generators of space translation and time translation are commutative, namely [P, H] = 0.
Then by Ehrenfest’s theorem for an arbitrary observable A

dhAi ∂A
=h i − ih[A, H]i, (3.28)
dt ∂t
where hAi = ψ ∗ (x, t)Aψ(x, t)dx is defined as the expectation value of A, we have
R

dhHi
= 0, (3.29)
dt
and

dhP i
= 0. (3.30)
dt
This means that the expectation values of energy and momentum are conserved for the
evolution of an isolated system. Moreover, for arbitrary functions f (H) and f (P ), we also
have

dhf (H)i
= 0, (3.31)
dt
and

dhf (P )i
= 0. (3.32)
dt
This is equivalent to the constancy of the expectation values of the generating functions or
spacetime translation operators U (a) ≡ e−iaH and T (a) ≡ e−iaP

dhU (a)i
= 0, (3.33)
dt
and

dhT (a)i
= 0. (3.34)
dt

43
 3.5. ON THE CONSERVATION OF ENERGY AND MOMENTUM

By these two equations it follows that the probability distributions of energy eigenvalues
and momentum eigenvalues are constant in time. This statement is usually defined as the
conservation of energy and momentum in quantum mechanics.
Now let’s analyze the implications of this derivation for the meaning of the conservation
of energy and momentum. First of all, we point out that the linearity of evolution is an
indispensable presupposition in the derivation. As we have stressed in the derivation of the
free Schrödinger equation, spacetime translation invariance does not lead to dH/dt = 0 and
[P, H] = 0 without assuming the linearity of evolution. Therefore, the common wisdom
that spacetime translation invariance implies laws of conservation only holds true for a
linear evolution. For a general nonlinear evolution H(ψ), energy and momentum will not be
conserved by Ehrenfest’s theorem9 :

dhH(ψ)i ∂H(ψ) ∂H(ψ)

=h i − ih[H(ψ), H(ψ)]i = h i=
6 0, (3.35)
dt ∂t ∂t
and

dhP i ∂P ∂H(ψ)
=h i − ih[P, H(ψ)]i = −ih[P, H(ψ)]i = −h i=6 0. (3.36)
dt ∂t ∂x
We can see the violation of the conservation of energy and momentum more clearly by an-
alyzing the nonlinear evolution of momentum eigenstates ei(px−Et) and their superpositions.
If a nonlinear evolution can conserve energy and momentum for momentum eigenstates, then
the momentum eigenstates must be the solutions of the nonlinear evolution equation; other-
wise the evolution will change the definite momentum eigenvalues or energy eigenvalues or
both and lead to the violation of the conservation of energy and momentum. Some nonlin-
ear evolutions can satisfy this requirement. For example, when H(ψ) = P 2 /2m + α|ψ|2 , the
solutions still include the momentum eigenstates ei(px−Et) , where E = p2 /2m + α, and thus
energy and momentum are conserved for such nonlinear evolutions of momentum eigenstates.
However, even if a nonlinear evolution can conserve energy and momentum for momentum
eigenstates, it cannot conserve energy and momentum for the superpositions of momentum
eigenstates. The reason is obvious. Only for a linear evolution the momentum eigenstates
and their superpositions can both be the solutions of the evolution equation. For any nonlin-
ear evolution H(ψ), if the momentum eigenstates are already its solutions, then their linear
superposition cannot be its solutions. This means that the coefficients of the momentum
eigenstates in the superposition will change with time during the evolution. The change of
the amplitudes of the coefficients directly leads to the change of the probability distribution
of momentum eigenvalues and energy eigenvalues, while the change of the phases of the
coefficients leads to the change of the momentum eigenvalues or energy eigenvalues, which
also leads to the change of the probability distribution of momentum eigenvalues or energy
eigenvalues. In fact, a nonlinear evolution may not only change the probability distributions
of energy and momentum eigenvalues, but also change the energy-momentum relation in gen-
eral cases (e.g. in the above example)10 . These results are understandable when considering
the fact that the nonlinear evolution of a spatial wave function will generally introduce a
time-dependent interaction between its different momentum eigenstates, which is equivalent
to adding a time-dependent external potential for its free evolution in some sense. Therefore,
it is not beyond expectation that a nonlinear evolution violates the conservation of energy
and momentum in general.
9
In order to ensure that the nonlinear evolution is unitary and thus the total probability is conserved in
time, the Hamiltonian H(ψ) must be Hermitian. Besides, this property is also required to ensure that the
energy eigenvalues (which satisfy the equation H(ψ)ψ(x) = Eψ(x)) are real. When the Hamiltonian H(ψ) is
Hermitian, the Ehrenfest theorem holds true.
10
This will violate the relativistic invariance of momentum eigenstates.

44
 3.5. ON THE CONSERVATION OF ENERGY AND MOMENTUM

Two points needs to be stressed here. First, energy and momentum are still defined as
usual for a nonlinear evolution in the above discussions. One may object that they should
be re-defined for a nonlinear evolution. However, this may be not the case. The reason is
as follows. Momentum is defined as the generator of space translation, and this definition
uniquely determines that its eigenstates are eipx . Similarly, energy is defined as the gener-
ator of time translation, and this definition uniquely determines that its eigenstates satisfy
H(ψ)ψ(x) = Eψ(x). Since these definitions are independent of whether the evolution of the
state is linear or nonlinear, they should have a fundamental status in a theory formulated in
space and time such as quantum mechanics. The second point is that the above argument
implicitly assumes that the nonlinear evolution H(ψ) is universal, i.e., that it applies to
all possible states. If the nonlinear evolution only applies to some special states, then the
evolution may still conserve energy and momentum. For example, suppose the nonlinear
evolution H(ψ) = P 2 /2m + α|ψ|2 applies only to the momentum eigenstates ei(px−Et) and
the linear evolution H(ψ) = P 2 /2m applies to the superpositions of momentum eigenstates,
then energy and momentum are still conserved during the evolution. On the other hand,
it has been argued that the universal nonlinear quantum dynamics has a serious drawback,
namely that the description of composite systems depends on a particular basis in the Hilbert
space (Czachor 1996). If a nonlinear quantum evolution only applies to certain privileged
bases due to some reason, then such nonlinear quantum dynamics may be logically consistent
and also conserve energy and momentum (Gao 2004).
The second implication of the above derivation of the conservation laws is that spacetime
translation invariance implies the conservation of energy and momentum for individual states,
not for an ensemble of identical states. As in the derivation of the free Schrödinger equation,
we only refer to an isolated system and never refer to an ensemble of identical systems
in the derivation of the conservation laws. Moreover, the transformations of spacetime
translation also apply to a single isolated system. Therefore, what the derivation tells us
is that spacetime translation invariance implies the conservation of energy and momentum
for the linear evolution of the states of an isolated system. The conservation of energy and
momentum for a single system means that the objective probability distributions of energy
eigenvalues and momentum eigenvalues are constant during the evolution of the state of the
system. As we have argued in the last chapter, the objective probability can be understood
according to the interpretation of the wave function in terms of random discontinuous motion.
Similarly, the above analysis of nonlinear evolutions also shows that a universal nonlinear
evolution violates the conservation of energy and momentum for individual systems.
This implication raises a further issue. It is well known that the conservation of energy
and momentum in quantum mechanics refers to an ensemble of identical systems, not to
individual systems, and its precise statement is that the probability distributions of the
measurement results of energy and momentum for an ensemble of identical isolated systems
are constant during the evolution of the systems in the ensemble. But as we have argued
above, the derivation of the conservation laws based on spacetime translation invariance is
for individual isolated systems, not for an ensemble of these systems. The derivation never
refers to the measurements of these systems either. Therefore, there is still a gap (which
may be very large) between the derivation and the conservation laws in quantum mechanics.
Undoubtedly we must analyze the measurement process in order to fill the gap. We will
postpone a detailed analysis of the measurement problem to the next section. Here we only
want to answer a more general question. If the conservation laws in quantum mechanics
are indeed valid as widely thought, then what are their implications for the evolution of
individual states?
First of all, the evolution of the state of an isolated system cannot contain a universal
deterministic nonlinear evolution, which applies to all possible states; otherwise the evolution

45
 3.5. ON THE CONSERVATION OF ENERGY AND MOMENTUM

will violate the conservation of energy and momentum not only at the individual level but
also at the ensemble level. Next, the evolution may contain linear evolutions as well as
special deterministic nonlinear evolutions that apply only to certain privileged states. They
can both conserve energy and momentum for individual states11 . Lastly, the evolution may
also contain a (universal) stochastic nonlinear evolution, which applies to all possible states.
Although the evolution cannot conserve energy and momentum for individual states, it may
conserve energy and momentum for an ensemble of identical states. As we will see in the
next chapter, the dynamical collapse of the wave function may be such a stochastic nonlinear
evolution.
To summarize, we have analyzed the relationships between the conservation of energy
and momentum, spacetime translation invariance and the linearity of quantum dynamics. It
has been often claimed that the conservation of energy and momentum is a conservation law
resulting from the requirement of spacetime translation invariance. However, this common-
sense view is not wholly right. Only when assuming the linearity of quantum dynamics,
can spacetime translation invariance lead to the conservation of energy and momentum.
Moreover, the connection between invariance of natural laws and conservation laws is for
individual states, not for an ensemble of identical states. Although a nonlinear evolution of
the wave function can readily satisfy spacetime translation invariance, the invariance can no
longer lead to the conservation of energy and momentum, let alone determining the form of
the nonlinear evolution. Rather, a universal nonlinear evolution that applies to all possible
states will inevitably violate the conservation of energy and momentum.
Since the conservation of energy and momentum is required by spacetime translation in-
variance only for the linear evolution of the wave function of an isolated system, the principle
cannot exclude the existence of a possible nonlinear evolution that may violate it. In other
words, spacetime translation invariance is no longer a reason to require that the evolution
of the wave function of an isolated system must conserve energy and momentum. On the
other hand, the conservation of energy and momentum may still hold true for an ensemble
of identical isolated systems as claimed by the standard quantum mechanics. Therefore, a
(universal) stochastic nonlinear evolution of the wave function may exist. Although such
evolutions cannot conserve energy and momentum for individual states, it may conserve
energy and momentum at the ensemble level. However, unlike the linear evolution, which
is natural in the sense that its form can be uniquely determined by the invariance require-
ments, the stochastic nonlinear evolution must have a physical origin, and its form can only
be determined by the underlying mechanism. In the next chapter, we will investigate the
possible stochastic nonlinear evolution of the wave function.

11
For more discussions about the arguments for linear quantum dynamics see Holman (2006) and references
therein.

46
 4
Was the wavefunction of the world waiting to jump for thousands of millions
of years until a single-celled living creature appeared? Or did it have to wait a
little longer, for some better qualified system ... with a Ph.D.? ... Do we not
have jumping then all the time?
— John Bell, 1990

A Suggested Solution to the Measurement Problem

In standard quantum mechanics, it is postulated that when a wave function is measured by a

macroscopic device, it no longer follows the linear Schrödinger equation, but instantaneously
collapses to one of the wave functions that correspond to definite measurement results.
However, this collapse postulate is ad hoc, and the theory does not tell us why and how
a definite measurement result appears (Bell 1990). There are in general two ways to solve
the measurement problem. The first one is to integrate the collapse evolution with the
normal Schrödinger evolution into a unified dynamics, e.g. in the dynamical collapse theories
(Ghirardi 2008). The second way is to reject the collapse postulate and assume that the
Schrödinger equation completely describes the evolution of the wave function. There are two
main alternative theories for avoiding collapse. The first one is the de Broglie-Bohm theory
(de Broglie 1928; Bohm 1952), which takes the wave function as an incomplete description
and adds some hidden variables to explain the emergence of definite measurement results.
The second one is the many-worlds interpretation (Everett 1957; DeWitt and Graham 1973),
which assumes the existence of many equally real worlds corresponding to all possible results
of quantum experiments and still regards the unitarily evolving wave function as a complete
description of the total worlds.
It has been in hot debate which solution to the measurement problem is the right one
or in the right direction. One of the main reasons, as we think, is that the physical mean-
ing of the wave function is not well understood. The failure of making sense of the wave
function is partly because the problem is only investigated in the context of conventional
impulse measurements. As we have shown in Chapter 2, the problem of interpreting the
wave function can be solved with the help of protective measurement, independent of how
to solve the measurement problem. There are three steps to reach this conclusion. First,
protective measurement shows that the mass and charge of a quantum system are distributed
throughout space, and the mass and charge density in each position is proportional to the
modulus square of its wave function there. Next, the superposition principle of quantum
mechanics requires that the mass and charge distributions are not real but effective, that
is, they are formed by the ergodic motion of a localized particle with the total mass and
charge of the system. Lastly, in order to form the mass and charge distributions predicted
by quantum mechanics, the ergodic motion of the particle must be discontinuous, and the
probability density of the particle appearing in every position must be equal to the mod-
ulus square of its wave function there. This result then implies that the wave function in
quantum mechanics describes the state of random discontinuous motion of particles, and at
a deeper level, it may represent the property of the particles that determines their random
discontinuous motion. Since the principle of protective measurement is independent of the
controversial process of wavefunction collapse and only based on the established parts of
quantum mechanics, namely the linear Schrödinger evolution of the wave function (for mi-
croscopic systems) and the Born rule, its implications, especially the resulting interpretation
of the wave function, can be used to examine the existing solutions to the measurement

47
 4.1. THE REALITY OF WAVEFUNCTION COLLAPSE

problem before experiments give the last verdict (cf. Marshall et al 2003)1 .
In this chapter, we will first investigate the existing solutions to the measurement prob-
lem and then propose a new solution based on the implications of protective measurement
and the meaning of the wave function. It is first shown that the two quantum theories
without wavefunction collapse, namely the de Broglie-Bohm theory and the many-worlds
interpretation, are inconsistent with protective measurement and the picture of random dis-
continuous motion of particles. This result strongly suggests that wavefunction collapse is a
real physical process. Next, it is argued that the random discontinuous motion of particles
may provide an appropriate random source to collapse the wave function. Moreover, the
wavefunction collapse is a discrete process, and the collapse states are energy eigenstates
when the principle of conservation of energy is satisfied. Based on these analyses, we fur-
ther propose a discrete model of energy-conserved wavefunction collapse. It is shown that
the model is consistent with existing experiments and our macroscopic experience. Lastly,
we also provide a critical analysis of other dynamical collapse models, including Penrose’s
gravity-induced collapse model and the CSL (Continuous Spontaneous Localization) model.

4.1 The reality of wavefunction collapse

At first sight, the main solutions to the measurement problem, i.e., the de Broglie-Bohm
theory, the many-worlds interpretation and dynamical collapse theories, seem apparently
inconsistent with the meaning of the wave function as implied by protective measurement.
Most interpretations of these theories take the wave function as a real physical entity on
configuration space or assume the wave function has a field-like spatiotemporal manifestation
in the real three-dimensional space (see, e.g. Ghirardi 1997, 2008; Wallace and Timpson
2009)2 . But the wave function is a description of the state of random discontinuous motion
of particles in real space, and at a deeper level, it represents the property of the particles
that determines their random discontinuous motion. However, some of these theories may
be not influenced by the interpretation of the wave function in a significant way. Therefore,
they may be still consistent with the meaning of the wave function after certain revision.

4.1.1 Against the de Broglie-Bohm theory

Let’s first investigate the de Broglie-Bohm theory (de Broglie 1928; Bohm 1952). According
to the theory, a complete realistic description of a quantum system is provided by the con-
figuration defined by the positions of its particles together with its wave function. The wave
function follows the linear Schrödinger equation and never collapses. The particles, called
Bohmian particles, are guided by the wave function via the guiding equation to undergo
deterministic continuous motion. The result of a measurement is indicated by the positions
of the Bohmian particles representing the pointer of the measuring device, and thus it is
1
Note that the possible existence of very slow collapse of the wave function for microscopic systems does
not influence the principle of protective measurement and its implications.
2
It has been argued that the wave function living on configuration space can hardly be considered as a
real physical entity due to its multi-dimensionality (see, e.g. Monton 2002, 2006 and references therein).
However, it seems that this common objection is not conclusive, and one can still insist on the reality of the
wave function living on configuration space by resorting to some metaphysical arguments. For example, a
general strategy is to show how a many-dimensional world can appear three-dimensional to its inhabitants,
and then argue on that basis that a wavefunction ontology is adequate to explain our experience (Albert
1996; Lewis 2004). As we have argued in Chapter 2, the existence of the effective mass and charge density
of a quantum system, which is measurable by protective measurement, poses a more serious objection to the
wavefunction ontology; even for a single quantum system the wave function cannot be taken as a field-like
entity in three-dimensional space either. Moreover, the reason is not metaphysical but physical, i.e., the
field-like interpretation contradicts both quantum mechanics and experimental observations.

48
 4.1. THE REALITY OF WAVEFUNCTION COLLAPSE

always definite. It can be shown that the de Broglie-Bohm theory gives the same predictions
of measurement results as standard quantum mechanics by means of a quantum equilibrium
hypothesis (so long as the latter gives unambiguous predictions). Concretely speaking, the
quantum equilibrium hypothesis provides the initial conditions for the guidance equation
which make the de Broglie-Bohm theory obey Born’s rule in terms of position distributions.
Moreover, since all measurements can be finally expressed in terms of position, e.g. pointer
positions, this amounts to full accordance with all predictions of quantum mechanics3 . In
this way, it seems that the de Broglie-Bohm theory can succeed in avoiding the collapse of
the wave function.
However, although the de Broglie-Bohm theory is mathematically equivalent to quantum
mechanics, there is no clear consensus with regard to its physical interpretation. The physical
contents of the theory contain three parts: the Bohmian particles, the wave function, and
the interaction between them. We first analyze the Bohmian particles and their physical
properties. It is fair to say that what physical properties a Bohmian particle has is still
an unsettled issue, and different proponents of the theory may have different opinions. For
example, it has been often claimed that a Bohmian particle has mass, as the guiding equation
for each Bohmian particle of a many-body system obviously contains the mass of each sub-
system (Goldstein 2009). Yet it seems unclear whether the mass is inertial mass or (passive
or active) gravitational mass or both or neither. On the other hand, it has been argued that
the mass of a quantum system should be possessed by its wave function, not by its Bohmian
particles (Brown, Dewdney and Horton 1995). It was even claimed (without proof) that a
Bohmian particle has no properties other than its position (Hanson and Thoma 2011). In
the last analysis, in order to know exactly what physical properties a Bohmian particle has,
we need to analyze the guiding equation that defines the laws of motion for them.
In the minimum formulation of the theory, which is usually called Bohmian mechanics
(Goldstein 2009)4 , the guiding equation for the Bohmian particle of a one-particle system
with mass m and charge e in the presence of an external electromagnetic field is5

dx ∇ψt
m = ~=[ ] − eA(x, t), (4.1)
dt ψt
where x is the position of the Bohmian particle, ψt is the wave function of the system
that obeys the Schrödinger equation, A(x, t) is the magnetic vector potential in position x.
According to this equation, the motion of a Bohmian particle is not only guided by the wave
function, but also influenced by the external vector potential A(x, t). The existence of the
term eA(x, t) in the guiding equation indicates that the Bohmian particle has charge e, the
charge of the system, and the charge is localized in its position6 . Besides, the appearance
of the mass of the system in the equation may indicate that the Bohmian particle also has
the (inertial) mass of the system. Therefore, according to Bohmian mechanics, the Bohmian
particle of a one-particle system such as an electron has the mass and charge of the system.
For example, in the ground state of a hydrogen atom, the Bohmian particle of the electron
in the atom has the mass and charge of the electron, and it is at rest in a random position
near the nucleus.
3
Certainly, as Albert (1992) noted, no theory can have exactly the same empirical content as quantum
mechanics does, as the latter (in the absence of any satisfactory account of wavefunction collapse) does not
have any exact empirical content.
4
For a critical analysis of this minimal formal interpretation see Belousek (2003).
5
Note that this guiding equation applies only for spin 0 particles, and for spin 1/2 particles there is also
a spin-dependent term (Holland and Philippidis 2003).
6
That a Bohmian particle has no properties other than its position is possible only when the mass and
charge terms disappear in the guiding equation, but the resulting theory will contradict quantum mechanics
and experiments.

49
 4.1. THE REALITY OF WAVEFUNCTION COLLAPSE

That the Bohmian particle of a one-particle system has the mass and charge of the system
can be seen more clearly from the quantum potential formulation of the de Broglie-Bohm
theory. By differentiating both sides of Eq. (4.1) relative to time and including an external
gravitational potential VG , we obtain

dẋ ∂A
m = −∇Q − m∇VG − e[∇A0 + − ẋ × (∇ × A)], (4.2)
dt ∂t
2
d ∂ ~ ∇ |ψt |
2
where dt = ∂t + ẋ · ∇, A0 is the electric scalar potential, and Q = − 2m |ψt | is the so-called
quantum potential. The electromagnetic interaction term −e[∇A0 + ∂A ∂t − ẋ × (∇ × A)]
indicates that the Bohmian particle has charge e, and the gravitational interaction term
−m∇VG indicates that the Bohmian particle also has (passive) gravitational mass m.
It can be seen that although a Bohmian particle has mass and charge, the functions of
these properties are not as complete as usual. For example, in Bohmian mechanics, a charged
Bohmian particle responds not to the electric scalar potential, but only to the magnetic vector
potential, and it has no gravitational mass but only inertial mass. This apparent abnormal-
ity is in want of a reasonable physical explanation. In addition, in the quantum potential
formulation, although the Bohmian particles of a quantum system respond to external grav-
itational and electromagnetic potentials, they don’t have gravitational and electromagnetic
influences on other charged quantum systems, including their Bohmian particles. Moreover,
the Bohmian particles of a quantum system do not have gravitational and electromagnetic
interactions with each other. Therefore, the (gravitational) mass and charge of a Bohmian
particle are always passive, i.e., a Bohmian particle is only a receptor of gravitational and
electromagnetic interactions. This characteristic may lead to some problems. For one, the
nonreciprocal interactions will violate the conservation of energy and momentum (except
that the Bohmian particles have no momentum and energy). At the worst, it already sug-
gests that the hypothetical Bohmian particles are redundant entities in the theory (and
their role in solving the measurement problem is ad hoc), since they have no any influence
on other entities in the theory such as the wave function. Note that these problems do not
exist for the wave function; the evolution of the wave function of a charged quantum system
is influenced by both electric scalar potential and magnetic vector potential, as well as by
gravitational potential, and the wave functions of two charged quantum systems also have
gravitational and electromagnetic interactions with each other.
Another suggestion of the non-existence of Bohmian particles concerns the mass and
charge distributions of a one-particle system such as an electron. As we have shown above,
the guiding equation in the de Broglie-Bohm theory requires that the Bohmian particle of
a one-particle system has the mass and charge of the system, and the mass and charge are
localized in a position where the Bohmian particle is. On the other hand, as noted before,
protective measurement shows that the mass and charge of a one-particle system such as an
electron are not localized in one position but distributed throughout space, and the mass and
charge density in each position is proportional to the modulus square of its wave function
there. Therefore, the de Broglie-Bohm theory is inconsistent with the results of protective
measurement concerning the mass and charge distributions of a quantum system7 . This
7
This conclusion relies on the common-sense assumption that an electron indeed has only one basic charge
(and one mass of the electron). The only way to avoid the inconsistency is to assume that an electron has
two basic charges: one for its wave function and the other for its Bohmian particle. In this case, since what
protective measurement measures is the mass and charge distributions relating to the wave function, not the
masses and charges of the Bohmian particles, the above inconsistency can be avoided. However, this theory
seems too clumsy and unnatural to be true. Moreover, it will introduce more problems. For one, there is
a dilemma concerning the electromagnetic interaction between the wave function and the Bohmian particle
of an electron. If they do have usual electromagnetic interaction, then the theory will be inconsistent with
quantum mechanics and experiments. If they have no electromagnetic interaction, then this will add more

50
 4.1. THE REALITY OF WAVEFUNCTION COLLAPSE

poses a serious objection to the de Broglie-Bohm theory.

Now let’s turn to the wave function in the de Broglie-Bohm theory. Admittedly, the
interpretation of the wave function in the theory has been debated by its proponents. For
example, the wave function has been regarded as a field similar to electromagnetic field
(Bohm 1952), an active information field (Bohm and Hiley 1993), a field carrying energy
and momentum (Holland 1993), a causal agent more abstract than ordinary fields (Valentini
1997), a component of physical law (Dürr, Goldstein and Zanghı̀ 1997; Goldstein and Teufel
2001), and a dispositional property of Bohmian particles (Belot 2011) etc. Notwithstanding
the differences between these existing interpretations, they are inconsistent with the meaning
of the wave function as implied by the results of protective measurement. To say the least,
they fail to explain the existence of the mass and charge density for a charged quantum sys-
tem, which is measurable by protective measurement and proportional to the modulus square
of the wave function of the system. Our previous analysis shows that the mass and charge
density of a quantum system is effective and formed by the ergodic motion of a localized
particle with the total mass and charge of the system, which is discontinuous and random
in nature. Thus the wave function describes the state of random discontinuous motion of
particles, and at a deeper level, it represents the property of the particles that determines
their random discontinuous motion. Since the principle of protective measurement is based
on the linear Schrödinger evolution of the wave function and the Born rule, which also hold
true in the de Broglie-Bohm theory, its implications, especially the resulting interpretation
of the wave function, are still valid in the theory.
The realistic interpretation of the wave function poses another serious threat against
the Bohmian-particles explanation of the guiding equation imposed by the de Broglie-Bohm
theory. The guiding equation is only a mathematical transformation of the relation between
the density ρ and the flux density j for the wave function; the relation is j = ρv, while the
guiding equation is v = j/ρ. Since the wave function of a quantum system is not merely a
probability amplitude for the predictions of measurement results, but also a realistic descrip-
tion of the physical state of the system as implied by protective measurement8 , the guiding
equation already has a physical explanation relating only to the realistic wave function. Inas-
much as a fundamental mathematical equation in a physical theory has a unique physical
explanation, the additional explanation of the guiding equation relating to the hypothetical
Bohmian particles will be improper9 . In addition, the positions of the Bohmian particles as
added (hidden) variables seem redundant too10 . In some sense, there are already additional
variables besides the wave function for the random discontinuous motion of particles. They
are the definite position, momentum and energy of the particles at each instant. Though
these variables are not continuous and deterministic, their random motion may just lead to
the stochastic collapse of the wave function and further account for the emergence of random
measurement results. We will discuss this possibility in detail later on.
problems. For instance, the manifestation of the charge of a Bohmian particle will be much stranger; it is
not only passive but also selective. One needs to explain why the charged Bohmian particle of an electron
responds not to the magnetic vector potential generated by the wave function of this electron, but to the
magnetic vector potential generated by the wave function of another electron. As we will see later, a more
serious objection concerns the guiding responsibility of the wave function.
8
This is also admitted by most interpretations of the de Broglie-Bohm theory.
9
This conclusion may not hold true if the guiding equation is not exactly the same as the above, e.g.
the guiding equation contains an additional stochastic damping term. Although such revised theories give
predictions different from quantum mechanics, they may be consistent with existing experiments.
10
The reality of the trajectories of the Bohmian particles has been questioned based on the analysis of
weak measurement and protective measurement (Englert, Scully, Süssmann and Walther 1992; Aharonov
and Vaidman 1996; Aharonov, Englert and Scully 1999; Aharonov, Erez and Scully 2004). However, these
objections may be answered by noticing that what protective measurement measures is the wave function,
not the Bohmian particles (see also Drezet 2006).

51
 4.1. THE REALITY OF WAVEFUNCTION COLLAPSE

Lastly, we analyze the hypothetical interaction between the Bohmian particles and the
wave function in the de Broglie-Bohm theory. It can be seen that the guiding responsibility
of the wave function assumed by the theory is inconsistent with the meaning of the wave
function. As noted above, the wave function represents the property of particles that de-
termines their random discontinuous motion. Accordingly, the wave function indeed guides
the motion of particles in some sense. However, the wave function guides the motion of the
particles not in a deterministic and continuous way as assumed by the de Broglie-Bohm the-
ory, but in a probabilistic and discontinuous way; the modulus square of the wave function
determines the probability density of the particles appearing in certain positions in space.
Moreover, the motion of these particles is ergodic. By contrast, the motion of the Bohmian
particles is not ergodic, and the time averages of the Bohmian particles positions typically
differ remarkably from the ensemble averages (Aharonov, Erez and Scully 2004).
Although one may assume that a quantum system contains additional Bohmian particles
besides its non-Bohmian particles that undergo random discontinuous motion, the motion of
the Bohmian particles cannot be guided by the wave function of the system. For the wave
function of the system represents the property of the non-Bohmian particles of the system,
and its efficiency is to guide the motion of these particles in a probabilistic way. In particular,
the wave function is neither a field-like entity distributing throughout space nor a property of
the Bohmian particles that may guide their motion, and at every instant there are only non-
Bohmian particles being in positions that are usually far from the positions of the Bohmian
particles. Note also that the non-Bohmian particles cannot have known interactions such as
gravitational and electromagnetic interactions with the Bohmian particles either; otherwise
the theory will contradict quantum mechanics and experiments. Without being guided by
the wave function in a proper way, the motion of the Bohmian particles will be unable to
generate the right measurement results in conventional impulse measurements.
In conclusion, we have shown that the de Broglie-Bohm theory is inconsistent with the
results of protective measurement and the meaning of the wave function implied by them
when considering its physical contents.

4.1.2 Against the many-worlds interpretation

Now let’s turn to the second approach to avoid wavefunction collapse, the many-worlds
interpretation. Although this theory is widely acknowledged as one of the main alternatives
to quantum mechanics, its many fundamental issues, e.g. the preferred basis problem and the
interpretation of probability, have not been completely solved yet (see Barrett 1999; Saunders
et al 2010 and references therein). In this subsection, we will mainly analyze whether the
existence of many worlds is consistent with the results of protective measurement and the
picture of random discontinuous motion of particles.
According to the many-worlds interpretation, each branch of the wave function of a
measuring device in which there is a definite measurement result corresponds to each world
among the many worlds (Barrett 2011). This means that in one world there is only one
branch of the superposed wave function in which there is a definite measurement result,
and the other branches do not exist in this world. This further means that in one world
the branches of the superposed wave function corresponding to the other worlds cannot be
measured11 . For if all branches can be observed in one world, then they will all exist in this
world. As a result, in every world the whole superposed wave function of the measuring
device cannot be measured.
It is unsurprising that the existence of such many worlds may be consistent with the
results of conventional impulse measurements, as the many-worlds interpretation is just in-
11
In other words, the many worlds are separate and mutually unobservable.

52
 4.1. THE REALITY OF WAVEFUNCTION COLLAPSE

vented to explain the emergence of these results, e.g. the definite measurement result in
each world always denotes the result of a conventional impulse measurement. However, this
does not guarantee consistency for all types of measurements. Indeed, it can be seen that
the existence of the many worlds defined above is inconsistent with the results of protective
measurements. The reason is that the whole superposed wave function of a quantum system
including a measuring device can be measured by a protective measurement12 . The result
of the protective measurement indicates that all branches of the superposed wave function
of the measuring device exist in the same world where the protective measurement is made.
Therefore, according to protective measurement, the branches of the superposed wave func-
tion of a measuring device, in each of which there is a definite measurement result, do not
correspond to many worlds, in each of which there is only one such branch; rather, the
whole superposed wave function of the device, if it exists, only exists in one world, namely
our world. In this way, protective measurement provides a strong argument against the
many-worlds interpretation13 .
Four points are worth stressing. First of all, the above argument does not depend on
how the many worlds are precisely defined in the many-worlds interpretation. The key point
is that all branches of the superposed wave function of a measuring device can be detected
by protective measurement in our world, and thus they all exist in one world. Therefore,
according to protective measurement, it is impossible that there are many worlds, in each
of which there is only one branch of the superposed wave function of a measuring device.
Next, the above argument is not influenced by environment-induced decoherence. On the
one hand, even if the superposition state of a measuring device is entangled with the states
of other systems, the entangled state of the whole system can also be measured by protective
measurement in principle (Anandan 1993). The method is by adding appropriate protection
procedure to the whole system so that its entangled state is a nondegenerate eigenstate of
the total Hamiltonian of the system together with the added potential. Then the entangled
state can be protectively measured. On the other hand, environment-induced decoherence
is not an essential element of the many-worlds interpretation. Even when a measuring
device is isolated from environment (and the measured particle is absorbed by the device),
the interpretation also requires that each branch of the wave function of the measuring
device in which there is a definite measurement result corresponds to each world among the
many worlds; otherwise the many-worlds interpretation will not give the same predictions of
measurement results as standard quantum mechanics (so long as the latter gives unambiguous
predictions).
Thirdly, the above argument does not require protective measurement to be able to
distinguish the superposed wave function of a measuring device (in each branch of which
there is a definite measurement result) from one of its branches, or whether the superposed
wave function collapses or not during a conventional impulse measurement. Since the deter-
mination demands the distinguishability of two non-orthogonal states, which is prohibited
by quantum mechanics, no measurements consistent with the theory including protective
measurement can do this. What protective measurement tells us is that such a superposed
wave function, which existence is assumed by the many-worlds interpretation, does not cor-
respond to the many worlds defined by the many-worlds interpretation. In other words,
protective measurement reveals inconsistency of the many-worlds interpretation. Lastly, we
12
Note that protective measurement in general requires that the measured wave function is known before-
hand so that an appropriate protective interaction can be added. But this requirement does not influence our
argument, as the superposed wave function of a measuring device can be prepared in a known form before
the protective measurement.
13
This objection does not apply to the de Broglie-Bohm theory, according to which the wave function of a
measuring device does not collapse either, but it exists only in one world.

53
 4.1. THE REALITY OF WAVEFUNCTION COLLAPSE

stress again that the principle of protective measurement is irrelevant to the controversial
process of wavefunction collapse and only depends on the linear Schrödinger evolution and
the Born rule. As a result, protective measurement can be used to examine the internal
consistency of the no-collapse solutions to the measurement problem, e.g. the many-worlds
interpretation, before experiments give the last verdict.
In the following, we will further show that the existence of many worlds is not consistent
with the picture of random discontinuous motion of particles either. In order to examine the
many-worlds interpretation, it is necessary to know exactly what a quantum superposition is.
No matter how to define the many worlds, they correspond to some branches of a quantum
superposition after all (e.g. the branches where measuring devices obtain definite results,
and in particular, observers have definite conscious experience). According to the picture of
random discontinuous motion of particles, a quantum superposition exists in a form of time
division. For a superposition of two positions A and B of a quantum system (e.g. the pointer
of a measuring device), the system randomly and discontinuously jumps between these two
positions. At some random and discontinuous instants the system is in position A, and at
other instants it is in position B. In this picture of quantum superposition, it is obvious that
there is only one system all along, which randomly and discontinuously moves throughout all
branches of the superposition, no matter the system is a microscopic particle or a measuring
device or an observer. In other words, there is only one world whose instantaneous state is
constantly changing in a random and discontinuous way.
This conclusion is also supported by a comparison between discontinuous motion and
continuous motion. For a quantum particle undergoing discontinuous motion, the position
of the particle changes discontinuously. For a classical particle, its position changes continu-
ously. There is no essential difference between these two kinds of changes. For both cases the
position of the particle is always definite at each instant, and the positions of the particle at
different instants may be different. Moreover, the discontinuous change, like the continuous
change, does not create the many worlds, because, among other reasons, the change happens
all the while but the creating process only happens once. Therefore, if there is only one
world in classical mechanics, then there is also one world in quantum mechanics according
to the picture of random discontinuous motion of particles, no matter how the many worlds
are precisely defined.
We have argued that there are no many (physical) worlds as claimed by the many-worlds
interpretation, and in particular, even if the physical state or brain state of an observer is in
a quantum superposition, there is still one physical observer. However, the argument does
not exclude the variants of the many-worlds interpretation that assume a distinct dynamics
for the evolution of an observer’s mental state, e.g. the many-minds theory (Albert and
Loewer 1988)14 . For example, although the superposed brain state of an observer does not
correspond to many physical observers, each of which has a definite measurement record, it
may correspond to many minds of a unique observer, each of which has the experience of
a definite measurement record, as assumed by the many-minds theory. Since what we can
immediately access is not the position of the pointer of a measuring device, but our immediate
conscious experience, it is indeed necessary to analyze the conscious experience of an observer
during a conventional impulse measurement. In the final analysis, the measurement problem
is the problem of explaining the apparent incompatibility of our determinate experience and
14
According to these theories, the physical state always evolves in a deterministic way and may be super-
posed and indefinite, while the mental state is always definite but evolves randomly. In some sense, these
theories can be regarded as hidden-variable theories like the de Broglie-Bohm theory. The latter assumes
the definite positions of Bohmian particles provide observers with definite measurement records, while the
former assumes the definite mental states of the observers, though which are non-physical parameters, directly
provide observers with definite measurement records.

54
 4.1. THE REALITY OF WAVEFUNCTION COLLAPSE

the existence of indeterminate quantum superpositions.

According to our existing experience, when an observer makes an impulse measurement
(by or not by a measuring device) on a quantum superposition of two states of a measured
system, each of which can lead to a determinate conscious perception of the observer, his
conscious perception is randomly one of the determinate perceptions corresponding to the two
states (with probability being equal to the objective probability of the respective state in the
superposition). The question is whether an observer in a quantum superposition of definite
brain states, which may be called a quantum observer, can have a determinate conscious
perception corresponding to one of these brain states in a probabilistic way consistent with
the above experience. We will argue below that the answer is negative.
According to the picture of random discontinuous motion of particles, for a quantum
observer there is still one physical observer whose brain state is definite at every instant
but undergoes random discontinuous change. There are three possibilities for the conscious
perception of such a quantum observer. The first possibility is that the conscious perception
of a quantum observer is irrelevant to his superposed brain state. Obviously this possibility
is inconsistent with the above experience. The second possibility is that the conscious per-
ception of a quantum observer depends on his superposed brain state, and the observer can
instantaneously be conscious of his brain state. In this case, the conscious perception of a
quantum observer, parallel to his brain state, will also undergo random and discontinuous
change between the determinate conscious perceptions corresponding to the brain states in
the superposition15 . This is not consistent with the above experience either. The third pos-
sibility is that the conscious perception of a quantum observer depends on his superposed
brain state, and the observer can be conscious of his brain state only during a finite time in-
terval. Then the conscious perception of the quantum observer will not undergo random and
discontinuous change between the conscious perceptions corresponding to the brain states in
the superposition, as the time average of his brain state during a finite time interval contains
no randomness. In other words, his conscious perception will be not random but fixed16 .
This is also inconsistent with the above experience.
To sum up, the above analysis shows that the de Broglie-Bohm theory and the many-
worlds interpretation are inconsistent with protective measurement and the resulting inter-
pretation of the wave function in terms of random discontinuous motion of particles. If
there are no hidden variables besides the wave function, then the state of a quantum system
including a measuring device will be represented only by its wave function. If there are
no many worlds either, then a definite measurement result, which is usually denoted by a
definite position of the pointer of a measuring device, will be represented by a local wave
packet of the pointer, rather than by a superposition of local wave packets. As a result,
the transition from microscopic uncertainty to macroscopic certainty (e.g. the emergence of
definite measurement results) can only be achieved by the collapse of the wave function. In
other words, wavefunction collapse will be a real physical process.
As noted earlier, however, the existing ontologies of the dynamical collapse theories that
admit the reality of wavefunction collapse, such as mass density ontology and flash ontology
15
As in the many-worlds case, the random discontinuous motion does not result in the emergence of many
minds either. Since the brain state of a quantum observer is definite and only assumes one brain state in
the superposition at a given instant, even if there are many minds with different conscious perceptions at the
instant, these perceptions are irrelevant to those corresponding to the brain states in the superposition except
the present brain state. Thus such a theory of many minds cannot be consistent with the above experience.
In addition, although the quantum observer has a dispositional property relating to his superposition state,
the property is still a definite property of the unique observer and thus cannot correspond to the existence of
many minds.
16
Moreover, it can be expected that the conscious perception of the observer is none of the perceptions
corresponding to the brain states in the superposition because these states have the same status.

55
 4.2. A CONJECTURE ON THE ORIGIN OF WAVEFUNCTION COLLAPSE

(Ghirardi, Grassi and Benatti 1995; Ghirardi 1997, 2008; Allori et al 2008), is inconsistent
with the results of protective measurement and the picture of random discontinuous motion
of particles. For example, the existence of the effective mass and charge density of a quan-
tum system already excludes the mass density ontology, which claims that the mass density
is real (Ghirardi 2008). In addition, the existing dynamical collapse theories are still phe-
nomenological models, and they are also plagued by some serious problems such as energy
non-conservation etc (Pearle 2007, 2009). In particular, the physical origin of the wavefunc-
tion collapse, including the origin of the randomness of the collapse process, is still unknown,
though there are already some interesting conjectures (see, e.g. Diósi 1989; Penrose 1996).
In the subsequent sections, we will try to solve these problems and propose a new dynamical
collapse model in terms of the random discontinuous motion of particles. A more detailed
review of the existing dynamical collapse theories will be given in the last section.

4.2 A conjecture on the origin of wavefunction collapse

It is well known that a ‘chooser’ and a ‘choice’ are needed to bring the required dynamical
collapse of the wave function (Pearle 1999). The chooser is the noise source that collapses
the wave function, and the choices are the states toward which the collapse tends. In this
section, we will first analyze these two relatively easier problems and then investigate the
more difficult problem, the physical origin of wavefunction collapse.

4.2.1 The chooser in discrete time

To begin with, let’s analyze the chooser problem. In the existing dynamical collapse models,
the chooser is generally assumed to be an unknown classical noise field independent of the
collapsed wave function (Pearle 2007, 2009). If what the wave function describes is the
random discontinuous motion of particles, then it seems natural to assume that the random
motion of particles is the appropriate noise source to collapse the wave function. This has
three merits at least. First, the noise source and its properties are already known. For
example, the probability of the particles being in certain position, momentum and energy at
each instant is given by the modulus square of their wave function at the instant. Next, this
noise source is not a classical field, and thus the model can avoid the problems introduced
by the field such as the problem of infinite energy etc (Pearle 2009). Last but not least, the
random discontinuous motion of particles can also manifest itself in the laws of motion by
introducing the collapse evolution of the wave function. In the following, we will give a more
detailed analysis.
According to the picture of random discontinuous motion of particles, the wave function
of a quantum particle represents an instantaneous property of the particle that determines
its random discontinuous motion. However, the wave function is not a complete description
of the instantaneous state of the particle. The instantaneous state of the particle at a given
instant also includes its random position, momentum and energy at the instant, which may
be called the random part of the instantaneous state of the particle. Although the probability
of the particle being in each random instantaneous state is completely determined by the
wave function, its stay in the state at each instant is a new physical fact independent of the
wave function. Therefore, it seems natural to assume that the random stays of the particle
may have certain physical efficiency that manifests in the complete equation of motion17 .
Since the motion of the particle is essentially random, its stay at an instant does not influence
17
This is distinct from the case of continuous motion. For the latter, the position of a particle at each
instant is completely determined by the deterministic instantaneous condition at the instant, and thus the
position of the particle has no influence on the deterministic instantaneous condition.

56
 4.2. A CONJECTURE ON THE ORIGIN OF WAVEFUNCTION COLLAPSE

its stays at other instants in any direct way. Then the random stays of the particle can only
manifest themselves in the equation of motion by their influences on the evolution of the wave
function18 . This forms a feedback in some sense; the wave function of a particle determines
the probabilities of its stays in certain position, momentum and energy, while its random
stay at each instant also influences the evolution of the wave function in a stochastic way19 .
However, the existence of the stochastic influences on the evolution of the wave function
relies on an important precondition: the discreteness of time. If time is continuous and
instants are durationless, the random stays of a particle can have no stochastic influence
on anything. The reason is as follows. First, the duration of each random stay of the
particle is zero in continuous time. Due to the randomness of motion, when there are at
least two possible instantaneous states a particle can move between, the particle cannot stay
in the same instantaneous state throughout a finite time. For the joint probability of the
particle being in the same instantaneous state for all infinitely uncountable instants in the
finite time interval is obviously zero, and the total probability of the particle being in other
instantaneous states is not zero at any instant in between either. In other words, in order
that a particle stays in the same instantaneous state for a finite time, the probability of
the particle being in this instantaneous state must be one all the while during the entire
interval. This is possible only for the banal case where there is only one instantaneous state
the particle can stay and thus there is no motion and its randomness at all throughout the
duration20 .
Secondly, the influence of the random stay of a particle at a durationless instant is zero.
This can be readily understood. If a physical influence is not zero at each durationless instant,
then it may accumulate to infinite during an arbitrarily short time interval, which should
be avoided in physics. Lastly, the accumulated influence of the random stays during a finite
time interval, even if it can be finite21 , contains no randomness. For the discontinuity and
randomness of motion exist only at each durationless instant, during which the influence
of the random stay is zero, and they don’t exist during a finite time interval or even an
infinitesimal time interval. For example, the state of random discontinuous motion in real
space, which is defined during an infinitesimal time interval at a given instant, is described
by the position density and position flux density, and they are continuous quantities that
contain no discontinuity and randomness.
Therefore, if time is continuous and instants are durationless, then the random stays of a
particle can have no stochastic effects. This also means that the random stays of a particle
can influence the evolution of its wave function in a stochastic way only when instants are not
zero-sized but finite-sized, i.e., when time is discrete or quantized. Once the duration of each
random stay of a particle is finite, each random stay can have a finite stochastic influence
on the evolution of the wave function. It is worth stressing again that if time is not discrete
but continuous, a particle cannot stay in one of the infinitely many instantaneous states
all through for a finite time; rather, it can only stay there for one zero-sized instant. By
contrast, if time is discrete and instants are not zero-sized but finite-sized, even if a particle
18
In fact, since the random stays of a particle as one part of its instantaneous state are completely random,
the complete evolution equation of the instantaneous state of the particle is only about the evolution of the
wave function. Therefore, the random stays of the particle can only manifest themselves in the complete
equation of motion by their stochastic influences on the evolution of the wave function.
19
In other words, the wave function of a particle determines its random discontinuous motion, while the
motion also influences the evolution of the wave function reciprocally.
20
Unfortunately, this banal case does not exist. Due to the uncertainty relation between position and
momentum in quantum mechanics, there are always infinitely many different instantaneous states (with
definite position and momentum) where a particle can stay at any time.
21
Our analysis of a concrete model in the next section will show that under some reasonable assumptions the
accumulated influence of the random stays during a finite time interval is still zero when time is continuous.

57
 4.2. A CONJECTURE ON THE ORIGIN OF WAVEFUNCTION COLLAPSE

stays in an instantaneous state only for one instant, the duration of its stay is also finite as
the instant is finite-sized. In some sense, the discreteness of time prevents a particle from
jumping from its present instantaneous state to another instantaneous state and makes the
particle stay in the present instantaneous state all through during each finite-sized instant22 .
Since it has been conjectured that the Planck scale is the minimum spacetime scale23 , we
will assume that the size of each discrete instant or the quantum of time is the Planck time
in our following analysis24 .
To sum up, the realization of the randomness and discontinuity of motion in the laws
of motion requires that time is discrete. In discrete time, a particle randomly stays in an
instantaneous state with definite position, momentum and energy at each discrete instant,
with a probability determined by the modulus square of its wave function at the instant.
Each random, finite stay of the particle may have a finite influence on the evolution of its wave
function. As we will show in the next section, the accumulation of such discrete and random
influences may lead to the correct collapse of the wave function, which can then explain the
emergence of definite measurement results. Accordingly, the evolution of the wave function
will be governed by a revised Schrödinger equation, which includes the normal linear terms
and a stochastic nonlinear term that describes the discrete collapse dynamics. Note that
the wave function (as an instantaneous property of particles) also exists in the discrete time,
which means that the wave function does not change during each discrete instant, and the
evolution of the wave function including the linear Schrödinger evolution is also discrete.

4.2.2 Energy conservation and the choices

Now let’s investigate the choice problem, namely the problem of determining the states
toward which the collapse tends. The random stay of a particle may have a stochastic
influence on the evolution of its wave function at each discrete instant. If the stochastic
influences accumulate and result in the collapse of the wave function, then what are the states
toward which collapse tends? This is the choice problem or preferred basis problem. It may
be expected that the stochastic influences of the motion of a particle on its wave function
should not be arbitrary but be restricted by some fundamental principles. In particular,
it seems reasonable to assume that the resulting dynamical collapse of the wave function
should also satisfy the conservation of energy. As a result, the collapse states or choices will
22
This means that the minimum duration of the random stay of a particle in a definite position or momentum
or energy is always a discrete instant. It can be imagined that the duration of the random stay of a particle
in an eigenvalue of energy is a discrete instant, but the duration of its random stay in each position is still
zero as in continuous time. In this case, however, the position probability distribution of the particle cannot
be uniquely determined during its stay in the definite energy for a general state of motion where the energy
branches are not wholly separated in space. Moreover, it seems that only the duration of the random stay
of a particle in the eigenvalue of every property is the same can the (objective) probability distributions of
all these properties be consistent with those given by the modulus square of the wave function in quantum
mechanics.
23
Note that the existing arguments, which are based on some sort of combination of quantum theory and
general relativity (see, e.g. Garay 1995 for a review), do not imply but only suggest that space and time
are discrete. Moreover, the meanings and realization of discrete spacetime are also different in the existing
models of quantum gravity.
24
It has been conjectured that a fundamental theory of physics may be formulated by three natural con-
stants: the Planck time (tP ), the Planck length (lP ) and the Planck constant (~), and all other physical con-
stants are expressed by the combinations of them (Gao 2006b). For example, the speed of light is c = lP /tP ,
and the Einstein gravitational constant is κ = 8πlP tP /~. In this sense, the quantum motion in discrete space
and time, represented by the above three constants, is more fundamental than the phenomena described by
the special and general theory of relativity. However, even if this conjecture turns out to be right, it is still a
big challenge how to work out the details (see Gao 2011c for an initial attempt).

58
 4.2. A CONJECTURE ON THE ORIGIN OF WAVEFUNCTION COLLAPSE

be the energy eigenstates of the total Hamiltonian of a given system25 . In the following,
we will give a more detailed analysis of the consequences of this assumption. Its possible
physical basis will be investigated in the next subsection.
As we have argued in the last chapter, for a deterministic evolution of the wave function
such as the linear Schrödinger evolution, the requirement of energy conservation applies to a
single isolated system. However, for a stochastic evolution of the wave function such as the
dynamical collapse process, the requirement of energy conservation cannot apply to a single
system in general but only to an ensemble of identical systems26 . It can be proved that only
when the collapse states are energy eigenstates of the total Hamiltonian for each identical
system in an ensemble, can energy be conserved at the ensemble level for wavefunction
collapse (See Pearle 2000 for a more detailed analysis). Note that for the linear Schrödinger
evolution under an external potential, energy is conserved but momentum is not conserved
even at the ensemble level, and thus it is not momentum conservation but energy conservation
that is a more universal restriction for wavefunction collapse.
The conservation of energy can not only help to solve the preferred basis problem, but
also further determine the law of dynamical collapse to a large extent. For each system in
the same quantum state in an ensemble, in order that the probability distribution of energy
eigenvalues of the state can keep constant for the whole ensemble (i.e. energy is conserved
at the ensemble level), the random stay of the system at each discrete instant can only
change its (objective) energy probability distribution27 , and moreover, the change must also
satisfy a certain restriction. Concretely speaking, the random stay in a definite energy Ei
will increase the probability of the energy branch |Ei > and decrease the probabilities of all
other energy branches pro rata. Moreover, the increasing amplitude must be proportional to
the total probability of all other energy branches, and the coefficient is related to the energy
uncertainty of the state. We will demonstrate this result in the next subsection.
A more important problem is whether this energy-conserved collapse model can explain
the emergence of definite measurement results and our macroscopic experience. At first
sight the answer appears negative. For example, the energy eigenstates being collapse states
seems apparently inconsistent with the localization of macroscopic objects. However, a
detailed analysis given in the subsequent subsections will demonstrate that the model can
be consistent with existing experiments and our macroscopic experience. The key is to
realize that the energy uncertainty driving the collapse of the entangled state of a many-
body system is not the uncertainty of the total energy of all sub-systems, but the sum of
the absolute energy uncertainty of every sub-system. As a result, the collapse states are
the product states of the energy eigenstates of the Hamiltonian of each sub-system for a
non-interacting or weakly-interacting many-body system. This gives a further collapse rule
for the superpositions of degenerate energy eigenstates of a many-body system.

4.2.3 In search of a deeper basis

In this subsection, we will investigate the possible physical basis of the energy conservation
restriction for wavefunction collapse.
It is well known that the conservation of energy refers to an ensemble of identical systems
in standard quantum mechanics. However, this standard view seems unnatural when the
25
For the superpositions of degenerate energy eigenstates of a many-particle system, a further collapse rule
is needed. We will discuss this issue later on.
26
As we will see later, the conservation of energy may also hold true at the individual level for the collapse
evolution of some special wave functions.
27
If the phase of an energy eigenstate also changes with time, then the probability distribution of energy
eigenvalues will in general be changed for each identical system in the ensemble, and as a result, energy will
be not conserved even at the ensemble level.

59
 4.2. A CONJECTURE ON THE ORIGIN OF WAVEFUNCTION COLLAPSE

wave function represents the physical state of a single system, e.g. the state of random
discontinuous motion of particles. An ensemble is not an actual system after all, and the
conservation of something for an ensemble seems physically meaningless. Moreover, since a
single system in the ensemble does not ‘know’ the other systems and the whole ensemble,
there must exist some underlying mechanism that can ensure the conservation of energy
for an ensemble. Then the conservation of energy for an ensemble of identical systems is
probably a result of the laws of motion for individual systems in the ensemble. Here is
a possible scheme. First of all, energy is conserved for the evolution of individual energy
eigenstates. Next, a superposition of energy eigenstates will dynamically collapse to one of
these energy eigenstates, and the probability of the collapse result satisfies the Born rule.
Then the wavefunction collapse will satisfy the conservation of energy for an ensemble of
identical systems.
In the following, we will further suggest a possible physical basis for this scheme of energy-
conserved wavefunction collapse. According to the picture of random discontinuous motion,
for a particle in a superposition of energy eigenstates, the particle stays in an instantaneous
state with a definite energy eigenvalue at a discrete instant, and at another instant it may
jump to another instantaneous state with another energy eigenvalue. It seems to be a reason-
able assumption that the particle has both the tendency to jump among the instantaneous
states with different energies and the tendency to stay in the instantaneous states with the
same energy, and their relative strength is determined by the energy uncertainty of the parti-
cle. This is satisfactory in logic, as there should exist two opposite tendencies in general, and
their relative strength is determined by certain condition. In some sense, the two tendencies
of a particle are related to the two parts of its instantaneous state; the jumping tendency
is related to the wave function, and it is needed to manifest the superposition of different
energy eigenstates, while the staying tendency is related to the random stays. These two
opposite tendencies together constitute the complete “temperament” of a particle.
It can be argued that the tendency to stay in the same energy for individual particles
may be the physical origin of the energy-conserved wavefunction collapse. For a particle
in a superposition of energy eigenstates, the particle stays in an instantaneous state with
definite energy at a discrete instant, and the staying tendency of the particle will increase its
probability of being in the instantaneous states with the present energy at next instant. In
other words, the random stay of a particle in an instantaneous state with an energy eigenvalue
will increase the probability of the energy eigenvalue (and correspondingly decrease the
probabilities of other energy eigenvalues pro rata). Moreover, the increase of probability
may relate to the energy uncertainty of the particle. By the continuity of the change of
staying tendency, the particle will jump more readily among the instantaneous states with
small energy uncertainty and more hardly among the instantaneous states with large energy
uncertainty (which can also be regarded as a restriction of energy change). Thus the larger
the energy uncertainty of the superposition is, the larger the increase of probability is during
each random stay. A detailed calculation, which will be given in the next section, shows
that such a random change of energy probability distribution can continuously accumulate
to lead to the collapse of the superposition of energy eigenstates to one of them.
It can be further argued that the probability distribution of energy eigenvalues should
remain constant during the random evolution of an ensemble of identical systems, and thus
the resulting wavefunction collapse will satisfy the Born rule. The reason is as follows. When
an initial superposition of energy eigenstates undergoes the dynamical collapse process, the
probability distribution of energy eigenvalues should manifest itself through the collapse
results for an ensemble of identical systems. At a deeper level, it is very likely that the
laws of nature permit nature to manifest itself, or else we will be unable to find the laws of
nature and verify them by experiments, and our scientific investigations will be also pointless.

60
 4.3. A DISCRETE MODEL OF ENERGY-CONSERVED WAVEFUNCTION
COLLAPSE

This may be regarded as a meta-law. Since the collapse evolution of individual systems is
completely random and irreversible, the diagonal density matrix elements for an ensemble
of identical systems must be precisely the same as the initial probability distribution at
every step of the evolution. Otherwise the frequency distribution of the collapse results
in the ensemble cannot reflect the initial probability distribution, or in other words, the
probability information contained in the initial state will be completely lost due to the
random and irreversible wavefunction collapse28 . As a consequence, the collapse evolution
will conserve energy at the ensemble level, and the collapse results will also satisfy the Born
rule in quantum mechanics.
Certainly, there is still a question that needs to be answered. Why energy? Why not
position or momentum? If there is only one property that undergoes the random discon-
tinuous motion (e.g. position), then the above tendency argument for the unique property
may be satisfying. But if there are many properties that undergoes the random discontin-
uous motion, then we need to answer why the tendency argument applies only to energy.
A possible answer is that energy is the property that determines the linear evolution of the
state of motion, and thus it seems natural and uniform that energy also determines the
nonlinear collapse evolution. Moreover, energy eigenstates are the states of motion that no
longer evolve (except an absolute phase) for the linear evolution. Then by analogy, it is likely
that energy eigenstates are also the states that no longer evolve for the nonlinear collapse
evolution, i.e., that energy eigenstates are the collapse states. However, we may never be
able to reach (and know we reach) the end point of explanation. Another important task is
to develop a concrete model and compare it with experiments. We do this in the subsequent
sections.

4.3 A discrete model of energy-conserved wavefunction collapse

After giving a speculative analysis of the origin of wavefunction collapse in terms of
the random discontinuous motion of particles, we will propose a discrete model of energy-
conserved wavefunction collapse based on the results obtained from the analysis.
Consider a multi-level system with a constant Hamiltonian. Its initial state is:
m
X
|ψ(0)i = ci (0) |Ei i, (4.3)
i=1

where |Ei i is the energy eigenstate of the Hamiltonian of the system, P Ei is the corresponding
energy eigenvalue, and ci (0) satisfies the normalization relation m |c
i=1 i (0)| 2 = 1.

According to our conjecture on the origin of wavefunction collapse, this superposition of

energy eigenstates will collapse to one of the eigenstates after a discrete dynamical process,
and the collapse evolution satisfies the conservation of energy at the ensemble level. The
physical picture of the dynamical collapse process is as follows. At the initial discrete instant
t0 = tP (where tP is the Planck time), the system randomly stays in a branch |Ei i with
probability Pi (0) ≡ |ci (0)|2 .29 This finite stay slightly increases the probability of the staying
branch and decreases the probabilities of all other branches pro rata. Similarly, at any
discrete instant t = ntP the system randomly stays in a branch |Ei i with probability Pi (t) ≡
|ci (t)|2 , and the random stay also changes the probabilities of the branches slightly. Then
28
Note that the reversible Schrödinger evolution conserves the information even for individual isolated
systems.
29
Strictly speaking, the description “branch” should be replaced by “instantaneous state”, e.g. the branch
|Ei i should be replaced by the instantaneous state with energy Ei . Yet the branch description may be more
succinct and visual, and we will use it in the following discussions.

61
 4.3. A DISCRETE MODEL OF ENERGY-CONSERVED WAVEFUNCTION
COLLAPSE

during a finite time interval much larger than tP , the probability of each branch will undergo
a discrete and stochastic evolution. In the end, the probability of one branch will be close to
one, and the probabilities of other branches will be close to zero. In other words, the initial
superposition will randomly collapse to one of the energy branches in the superposition.
Now we will give a concrete analysis of this dynamical collapse process. Since the linear
Schrödinger evolution does not change the energy probability distribution, we may only
consider the influence of dynamical collapse on the energy probability distribution. Suppose
the system stays in branch |Ei i at the discrete instant t = ntP , and the stay changes the
probability of this branch, Pi (t), to

Pii (t + tP ) = Pi (t) + ∆Pi , (4.4)

where the superscript i denotes the staying branch, and ∆Pi is a functional of Pi (t). Due
to the conservation of probability, the increase of the probability of one branch can only
come from the scale-down of the probabilities of all other branches. This means that the
probability of another branch Pj (t) (j 6= i) correspondingly turns to be30

Pj (t)∆Pi
Pji (t + tP ) = Pj (t) − , (4.5)
1 − Pi (t)
where the superscript i still denotes the staying branch. The probability of this random
stay at the instant is p(Ei , t) = Pi (t). Then we can work out the diagonal density matrix
elements of the evolution31 :

m
p(Ej , t)Pij (t + tP )
X
ρii (t + tP ) =
j=1
X Pi (t)∆Pj (t)
= Pi (t)[Pi (t) + ∆Pi ] + Pj (t)[Pi (t) − ]
1 − Pj (t)
j6=i
X ∆Pj (t)
= ρii (t) + Pi (t)[∆Pi − Pj (t) ]. (4.6)
1 − Pj (t)
j6=i

Here we shall introduce the first rule of dynamical collapse, which says that the probabil-
ity distribution of energy eigenvalues for an ensemble of identical systems is constant during
the dynamical collapse process. As we have argued in the last subsection, this rule is required
by the principle of energy conservation at the ensemble level, and it may also have a physical
basis relating to the manifestability of nature. By this rule, we have ρii (t + tP ) = ρii (t) for
any i. This leads to the following set of equations:

X Pj (t)∆Pj (t)
∆P1 (t) − = 0,
1 − Pj (t)
j6=1
X Pj (t)∆Pj (t)
∆P2 (t) − = 0,
1 − Pj (t)
j6=2
...
X Pj (t)∆Pj (t)
∆Pm (t) − = 0. (4.7)
1 − Pj (t)
j6=m
30
One can also obtain this result by first increasing the probability of the staying branch and then normal-
Pj (t)
izing the probabilities of all branches. This means that Pi (t + tP ) = Pi1+∆(t)+∆
and Pj (t + tP ) = 1+∆ for any
∆ ∆
j 6= i. In this way, we have ∆Pi = 1+∆ (1 − Pi (t)) and ∆Pj = 1+∆ Pj (t) for any j 6= i.
31
The density matrix describes the ensemble of states which arise from all possible random stays.

62
 4.3. A DISCRETE MODEL OF ENERGY-CONSERVED WAVEFUNCTION
COLLAPSE

By solving this equations set (e.g. by subtracting each other), we find the following relation
for any i:

∆Pi
= k, (4.8)
1 − Pi (t)
where k is an undetermined dimensionless quantity that relates to the state |ψ(t)i.
By using Eq. (4.8), we can further work out the non-diagonal density matrix elements of
the evolution. But it is more convenient to calculate the following variant of non-diagonal
density matrix elements:

m
X
ρij (t + tP ) = p(El , t)Pil (t + tP )Pjl (t + tP )
l=1
X
= Pl (t)[Pi (t) − kPi (t)][Pj (t) − kPj (t)]
l6=i,j
+Pi (t)[Pi (t) + k(1 − Pi (t))][Pj (t) − kPj (t)]
+Pj (t)[Pj (t) + k(1 − Pj (t))][Pi (t) − kPi (t)]
= (1 − k 2 )ρij (t). (4.9)

Since the usual collapse time, τc , is defined by the relation ρij (τc ) = 21 ρij (0), we may use a
proper approximation, where k is assumed to be the same as its initial value during the time
interval [0, τc ], to simplify the calculation of the collapse time. Then we have:

ρij (t) ≈ (1 − k 2 )n ρij (0). (4.10)

The corresponding collapse time is in the order of:
1
τc ≈ tP , (4.11)
k2
In the following, we shall analyze the formula of k defined by Eq. (4.8). To begin with,
the probability restricting condition 0 6 Pi (t) 6 1 for any i requires that 0 6 k 6 1. When
k = 0, no collapse happens, and when k = 1, collapse happens instantaneously. Note that k
cannot be smaller than zero, as this will lead to the negative value of Pi (t) in some cases. For
|k|
instance, when k is negative and Pi (t) < 1+|k| , Pi (t+tP ) = Pi (t)+k[1−Pi (t)] will be negative
and violate the probability restricting condition. That k is positive indicates that each
random stay increases the probability of the staying branch and decreases the probabilities
of other branches, which is consistent with the analysis given in the last subsection.
Next, k is proportional to the duration of stay. The influence of each stay on the proba-
bility of the staying branch is an accumulating process. When the duration of stay is zero as
in continuous time, no influence exists and no collapse happens. When the duration of stay,
tP , is longer, the probability of the staying branch will increase more. Thus we have k ∝ tP .
Thirdly, k is also proportional to the energy uncertainty of the superposition of energy
eigenstates. First, from a dimensional analysis k should be proportional to an energy term
in order to cancel out the dimension of time. Next, the energy term should be the energy
uncertainty of the superposition defined in an appropriate way according to the analysis of
the last subsection. When the energy uncertainty is zero, i.e., when the state is an energy
eigenstate, no collapse happens. When the energy uncertainty is not zero, collapse happens.
Moreover, the larger the energy uncertainty is, the larger the increase of the probability of the
staying branch for each random stay is, namely the larger k is. Therefore, k is proportional
to the energy uncertainty of the superposition. How to define the energy uncertainty then?

63
 4.3. A DISCRETE MODEL OF ENERGY-CONSERVED WAVEFUNCTION
COLLAPSE

Since k is invariant under the swap of any two branches (Pi , Ei ) and (Pj , Ej ) according to
Eq. (4.8), the most natural definition of the energy uncertainty of a superposition of energy
eigenstates is32 :
m
1 X
∆E = Pi Pj |Ei − Ej |. (4.12)
2
i,j=1

For the simplest two-level system, we have

∆E = P1 P2 |E1 − E2 |. (4.13)
It seems a little counterintuitive that k contains the energy uncertainty term that relates
to the whole energy distribution. The puzzle is two-fold. First, this means that the increase
of the probability of the staying branch relates not to the energy difference between the
staying branch and all other branches, but to the energy uncertainty of the whole state.
This is reflected in the formula of ∆E in the existence of the energy difference between
any two branches, |Ei − Ej | for any i and j. Next, the increase of the probability of the
staying branch relates also to the energy probability distribution that determines the energy
uncertainty. This is reflected in the formula of ∆E in the existence of Pi Pj . In fact, these
seemingly puzzling aspects are still understandable. The first feature is required by the first
rule of dynamical collapse that ensures energy conservation at the ensemble level. This can
be clearly seen from Eq. (4.8). If the increase of the probability of the staying branch relates
to the difference between the energy of the staying branch and the average energy of all other
branches, then Eq. (4.8) will not hold true because the swap symmetry of k will be violated,
and as a result, the first rule of dynamical collapse will be broken. The second feature can
be understood as follows. In the picture of random discontinuous motion, the probability
distribution contains the information of staying time distribution. An energy branch with
small probability means that the system jumps through it less frequently. Thus this energy
branch only makes a small contribution to the restriction of energy change or the increase of
the staying tendency. As a result, the increase of the probability of the staying branch and
k will relate not only to energy difference, but also to the energy probability distribution.
Then after omitting a coefficient in the order of unity, we can get the formula of k in the
first order:

k ≈ ∆EtP /~. (4.14)

This is the second rule of dynamical collapse. By inputting Eq. (4.14) into Eq. (4.11), we
can further get the collapse time formula:

~EP
τc ≈ , (4.15)
(∆E)2
where EP = h/tP is the Planck energy, and ∆E is the energy uncertainty of the initial
state33 .
Here it is worth pointing out that k must contain the first order term of ∆E. For
the second order or higher order term of ∆E will lead to much longer collapse time for
some common measurement situations, which contradicts experiments (Gao 2006a, 2006b).
Besides, a similar analysis of the consistency with experiments may also provide a further
32
Note that the common RMS (mean square root) uncertainty also satisfies the swap symmetry. Thus it
still needs to be studied what the exact form of k is.
33
This collapse time formula indicates that there is no wavefunction collapse in continuous time because
tP → 0 leads to τc → ∞. A premise of this conclusion is that the influence of each random stay is proportional
to the duration of stay.

64
 4.3. A DISCRETE MODEL OF ENERGY-CONSERVED WAVEFUNCTION
COLLAPSE

support for the energy-conserved collapse model in which the collapse states are energy
eigenstates. First of all, if the collapse states are not energy eigenstates but momentum
eigenstates, then the energy uncertainty will be replaced by momentum uncertainty in the
~EP
collapse time formula Eq. (4.15), namely τc ≈ (∆pc) 2 . As a result, the collapse time will

be too short to be consistent with experiments for some situations. For example, for the
ground state of hydrogen atom the collapse time will be about several days. Note that the
second order or higher order term of ∆p will also lead to much longer collapse time for some
common measurement situations, which contradicts experiments.
Next, if the collapse states are position eigenstates34 , then the collapse time formula
l2 tP
Eq. (4.15) will be replaced by something like τc ≈ (∆x) 2 , where l is certain length scale

relating to the collapsing state. No matter what length scale l is, the collapse time of a
momentum eigenstate will be zero as its position uncertainty is infinite. This means that
the momentum eigenstates of any quantum system will collapse instantaneously to one of
its position eigenstates and thus cannot exist. Moreover, the superposition states with very
small momentum uncertainty will also collapse very quickly even for microscopic particles.
These results are apparently inconsistent with quantum mechanics. Although it may be
possible to adjust the length scale l to make the model consistent with existing experience,
the collapse time formula will be much more complex than that in the above energy-conserved
collapse model. Let’s give a little more detailed analysis here. There are two universal
length scales for a quantum system: its Compton wavelength λc and the Planck length lP .
It is obvious that both of them cannot be directly used as the length scale in the collapse
l2 tP
time formula τc ≈ (∆x) 2 . Then the formula can only be written in a more complex form:
2
λc tP
τc ≈ ( λlPc )α · (∆x)2 . Moreover, experiments such as the SQUID experiments and our everyday

macroscopic experience require α ≈ 8. It seems very difficult to explain this unusually large
exponent in theory. To sum up, the collapse states can hardly be position eigenstates when
considering the consistency with experiments and the simplicity of theory.
Based on the above analysis, the state of the multi-level system at instant t = ntP will
be:
m
X
|ψ(t)i = ci (t)e−iEi t/~ |Ei i, (4.16)
i=1

Besides the linear Schrödinger evolution, the collapse dynamics adds a discrete stochastic
evolution for Pi (t) ≡ |ci (t)|2 :

∆E
Pi (t + tP ) = Pi (t) + [δEs Ei − Pi (t)], (4.17)
EP
where ∆E is the energy uncertainty of the state at instant t defined by Eq. (4.12), Es is a
random variable representing the random stay of the system, and its probability of assuming
Ei at instant t is Pi (t). When Es = Ei , δEs Ei = 1, and when Es 6= Ei , δEs Ei = 0.
This equation of dynamical collapse can be directly extended to the entangled states of
a many-body system. The difference only lies in the definition of the energy uncertainty
∆E. As noted in the last subsection, for a non-interacting or weakly-interacting many-
body system in an entangled state, for which the energy uncertainty of each sub-system can
be properly defined, ∆E is the sum of the absolute energy uncertainty of all sub-systems,
namely
34
In continuous space and time, a position eigenstate has infinite average energy and cannot be physically
real. But in discrete space and time, position eigenstates will be the states whose spatial dimension is about
the Planck length, and they may exist.

65
 4.3. A DISCRETE MODEL OF ENERGY-CONSERVED WAVEFUNCTION
COLLAPSE

n m
1XX
∆E = Pi Pj |Eli − Elj |, (4.18)
2
l=1 i,j=1

where n is the total number of the entangled sub-systems, m is the total number of energy
branches in the entangled state, and Eli is the energy of sub-system l in the i -th energy
branch of the state. Correspondingly, the collapse states are the product states of the energy
eigenstates of the Hamiltonian of each sub-system. It should be stressed here that ∆E is
not defined as the uncertainty of the total energy of all sub-systems as in the energy-driven
collapse models (see, e.g. Percival 1995, 1998a; Hughston 1996). For each sub-system has its
own energy uncertainty that drives its collapse, and the total driving “force” for the whole
entangled state should be the sum of the driving “forces” of all sub-systems, at least in
the first order approximation. Although these two kinds of energy uncertainty are equal in
numerical values in some cases (e.g. for a strongly-interacting many-body system), there are
also some cases where they are not equal. For example, for a superposition of degenerate
energy eigenstates of a non-interacting many-body system, which may arise during a common
measurement process, the uncertainty of the total energy of all sub-systems is exactly zero,
but the absolute energy uncertainty of each sub-system and their sum may be not zero. As
a result, the superpositions of degenerate energy eigenstates of a many-particle system may
also collapse. As we will see later, this is an important feature of our model, which can avoid
Pearle’s (2004) serious objections to the energy-driven collapse models.
It can be seen that the equation of dynamical collapse, Eq.(4.17), has an interesting
property, scale invariance. After one discrete instant tP , the probability increase of the
staying branch |Ei i is ∆Pi = ∆E EP (1 − Pi ), and the probability decrease of the neighboring
∆E
branch |Ei+1 i is ∆Pi+1 = EP Pi+1 . Then the probability increase of these two branches is

∆E
∆(Pi + Pi+1 ) = [1 − (Pi + Pi+1 )]. (4.19)
EP
Similarly, the equation ∆P = ∆E EP (1 − P ) holds true for the total probability of arbitrarily
many branches (one of which is the staying branch). This property of scale invariance may
simplify the analysis in many cases. For instance, for a superposition of two wavepackets
with energy difference, ∆E12 , much larger than the energy uncertainty of each wavepacket,
∆E1 = ∆E2 , we can calculate the collapse dynamics in two steps. First, we use Eq.(4.17)
and Eq.(4.13) with |E1 − E2 | = ∆E12 to calculate the time of the superposition collapsing
into one of the two wavepackets35 . Here we need not to consider the almost infinitely many
energy eigenstates constituting each wavepacket and their probability distribution. Next, we
use Eq.(4.17) with ∆E = ∆E1 to calculate the time of the wavepacket collapsing into one
of its energy eigenstates. In general, this collapse process is so slow that its effect can be
ignored.
Lastly, we want to stress another important point. As we have argued before, the dis-
continuity of motion requires that the collapse dynamics must be discrete in nature, and
moreover, the collapse states must be energy eigenstates in order that the collapse dynamics
satisfies the conservation of energy at the ensemble level. As a result, the energy eigenvalues
must be also discrete for any quantum system. This result seems to contradict quantum
mechanics, but when considering that our universe has a finite size (i.e. a finite event hori-
zon), the momentum and energy eigenvalues of any quantum system in the universe may be
35
Note that most collapse states in an ensemble of identical systems keep the shape of the wavepacket
almost precisely.

66
 4.4. ON THE CONSISTENCY OF THE MODEL AND EXPERIMENTS

indeed discrete36 . The reason is that all quantum systems in the universe are limited by the
finite horizon, and thus no free quantum systems exist in the strict sense. For example, the
hc
energy of a massless particle (e.g. photon) can only assume discrete values En = n2 4R U
,
hc −33 25
and the minimum energy is E1 = 4RU ≈ 10 eV , where RU ≈ 10 m is the radius of the
horizon of our universe37 . Besides, for a free particle with mass m0 , its energy also assumes
2
discrete values En = n2 32mh R2 . For instance, the minimum energy is E1 ≈ 10−72 eV for free
0 U
electrons, which is much smaller than the minimum energy of photons38 .
It is interesting to see whether this tiny discreteness of energy makes the collapse dy-
namics more abrupt. Suppose the energy uncertainty of a quantum state is ∆E ≈ 1eV , and
its energy ranges between
q the minimum energy E1 and 1eV . Then we can get the maximum
1eV
energy level lmax ≈ 10−33 eV
≈ 1016 . The probability of most energy eigenstates in the
superposition will be about P ≈ 10−16 . During each discrete instant tP , the probability
increase of the staying energy branch is ∆P ≈ ∆E −28 . This indicates that the
EP (1 − P ) ≈ 10
probability change during each random stay is still very tiny. Only when the energy uncer-
tainty is larger than 1023 eV or 10−5 EP , will the probability change during each random stay
be sharp. Therefore, the collapse evolution is still very smooth for the quantum states with
energy uncertainty much smaller than the Planck energy.

4.4 On the consistency of the model and experiments

In this section, we will analyze whether the discrete model of energy-conserved wave-
function collapse is consistent with existing experiments and our macroscopic experience.
Note that Adler (2002) has already presented a detailed consistency analysis in the context
of energy-driven collapse models, and as we will see below, most of his analysis also applies
to our model.

4.4.1 Maintenance of coherence

First of all, the model satisfies the constraint of predicting the maintenance of coherence
when this is observed. Since the energy uncertainty of the state of a microscopic particle is
very small in general, its collapse will be too slow to have any detectable effect in present
experiments on these particles. For example, the energy uncertainty of a photon emitted from
an atom is in the order of 10−6 eV , and the corresponding collapse time is 1025 s according to
Eq. (4.15) of our collapse model, which is much longer than the age of the universe, 1017 s.
This means that the collapse states (i.e. energy eigenstates) are never reached for a quantum
system with small energy uncertainty even during a time interval as long as the age of the
universe. As another example, consider the SQUID experiment of Friedman et al (2000),
where the coherent superpositions of macroscopic states consisting of oppositely circulating
36
There might exist a subtle connection here. It seems that the energy-conserved wavefunction collapse in
discrete time requires a finite event horizon to ensure the energy eigenvalues of any system are discrete. On
the other hand, it seems that discrete spacetime permits the existence of quantum fluctuations of spacetime
(as a possible form of dark energy) to lead to acceleration and finite event horizon (Gao 2005). In any case,
the existence of a cosmological constant also leads to the existence of a finite event horizon.
37
Note that the present upper bound on the photon mass is about mγ < 10−18 eV /c2 (Nakamura et al,
2010).
38
Whether this heuristic analysis is (approximately) valid depends on the application of the final theory
of quantum gravity to our finite universe. However, it is worth noting that the existence of discrete energy
levels for a free quantum system limited in our universe is also supported by the hypothetical holographic
principle, which implies that the total information within a universe with a finite event horizon is finite. If
the energy of a quantum system is continuous, then the information contained in the system will be infinite.

67
 4.4. ON THE CONSISTENCY OF THE MODEL AND EXPERIMENTS

supercurrents are observed. In the experiment, each circulating current corresponds to the
collective motion of about 109 Cooper pairs, and the energy uncertainty is about 8.6×10−6 eV .
Eq. (4.15) predicts a collapse time of 1023 s, and thus maintenance of coherence is expected
despite the macroscopic structure of the state39 . For more examples see Adler (2002).

4.4.2 Rapid localization in measurement situations

In the following, we will investigate whether the discrete model of energy-conserved wave-
function collapse can account for the emergence of definite measurement results.
Let’s first see a simple position measurement experiment. Consider an initial state de-
scribing a particle in a superposition of two locations (e.g. a superposition of two Gaussian
wavepackets separated by a certain distance). After the measurement interaction, the po-
sition measuring device evolves to a superposition of two macroscopically distinguishable
states:

(c1 ψ1 + c2 ψ2 )ϕ0 → c1 ψ1 ϕ1 + c2 ψ2 ϕ2 , (4.20)

where ψ1 , ψ2 are the states of the particle in different locations, ϕ0 is the initial state of the
position measuring device, and ϕ1 , ϕ2 are the different outcome states of the device. For an
ideal measurement, the two particle/device states ψ1 ϕ1 and ψ2 ϕ2 have precisely the same
energy spectrum. Then it appears that this superposition will not collapse according to the
energy-conserved collapse model.
However, this is not the case. The key is to see that the two states of the particle in
the superposition are detected in different parts of the measuring device, and they interact
with different atoms or molecules in these parts. Thus we should rewrite the device states
explicitly as ϕ0 = χA (0)χB (0), ϕ1 = χA (1)χB (0), and ϕ2 = χA (0)χB (1), where χA (0) and
χB (0) denote the initial states of the device in the parts A and B, respectively, and χA (1)
and χB (1) denote the outcome states of the device in the parts A and B, respectively. Then
we have

(c1 ψ1 + c2 ψ2 )χA (0)χB (0) → c1 ψ1 χA (1)χB (0) + c2 ψ2 χA (0)χB (1) (4.21)

This reformulation clearly shows that there exists energy difference between the sub-systems
in the different outcome states of the device. Since there is always some kind of measurement
amplification from the microscopic state to the macroscopic outcome in the measurement
process, there is a large energy difference between the states χA (0), χB (0) and χA (1), χB (1).
As a result, the total energy difference ∆E = |∆EA | + |∆EB | is also very large, and it will
result in the rapid collapse of the above superposition into one of its branches according to
the energy-conserved collapse model40 .
Let’s see a more realistic example, a photon being detected via photoelectric effect (e.g.
by a single-photon avalanche diode). In the beginning of the detection, the spreading spatial
wave function of the photon is entangled with the states of a large number of surface atoms
of the detector. In each local branch of the entangled state, the total energy of the photon
39
A potentially more promising case is provided by certain long-lived nuclear isomers, which have large
energy gaps from their ground states (see Adler 2002 and references therein). For example, the metastable
isomer of 180 Ta, the only nuclear isomer to exist naturally on earth, has a half-life of more than 1015 years
and an energy gap of 75keV from the ground state. According to Eq. (4.15), a coherent superposition of the
ground state and metastable isomer of 180 Ta will spontaneously collapse to either the isomeric state or the
ground state, with a collapse time of order 20 minutes. It will be a promising way to test our collapse model
by examining the maintenance of coherence of such a superposition.
40
Since the uncertainty of the total energy of the whole entangled system is still zero, the energy-driven
collapse models will predict that no wavefunction collapse happens and no definite measurement result appears
for the above measurement process, which contradicts experimental observations (Pearle 2004).

68
 4.4. ON THE CONSISTENCY OF THE MODEL AND EXPERIMENTS

is wholly absorbed by the electron in the local atom interacting with the photon. This
is clearly indicated by the term δ(Ef − Ei − ~ω) in the transition rate of photoelectric
effect. The state of the ejecting electron is a (spherical) wavepacket moving outward from
the local atom, whose average direction and momentum distribution are determined by the
momentum and polarization of the photon. The small energy uncertainty of the photon will
also be transferred to the ejecting electron41 .
This microscopic effect of ejecting electron is then amplified (e.g. by an avalanche process
of atoms) to form a macroscopic signal such as the shift of the pointer of a measuring device.
During the amplification process, the energy difference is constantly increasing between the
branch in which the photon is absorbed and the branch in which the photon is not absorbed
near each atom interacting with the photon. This large energy difference will soon lead to
the collapse of the whole superposition into one of the local branches, and thus the photon
is only detected locally. Take the single photon detector - avalanche photodiode as a typical
example42 . Its energy consumption is sharply peaked in a very short measuring interval.
One type of avalanche photodiode operates at 105 cps and has a mean power dissipation of
4mW (Gao 2006a). This corresponds to an energy consumption of about 2.5 × 1011 eV per
measuring interval 10−5 s. By using the collapse time formula Eq. (4.15), where the energy
uncertainty is ∆E ≈ 2.5 × 1011 eV , we find the collapse time is τc ≈ 1.25 × 10−10 s. This
collapse time is much smaller than the measuring interval.
One important point needs to be stressed here. Although a measured particle is detected
locally in a detector (e.g. the spatial size of its collapse state is in the order of the size of
an atom), its wave function does not necessarily undergo the position collapse as assumed
by standard quantum mechanics in an ideal position measurement, and especially, energy
can be conserved during the localization process according to our model. The reason can
be summarized as follows. The wave function of the measured particle is usually a spherical
wave (e.g. a spherically symmetric wave function) in three-dimensional space. Its momentum
is along the radial direction, but the local and random measurement result distributes along
the sphere, perpendicular to the radial direction. During the detection, the measured particle
interacts with a single atom of the detector by an ionizing process in each local branch of
the entangled state of the whole system including the particle and the atoms in the detector.
The particle is usually absorbed by the atom or bound in the atom, and its energy is wholly
transferred to the newly-formed atom and the ejecting electrons during the ionizing process
in each branch. Then the amplification process such as an avalanche process of atoms
41
In more general measurement situations, the measured particle (e.g. electron) is not annihilated by the
detector. However, in each local branch of the entangled state of the whole system, the particle also interacts
with a single atom of the detector by an ionizing process, and its total energy is also wholly transferred to
the atom and the ejecting electrons.
42
We take the widely-used Geiger counter as another illustration of the amplification process during mea-
surement. A Geiger counter is an instrument used to detect particles such as α particles, β particles and γ
rays etc. It consists of a glass envelope containing a low-pressure gas (usually a mixture of methane with
argon and neon) and two electrodes, with a cylindrical mesh being the cathode and a fine-wire anode running
through the centre of the tube. A potential difference of about 103 V relative to the tube is maintained between
the electrodes, therefore creating a strong electric field near the wire. The counter works on the mechanism
of gas multiplication. Ionization in the gas is caused by the entry of a particle. The ions are attracted to
their appropriate electrode, and they gain sufficient energy to eject electrons from the gas atoms as they pass
through the gas. This further causes the atoms to ionize. Therefore, electrons are produced continuously by
this process and rapid gas multiplication takes place (especially in the central electrode because of its strong
electric field strength). Its effect is that more than 106 electrons are collected by the central electrode for every
ion produced in the primary absorption process. These “electron avalanches” create electric pulses which then
can be amplified electronically and counted by a meter to calculate the number of initial ionization events.
In this way, a Geiger counter can detect low-energy radiation because even one ionized particle produces a
full pulse on the central wire. It can be estimated that the introduced energy difference during a detection is
∆E ≈ 109 eV , and the corresponding collapse time is τc ≈ 10−5 s according to our collapse model.

69
 4.4. ON THE CONSISTENCY OF THE MODEL AND EXPERIMENTS

introduces very large energy difference between the detected branch and the empty branch,
and as a result, the whole superposition will soon collapse into one of its local branches in
a random way according to the energy-conserved collapse model43 . After the collapse, the
state of the measured particle is localized in the spatial region of one atom. Moreover, since
all local branches of the entangled state of the particle and the detector have the same energy
spectrum, the collapse process also conserves energy at the individual level.

4.4.3 Emergence of the classical world

In this subsection, we will analyze whether the discrete model of energy-conserved wave-
function collapse is consistent with our macroscopic experience. It seems that there is an
apparent inconsistency here. According to the model, when there is a superposition of a
macroscopic object in an identical physical state (an approximate energy eigenstate) at two
different, widely separated locations, the superposition does not collapse, as there is no en-
ergy difference between the two branches of the superposition. But the existence of such
superpositions is obviously inconsistent with our macroscopic experience; the macroscopic
objects are localized. This common objection has been basically answered by Adler (2002).
The crux of the matter lies in the influences of environment. The collisions and especially
the accretions of environmental particles will quickly increase the energy uncertainty of the
entangled state of the whole system including the object and environmental particles, and
thus the initial superposition will soon collapse to one of the localized branches according
to our model. Accordingly, the macroscopic objects can always be localized due to environ-
mental influences. Note that the energy uncertainty here denotes the sum of the absolute
energy uncertainty of each sub-system in the entangled state as defined in our model44 .
As a typical example, we consider a dust particle of radius a ≈ 10−5 cm and mass
m ≈ 10−7 g. It is well known that localized states of macroscopic objects spread very slowly
under the free Schrödinger evolution. For instance, for a Gaussian wave packet with initial
(mean square) width ∆, the wave packet will spread so that the width doubles in a time
t = 2m∆2 /~. This means that the double time is almost infinite for a macroscopic object. If
the dust particle had no interactions with environment and its initial state is a Gaussian wave
packet with width ∆ ≈ 10−5 cm, the doubling time would be about the age of the universe.
However, if the dust particle is in interaction with environment, the situation turns out to
be very different. Although the different components that couple to the environment will be
individually incredibly localised, collectively they can have a spread that is many orders of
magnitude larger. In other words, the state of the dust particle and the environment could
be a superposition of zillions of very well localised terms, each with slightly different posi-
tions, and which are collectively spread over a macroscopic distance (Bacciagaluppi 2008).
According to Joos and Zeh (1985), the spread in an environment full of thermal radiation
only is proportional to mass times the cube of time for large times, namely (∆x)2 ≈ Λmτ 3 ,
where Λ is the localization rate depending on the environment, defined by the evolution
0 2
equation of density matrix ρt (x, x0 ) = ρ0 (x, x0 )e−Λt(x−x ) . For example, if the above dust
particle interacts with thermal radiation at T = 300K, the localization rate is Λ = 1012 , and
the overall spread of its state is of the order of 10m after a second (Joos and Zeh 1985). If
the dust particle interacts with air molecules, e.g. floating in the air, the spread of its state
will be much faster.
43
In a similar way, a spherically symmetric wave function will be detected as one linear track in a cloud
chamber (cf. Mott 1929).
44
The uncertainty of the total energy of the whole system is still very small even if the influences of
environment are counted. Thus no observable collapse happens for the above situation according to the
energy-driven collapse models (Pearle 2004).

70
 4.4. ON THE CONSISTENCY OF THE MODEL AND EXPERIMENTS

Let’s see whether the energy-conserved collapse in our model can prevent the above
spreading of the wave packet. Suppose the dust particle is in a superposition of two identical
localized states that are separated by 10−5 cm in space. The particle floats in the air, and
its average velocity is about zero. At standard temperature and pressure, one nitrogen
molecule accretes in the dust particle, which area is 10−10 cm2 , during a time interval of
10−14 s in average (Adler 2002). Since the mass of the dust particle is much larger than the
mass of a nitrogen molecule, the velocity change of the particle is negligible when compared
with the velocity change of the nitrogen molecules during the process of accretion. Then
the kinetic energy difference between an accreted molecule and a freely moving molecule
is about ∆E = 23 kT ≈ 10−2 eV . When one nitrogen molecule accretes in one localized
branch of the dust particle (the molecule is freely moving in the other localized branch), it
will increase the energy uncertainty of the total entangled state by ∆E ≈ 10−2 eV . Then
after a time interval of 10−4 s, the number of accreted nitrogen molecules is about 1010 ,
and the total energy uncertainty is about 108 eV . According to Eq. (4.15) in our collapse
model, the corresponding collapse time is about 10−4 s. Since the two localized states in the
superposition have the same energy spectra, the collapse also conserves energy.
In the energy-conserved collapse model, the collapse states are energy eigenstates, and
in particular, they are nonlocal momentum eigenstates for free quantum systems. Thus it is
indeed counterintuitive that the energy-conserved collapse can make the states of macroscopic
objects local. As shown above, this is due to the constant influences of environmental
particles. When the spreading of the state of a macroscopic object becomes larger, its
interaction with environmental particles will introduce larger energy difference between its
different local branches, and this will then collapse the spreading state again into a more
localized state45 . As a result, the states of macroscopic objects in an environment will
never reach the collapse states, namely momentum eigenstates, though they do continuously
undergo the energy-conserved collapse. To sum up, there are two opposite processes for a
macroscopic object constantly interacting with environmental particles. One is the spreading
process due to the linear Schrödinger evolution, and the other is the localization process due
to the energy-conserved collapse evolution. The interactions with environmental particles not
only make the spreading more rapidly but also make the localization more frequently. In the
end these two processes will reach an approximate equilibrium. The state of a macroscopic
object will be a wave packet narrow in both position and momentum, and this narrow wave
packet will follow approximately Newtonian trajectories (if the external potential is uniform
enough along the width of the packet) by Ehrenfest’s theorem (see Bacciagaluppi 2008 for
an insightful analysis in the context of decoherence)46 . In some sense, the emergence of
the classical world around us is “conspired” by environmental particles according to the
energy-conserved collapse model.
Ultimately, the energy-conserved collapse model should be able to account for our defi-
45
It is interesting to note that the state of a macroscopic object can also be localized by the linear
Schrödinger evolution via interactions with environment, e.g. by absorbing an environmental particle with
certain energy uncertainty. For example, if a macroscopic object absorbs a photon (emitted from an atom)
with momentum uncertainty of ∆p ≈ 10−6 eV /c, the center-of-mass state of the object, even if being a mo-
mentum eigenstate initially, will have the same momentum uncertainty by the linear Schrödinger evolution,
and thus it will become a localized wavepacket with width about 0.1m. Note that there is no vicious circle
here. The energy spreading state of a microscopic particle can be generated by an external potential (e.g.
an electromagnetic potential in general) via the linear Schrödinger evolution, and especially they don’t nec-
essarily depend on the localization of macroscopic objects such as measuring devices. Thus we can use the
existence of these states to explain the localization of macroscopic objects.
46
When assuming the energy uncertainty of an object is in the same order of its thermal energy fluctuation,
we can estimate the rough size of its wavepacket. For instance, for a dust particle of mass m = 10−7 g, its
root mean square energy fluctuation is about 103 eV at room temperature T = 300K (Adler 2002), and thus
the width of its wavepacket is about 10−10 m.

71
 4.5. CRITICAL COMMENTS ON OTHER DYNAMICAL COLLAPSE MODELS

nite conscious experience. According to recent neuroscience literature, the appearance of a

(definite) conscious perception in human brains involves a large number of neurons chang-
ing their states from resting state (resting potential) to firing state (action potential). In
each neuron, the main difference of these two states lies in the motion of 106 N a+ s passing
through the neuron membrane. Since the membrane potential is in the order of 10−2 V , the
energy difference between firing state and resting state is ∆E ≈ 104 eV . According to the
energy-conserved collapse model, the collapse time of a quantum superposition of these two
states of a neuron is

~EP 2.8M eV 2
τc ≈ 2
≈( ) ≈ 105 s, (4.22)
(∆E) 0.01M eV
where the Planck energy EP ≈ 1019 GeV . When considering the number of neurons that can
form a definite conscious perception is usually in the order of 107 , the collapse time of the
quantum superposition of two different conscious perceptions will be

2.8M eV 2
τc ≈ ( ) ≈ 10−9 s. (4.23)
100GeV
Since the normal conscious time of a human being is in the order of several hundred mil-
liseconds, the collapse time is much shorter than the normal conscious time. Therefore, our
conscious perceptions are always definite according to the energy-conserved collapse model.

4.5 Critical comments on other dynamical collapse models

In this section, we will give a critical analysis of other dynamical collapse models. These
models can be sorted into two categories. The first one may be called spontaneous collapse
models, in which the dynamical collapse of the wave function is assumed to happen even for
an isolated system. They include the gravity-induced wavefunction collapse model (Diósi
1989; Penrose 1996), the GRW model (Ghirardi, Rimini and Weber 1986)47 etc. The second
category may be called interaction-induced collapse models, which assume that the dynamical
collapse of the wave function of a given system results from its particular interaction with a
noise field. One typical example is the CSL model (Pearle 1989; Ghirardi, Pearle and Rimini
1990)48 . In the following, we will primarily analyze Penrose’s gravity-induced wavefunction
collapse model and the CSL model, which are generally regarded as two of the most promising
models of wavefunction collapse.

4.5.1 Penrose’s gravity-induced wavefunction collapse model

It seems very natural to guess that the collapse of the wave function is induced by gravity.
The reasons include: (1) gravity is the only universal force being present in all physical
interactions; (2) gravitational effects grow with the size of the objects concerned, and it
is in the context of macroscopic objects that linear superpositions may be violated. The
47
The GRW model was originally referred to as QMSL (Quantum Mechanics with Spontaneous Localiza-
tions). In this model, it is assumed that each elementary constituent of any physical system is subjected, at
random times, to random and spontaneous localization processes (or hittings) around appropriate positions.
The random hittings happen much less frequently for a microscopic system, e.g. an electron undergoes a
hitting, on average, every hundred million years. If these hittings are assumed to be brought about by an
external system, then the GRW model should be regarded not as a spontaneous collapse model but as an
interaction-induced collapse model.
48
If the involved noise field in the CSL model is not taken as real, then the model should be regarded as a
spontaneous collapse model.

72
 4.5. CRITICAL COMMENTS ON OTHER DYNAMICAL COLLAPSE MODELS

gravity-induced collapse conjecture can be traced back to Feynman (1995)49 . In his Lectures
on Gravitation, he considered the philosophical problems in quantizing macroscopic objects
and contemplates on a possible breakdown of quantum theory. He said, “I would like to
suggest that it is possible that quantum mechanics fails at large distances and for large
objects, it is not inconsistent with what we do know. If this failure of quantum mechanics is
connected with gravity, we might speculatively expect this to happen for masses such that
GM 2 /~c = 1, of M near 10−5 grams.”
Penrose (1996) further proposed a concrete gravity-induced collapse argument. The ar-
gument is based on a profound and fundamental conflict between the general covariance
principle of general relativity and the superposition principle of quantum mechanics. The
conflict can be clearly seen by considering the superposition state of a static mass distribu-
tion in two different locations, say position A and position B. On the one hand, according
to quantum mechanics, the valid definition of such a superposition requires the existence of
a definite space-time background, in which position A and position B can be distinguished.
On the other hand, according to general relativity, the space-time geometry, including the
distinguishability of position A and position B, cannot be predetermined, and must be dy-
namically determined by the position superposition state. Since the different position states
in the superposition determine different space-time geometries, the space-time geometry
determined by the whole superposition state is indefinite, and as a result, the superposi-
tion and its evolution cannot be consistently defined. In particular, the definition of the
time-translation operator for the superposed space-time geometries involves an inherent ill-
definedness, and this leads to an essential uncertainty in the energy of the superposed state.
Then by analogy Penrose argued that this superposition, like an unstable particle in usual
quantum mechanics, is also unstable, and it will decay or collapse into one of the two states
in the superposition after a finite lifetime. Furthermore, Penrose suggested that the essential
energy uncertainty in the Newtonian limit is proportional to the gravitational self-energy E∆
of the difference between the two mass distributions, and the collapse time, analogous to the
half-life of an unstable particle, is

T ≈ ~/E∆ (4.24)
This criterion is very close to that put forward by Diósi (1989) earlier, and it is usually called
the Diósi-Penrose criterion. Later, Penrose (1998) further suggested that the collapse states
are the stationary solutions of the Schrödinger-Newton equation.
Let’s now analyze Penrose’s argument. The crux of the matter is whether the conflict
between quantum mechanics and general relativity requires that a quantum superposition of
two space-time geometries must collapse after a finite time. We will argue in the following
that the answer is negative. First of all, although it is widely acknowledged that there exists
a fundamental conflict between the general covariance principle of general relativity and the
superposition principle of quantum mechanics, it is still a controversial issue what the exact
nature of the conflict is and how to solve it. For example, it is possible that the conflict
may be solved by reformulating quantum mechanics in a way that does not rely on a definite
spacetime background (see, e.g. Rovelli 2011).
Next, Penrose’s argument seems too weak to establish a necessary connection between
the conflict and wavefunction collapse. Even though there is an essential uncertainty in the
energy of the superposition of different space-time geometries, this kind of energy uncertainty
is different in nature from the energy uncertainty of unstable particles or unstable states
in usual quantum mechanics (Gao 2010). The former results from the ill-definedness of
49
It is interesting to note that Feynman considered this conjecture even earlier at the 1957 Chapel Hill
conference (see DeWitt and Rickles 2011, ch.22).

73
 4.5. CRITICAL COMMENTS ON OTHER DYNAMICAL COLLAPSE MODELS

the time-translation operator for the superposed space-time geometries (though its nature
seems still unclear), while the latter exists in a definite spacetime background, and there is a
well-defined time-translation operator for the unstable states. Moreover, the decay of these
unstable states is a natural result of the linear Schrödinger evolution, and the process is not
random but deterministic. By contrast, the hypothetical spontaneous decay or collapse of
the superposed space-time geometries is nonlinear and random. In addition, the decay of
an unstable state (e.g. excited state of an atom) is actually not spontaneous but caused by
the background field constantly interacting with it. In some extreme situations, the state
may not decay at all when in a very special background field with bandgap (Yablonovitch
1987). In short, there exists no convincing analogy between a superposition of different
space-time geometries and an unstable state in usual quantum mechanics. Accordingly, one
cannot argue for the decay or collapse of the superposition of different space-time geometries
by this analogy. Although an unstable state in quantum mechanics may decay after a
very short time, this does not imply that a superposition of different space-time geometries
should also decay - and, again, sometimes an unstable state does not decay at all under
special circumstances. To sum up, Penrose’s argument by analogy only has a very limited
force, and especially, it is not strong enough to establish a necessary connection between
wavefunction collapse and the conflict between quantum mechanics and general relativity.
Thirdly, it can be further argued that the conflict does not necessarily lead to the wave-
function collapse. The key is to realize that the conflict also needs to be solved before
the wavefunction collapse finishes, and when the conflict has been solved, the wavefunc-
tion collapse will lose its basis relating to the conflict. As argued by Penrose, the quantum
superposition of different space-time geometries and its evolution are both ill-defined due
to the fundamental conflict between the general covariance principle of general relativity
and the superposition principle of quantum mechanics. The ill-definedness seems to require
that the superposition must collapse into one of the definite space-time geometries, which
has no problem of ill-definedness. However, the wavefunction collapse seems too late to
save the superposition from the “suffering” of the ill-definedness during the collapse. In
the final analysis, the conflict or the problem of ill-definedness needs to be solved before
defining a quantum superposition of different space-time geometries and its evolution. In
particular, the possible collapse evolution of the superposition also needs to be consistently
defined, which again indicates that the wavefunction collapse does not solve the problem
of ill-definedness. On the other hand, once the problem of ill-definedness is solved and a
consistent description obtained (however this is still an unsolved issue in quantum gravity),
the wavefunction collapse will completely lose its connection with the problem50 . Therefore,
contrary to Penrose’s expectation, it seems that the conflict between quantum mechanics
and general relativity does not entail the existence of wavefunction collapse.
Even though Penrose’s gravity-induced collapse argument is debatable, the wavefunction
collapse may still exist due to other reasons, and thus Penrose’s concrete suggestions for the
collapse time formula and collapse states also need to be further examined as some aspects
of a phenomenological model. First of all, let’s analyze Penrose’s collapse time formula Eq.
(4.24), according to which the collapse time of a superposition of two mass distributions is
50
Note that if the problem of ill-definedness cannot be solved in principle for the superpositions of very
different space-time geometries, then the wavefunction collapse may be relevant here. Concretely speaking,
if the superpositions of very different space-time geometries cannot be consistently defined in nature, then
it is very likely that these superpositions cannot exist, which means that they must have collapsed into
one of the definite space-time geometries before formed from the superpositions of minutely different space-
time geometries. In this case, the large difference of the space-time geometries in the superposition will
set a upper limit for wavefunction collapse. Though the limit may be loose, it does imply the existence of
wavefunction collapse. However, this possibility might be very small, as it seems that there is always some
kind of approximate sense in which two different spacetimes can be pointwise identified.

74
 4.5. CRITICAL COMMENTS ON OTHER DYNAMICAL COLLAPSE MODELS

inversely proportional to the gravitational self-energy of the difference between the two mass
distributions. As we have argued above, the analogy between such a superposition and an
unstable state in quantum mechanics does not exist, and gravity does not necessarily induce
wavefunction collapse either. Thus this collapse time formula, which is based on a similar
application of Heisenberg’s uncertainty principle to unstable states, will lose its original phys-
ical basis. In particular, the appearance of the gravitational self-energy term in the formula
is in want of a reasonable explanation. In fact, it has already been shown that this gravita-
tional self-energy term does not represent the ill-definedness of time-translation operator (or
the fuzziness of the identification between two spacetimes) in the strictly Newtonian regime
(Christian 2001). In this regime, the time-translation operator can be well defined, but
the gravitational self-energy term is not zero. In addition, as Diósi (2007) pointed out, the
microscopic formulation of the collapse time formula is unclear and still has some problems
(e.g. the cut-off difficulty).
Next, let’s examine Penrose’s suggestion for the collapse states. According to Penrose
(1998), the collapse states are the stationary solutions of the Schrödinger-Newton equation,
namely Eq. (2.31). The equation describes the gravitational self-interaction of a single
quantum system, in which the mass density m|ψ(x, t)|2 is the source of the classical gravi-
tational potential. As we have argued in Chapter 2, although a quantum system has mass
density that is measurable by protective measurement, the density is not real but effective,
and it is formed by the ergodic motion of a localized particle with the total mass of the
system. Therefore, there does not exist a gravitational self-interaction of the mass density.
This conclusion can also be reached by another somewhat different argument. Since charge
always accompanies mass for a charged particle such as an electron51 , the existence of the
gravitational self-interaction, though which is too weak to be excluded by present experi-
ments, may further entail the existence of a remarkable electrostatic self-interaction of the
particle52 , which already contradicts experiments. This analysis poses a serious objection to
the Schrödinger-Newton equation and Penrose’s suggestion for the collapse states53 .
Lastly, we briefly discuss another two problems of Penrose’s collapse scheme. The first
one is the origin of the randomness of collapse results. Penrose did not consider this issue
in his collapse scheme. If the collapse is indeed spontaneous as implied by his gravity-
induced collapse argument, then the randomness cannot result from any external influences
such as an external noise field, and it can only come from the studied quantum system
and its wave function. The second problem is energy non-conservation. Although Penrose
did not give a concrete model of wavefunction collapse, his collapse scheme requires the
collapse of superpositions of different positions, while this kind of space collapse inevitably
violates energy conservation54 . Since the gravitational energy of a quantum system is usually
51
However, the concomitance of mass and charge in space for a charged particle does not necessarily
require that they must satisfy the same law of interaction. For example, the fact that electromagnetic fields
are quantized in nature does not necessarily imply that gravitational fields must be also quantized.
52
If there is a gravitational self-interaction but no electrostatic self-interaction for a charged particle, e.g.
an electron, then the charge and mass of an electron will be located in different positions and have different
density distributions in space, though they are described by the same wave function. Concretely speaking, the
mass density of an electron is me |ψ(x, t)|2 as in the Schrödinger-Newton equation, whereas its charge density
is not e|ψ(x, t)|2 but only localized in a single position (which permits no electrostatic self-interaction). This
result seems very unnatural and has no experimental support either.
53
Since the Schrödinger-Newton equation is the non-relativistic realization of the typical model of semi-
classical gravity, in which the source term in the classical Einstein equation is taken as the expectation of
the energy momentum operator in the quantum state (Rosenfeld 1963), our analysis also presents a serious
objection to the approach of semiclassical gravity. Although the existing arguments against the semiclassical
gravity models seem so strong, they are still not conclusive (Carlip 2008; Boughn 2009). This new analysis
of the Schrödinger-Newton equation may shed some new light on the solution of the issue.
54
Diósi (2007) explicitly pointed out that the von-Neumann-Newton equation, which may be regarded as one
realization of Penrose’s collapse scheme, obviously violates conservation of energy. Another way to understand

75
 4.5. CRITICAL COMMENTS ON OTHER DYNAMICAL COLLAPSE MODELS

much smaller than the energy of the system, Penrose’s collapse scheme still violates energy
conservation even if the gravitational field is counted. As we have noted earlier, for an isolated
system only the collapse states are energy eigenstates can energy conserve (at the ensemble
level) during the collapse. If the principle of conservation of energy is indeed universal as
widely thought, then the spontaneous collapse models that violate energy conservation will
have been excluded. By contrast, although the interaction-induced collapse models such as
the CSL model also violate energy conservation in their present formulations, there is still
hope that when counting the energy of external noise field the total energy may be conserved
in these models (Pearle 2000; Bassi, Ippoliti and Vacchini 2005). Let’s turn to the CSL model
now.

4.5.2 The CSL model

In the CSL model, the collapse of the wave function of a quantum system is assumed to be
caused by its interaction with a classical scalar field, w(x, t). The collapse states are the
eigenstates of the smeared mass density operator, and the mechanism leading to the sup-
pression of the superpositions of macroscopically different states is fundamentally governed
by the integral of the squared differences of the mass densities associated to the superposed
states. It may be expected that the introduction of the noise field can help to solve the
problems plagued by the spontaneous collapse models, e.g. the problems of energy non-
conservation and the origin of randomness etc. However, one must first answer what field
the noise field is and especially why it can collapse the wave functions of all quantum sys-
tems. The validity of the CSL model strongly depends on the existence of this hypothetical
noise field. In the following, we will mainly analyze this important legitimization problem of
the CSL model55 .
Whatever the nature of the noise field w(x, t) is, it cannot be quantum in the usual sense
since its coupling to a quantum system is not a standard coupling between two quantum
systems. The coupling is anti-Hermitian (Bassi 2007), and the equation of the resulting
dynamical collapse is not the standard Schrödinger equation with a stochastic potential
either. According to our current understandings, the gravitational field is the only universal
field that might be not quantized, though this possibility seems extremely small in the view of
most researchers. Therefore, it seems natural to identify this noise field with the gravitational
field. In fact, it has been argued that in the CSL model the w-field energy density must have
a gravitational interaction with ordinary matter (Pearle and Squires 1996; Pearle 2009). The
this conclusion is to realize that the energy-conserved wavefunction collapse cannot result from the spacetime
geometry difference between the branches in a superposition as suggested by Penrose’s collapse scheme. The
reason is that there is no difference of spacetime geometries for two different momentum eigenstates. A
momentum eigenstate does not influence its background spacetime geometry, as its energy density is zero
throughout the whole space. Thus if a superposition of two momentum eigenstates does collapse into one of
them, the collapse cannot result from the difference of spacetime geometries in the superposition. As a result,
Penrose’s gravity-induced collapse argument does not lead to the energy-conserved wavefunction collapse,
and if it does lead to some sort of wavefunction collapse, the collapse cannot conserve energy.
55
As admitted by Pearle (2009), “When, over 35 years ago, ... I had the idea of introducing a randomly
fluctuating quantity to cause wave function collapse, I thought, because there are so many things in nature
which fluctuate randomly, that when the theory is better developed, it would become clear what thing in
nature to identify with that randomly fluctuating quantity. Perhaps ironically, this problem of legitimizing
the phenomenological CSL collapse description by tying it in a natural way to established physics remains
almost untouched.” Related to this legitimization problem is that the two parameters which specify the model
are ad hoc (Pearle 2007). These two parameters, which were originally introduced by Ghirardi, Rimini and
Weber (1986), are a distance scale, a ≈ 105 cm, characterising the distance beyond which the collapse becomes
effective, and a time scale, λ−1 ≈ 1016 sec, giving the rate of collapse for a microscopic system. If wavefunction
collapse is a fundamental physical process related to other fundamental processes, the parameters should be
able to be written in terms of other physical constants.

76
 4.5. CRITICAL COMMENTS ON OTHER DYNAMICAL COLLAPSE MODELS

argument of Pearle and Squires (1996) can be summarized as follows56 .

There are two equations which characterize the CSL model. The first equation is a
modified Schrödinger equation, which expresses the influence of an arbitrary field w(x, t) on
the quantum system. The second equation is a probability rule which gives the probability
that nature actually chooses a particular w(x, t). This probability rule can also be interpreted
as expressing the influence of the quantum system on the field. As a result, w(x, t) can be
written as follows:

w(x, t) = w0 (x, t) + hA(x, t)i, (4.25)

where A(x, t) is the mass density operator smeared over the GRW scale a, hA(x, t)i is its
quantum expectation value, and w0 (x, t) is a Gaussian randomly fluctuating field with zero
drift, temporally white noise in character and with a particular spatial correlation function.
Then the scalar field w(x, t) that causes collapse can be interpreted as the gravitational
curvature scalar with two sources, the expectation value of the smeared mass density operator
and an independent white noise fluctuating source. This indicates that the CSL model
is based on the semi-classical gravity, and the smeared mass density is the source of the
gravitational potential. Note that the reality of the field w(x, t) requires that the smeared
mass density of a quantum system is real57 .
According to our previous analysis in Chapter 2, however, a quantum system does not
have a real mass density distribution in space, no matter it is smeared or not. Moreover,
although the approach of semi-classical gravity may be consistent in the context of dynamical
collapse models (Pearle and Squires 1996; Ghirardi 2008), it may have been excluded as we
have argued in the last subsection. Besides, protective measurement shows that the effective
mass density of a quantum system is proportional to the modulus square of its wave function.
Thus the assumed existence of the smeared mass density in the CSL model, even if it is
effective, also contradicts protective measurement. Note that it is crucial that the mass
density be smeared over the GRW scale a in the CSL model; without such a smearing the
energy excitation of particles undergoing collapse would be beyond experimental constraints
(Pearle and Squires 1996). In conclusion, it seems that the noise field introduced in the CSL
model cannot have a gravitational origin as required by the model, and this may raise strong
doubts about the reality of the field.
On the other hand, even though the approach of semi-classical gravity is viable and the
noise field in the CSL model can be the gravitational field, one still need to answer why the
gravitational field has the very ability to collapse the wave functions of all quantum systems
as required by the model. It is worth noting that the randomly fluctuating field in the model,
w0 (x, t), is not the gravitational field of the studied quantum system but the background
gravitational field. Thus Penrose’s gravity-induced wavefunction collapse argument, even if
valid, does not apply to the CSL model, which is essentially an interaction-induced model
of wavefunction collapse. The fluctuations of the background gravitational field can readily
56
Pearle (2009) further argued that compatibility with general relativity requires a gravitational force
exerted upon matter by the w-field. However, as Pearle (2009) admitted, no convincing connection (for
example, identification of metric fluctuations, dark matter or dark energy with w(x, t)) has yet emerged, and
the legitimization problem (i.e. the problem of endowing physical reality to the noise field) is still in its
infancy.
57
In fact, Ghirardi, Grassi and Benatti (1995) and Ghirardi (1997) already explicitly proposed the so-called
mass density ontology in the context of dynamical collapse theories. According to Ghirardi (2008), “what
the theory is about, what is real ‘out there’ at a given space point x, is just a field, i.e. a variable m(x, t)
given by the expectation value of the mass density operator M (x) at x obtained by multiplying the mass of
any kind of particle times the number density operator for the considered type of particle and summing over
all possible types of particles.”

77
 4.5. CRITICAL COMMENTS ON OTHER DYNAMICAL COLLAPSE MODELS

lead to the decoherence of the wave function of a quantum system, but it seems that they
have no ability to cause the collapse of the wave function.
Lastly, we will briefly discuss another two problems of the CSL model. The first one
is the well-known problem of energy non-conservation. The collapse in the model narrows
the wave function in position space, thereby producing an increase of energy58 . A possible
solution is that the conservation laws may be satisfied when the contributions of the noise
field w(x, t) to the conserved quantities are taken into account. It has been shown that the
total mean energy can be conserved (Pearle 2004), and the energy increase can also be made
finite when further revising the coupling between the noise field and the studied quantum
system (Bassi, Ippoliti and Vacchini 2005). But a complete solution has not been found yet,
and it is still unknown whether such a solution indeed exists. The second problem is to make
a relativistic quantum field theory which describes collapse (Pearle 2009). Notwithstanding
a good deal of effort, a satisfactory theory has not been obtained at present (see Beding-
ham 2011 for a recent attempt). The main difficulty is that the hypothetical interaction
responsible for collapse will produce too many particles out of the vacuum, amounting to
infinite energy per sec per volume, in the relativistic extension of these interaction-induced
collapse models. Note that the spontaneous collapse models without collapse interaction
(e.g. the energy-conserved collapse model) don’t face this difficulty. We will discuss the
problem of compatibility between wavefunction collapse and the principle of relativity in the
next Chapter.

58
Note that with appropriate choice for the parameters in the CSL model, such a violation of energy
conservation is very tiny and hardly detectable by present day technology.

78
 5
We have an apparent incompatibility, at the deepest level, between the two
fundamental pillars of contemporary theory ... It may be that a real synthesis
of quantum and relativity theories requires not just technical developments but
radical conceptual renewal.
— John Bell, 1986

Random Discontinuous Motion and Special

Relativity

In this chapter, we will briefly analyze random discontinuous motion of particles and its col-
lapse evolution in the relativistic domain1 . It is first shown that the Lorentz transformation
seriously distorts the picture of random discontinuous motion of particles, and the distortion
results from the relativity of simultaneity. We then argue that absolute simultaneity is not
only possible in the relativistic domain, but also necessitated by the existence of random
discontinuous motion of particles and its collapse evolution. This leads to the emergence of
a preferred Lorentz frame when combined with the requirement of the constancy of speed
of light. It is further shown that the collapse dynamics may provide a way to detect the
frame according to the energy-conserved collapse model. If quantum mechanics indeed de-
scribes random discontinuous motion of particles as protective measurement suggests, then
this analysis may be helpful for solving the problem of the incompatibility between quantum
mechanics and special relativity2 .

5.1 The picture of motion distorted by the Lorentz transformation

Let’s first see how the picture of random discontinuous motion is distorted by the Lorentz
transformation.

5.1.1 Single particle picture

For the random discontinuous motion of a particle, the particle has a tendency to be in any
possible position at a given instant, and the probability density of the particle appearing in
each position x at a given instant t is determined by the modulus square of its wave function,
namely ρ(x, t) = |ψ(x, t)|2 . The physical picture of the motion of the particle is as follows. At
a discrete instant the particle randomly stays in a position, and at the next instant it will still
stay there or randomly appear in another position, which is probably not in the neighborhood
of the previous position. In this way, during a time interval much larger than the duration of
one instant, the particle will move discontinuously throughout the whole space with position
probability density ρ(x, t). Since the distance between the locations occupied by the particle
at two neighboring instants may be very large, this jumping process is obviously nonlocal.
In the non-relativistic domain where time is absolute, the nonlocal jumping process is the
same in every inertial frame. But in the relativistic domain, the jumping process will look
1
Our analysis is in the low-energy regime and does not consider the high-energy processes described by
relativistic quantum field theory, e.g. the creation and annihilation of particles.
2
There is no consensus among contemporary philosophers and physicists concerning the solution to this
incompatibility problem. For a comprehensive discussion of this issue see Maudlin (2002) and references
therein.

79
 5.1. THE PICTURE OF MOTION DISTORTED BY THE LORENTZ
TRANSFORMATION

different in different inertial frames due to the Lorentz transformation. Let’s give a concrete
analysis.
Suppose a particle is in position x1 at instant t1 and in position x2 at instant t2 in an
inertial frame S. In another inertial frame S 0 with velocity v relative to S, the Lorentz
transformation leads to:

0 t1 − x1 v/c2
t1 = p , (5.1)
1 − v 2 /c2
0 t2 − x2 v/c2
t2 = p , (5.2)
1 − v 2 /c2
0 x1 − vt1
x1 = p , (5.3)
1 − v 2 /c2
0 x2 − vt2
x2 = p . (5.4)
1 − v 2 /c2
Since the jumping process of the particle is nonlocal, the two events (t1 , x1 ) and (t2 , x2 )
may readily satisfy the spacelike separation condition |x2 − x1 | > c|t2 − t1 |. Then we can
0 0
always select a possible velocity v < c that leads to t2 = t1 :
t2 − t1 2
v= c . (5.5)
x2 − x1
0 0
But obviously the two positions of the particle in frame S 0 , namely x1 and x2 , are not equal.
0 0
This means that in frame S 0 the particle will be in two different positions x1 and x2 at the
0
same time at instant t1 . In other words, it seems that there are two identical particles at
0
instant t1 in frame S 0 . Note that the velocity of S 0 relative to S may be much smaller than
the speed of light, and thus the appearance of the two-particle picture is irrelevant to the
high-energy processes described by relativistic quantum field theory, e.g. the creation and
annihilation of particles.
The above result shows that for any pair of events in frame S that satisfies the spacelike
separation condition, there always exists an inertial frame in which the two-particle picture
will appear. Since the jumping process of the particle in frame S is essentially random, it can
be expected that the two-particle picture will appear in the infinitely many inertial frames
with the same probability. Then during an arbitrary finite time interval, in each inertial
frame the measure of the instants at which there are two particles in appearance, which is
equal to the finite time interval divided by the total number of the frames that is infinite,
will be zero. Moreover, there may also exist situations where the particle is at arbitrarily
many positions at the same time at an instant in an inertial frame, though the measure
of these situations is also zero. Certainly, at nearly all instants whose measure is one, the
particle is still in one position at an instant in all inertial frames. Therefore, the many-
particle appearance of the random discontinuous motion of a particle cannot be measured
in principle.
However, for the random discontinuous motion of a particle, in any inertial frame different
from S, the Lorentz transformation will inevitably make the time order of the random stays
of the particle in S reversal and disorder, as the discontinuous motion of the particle is
nonlocal and most neighboring random stays are spacelike separated events. In other words,
the time order is not Lorentz invariant. Moreover, the set of the instants at which the time
order of the random stays of the particle is reversed has finite measure, which may be close
to one. As we will see below, this reversal and disorder of time order will lead to more
distorted pictures for quantum entanglement and wavefunction collapse.

80
 5.1. THE PICTURE OF MOTION DISTORTED BY THE LORENTZ
TRANSFORMATION

5.1.2 Picture of quantum entanglement

Now let’s analyze the motion of two particles in quantum entanglement. For the random
discontinuous motion of two particles in an entangled state, the two particles have a joint
tendency to be in any two possible positions, and the probability density of the two particles
appearing in each position pair x1 and x2 at a given instant t is determined by the modulus
square of their wave function at the instant, namely ρ(x1 , x2 , t) = |ψ(x1 , x2 , t)|2 .
Suppose two particles are in an entangled state ψu ϕu + ψd ϕd , where ψu and ψd are
two spatially separated states of particle 1, ϕu and ϕd are two spatially separated states of
particle 2, and particle 1 and particle 2 are also separated in space. The physical picture of
this entangled state is as follows. Particles 1 and 2 are randomly in the state ψu ϕu or ψd ϕd
at an instant, and then they will still stay in this state or jump to the other state at the
next instant. During a very short time interval, the two particles will discontinuously move
throughout the states ψu ϕu and ψd ϕd with the same probability 1/2. In this way, the two
particles form an inseparable whole, and they jump in a precisely simultaneous way. At an
arbitrary instant, if particle 1 is in the state ψu or ψd , then particle 2 must be in the state
ϕu or ϕd , and vice versa. Moreover, when particle 1 jumps from ψu to ψd or from ψd to ψu ,
particle 2 must simultaneously jump from ϕu to ϕd or from ϕd to ϕu , and vice versa. Note
that this kind of random synchronicity between the motions of particle 1 and the motion of
particle 2 is irrelevant to the distance between them, and it can only be explained by the
existence of joint tendency of the two particles as a whole.
The above picture of quantum entanglement is assumed to exist in one inertial frame. It
can be expected that when observed in another inertial frame, this perfect picture will be
distorted in a similar way as for the single particle case. Let’s give a concrete analysis below.
Suppose in an inertial frame S, at instant ta particle 1 is at position x1a and in state ψu and
particle 2 at position x2a and in state ϕu , and at instant tb particle 1 is at position x1b and
in state ψd and particle 2 at position x2b and in state ϕd . Then according to the Lorentz
transformation, in another inertial frame S 0 with velocity v 0 relative to S, where v 0 satisfies:
ta − tb 2
v0 = c , (5.6)
x1a − x2b
the instant at which particle 1 is at position x01a and in state ψu is the same as the instant
at which particle 2 is at position x02b and in state ϕd , namely
1 x1a tb − x2b ta
t01a = t02b = p · . (5.7)
1 − v 02 /c2 x1a − x2b
This means that in S 0 there exists an instant at which particle 1 is in state ψu but particle
2 is in state ϕd . Similarly, in another inertial frame S 00 with velocity v 00 relative to S, there
also exists an instant t00 at which particle 1 is in state ψd but particle 2 is in state ϕu , where
v 00 and t00 satisfy the following relations:
ta − tb 2
v 00 = c , (5.8)
x2a − x1b
1 x2a tb − x1b ta
t00 = p · . (5.9)
1 − v 002 /c2 x2a − x1b
Note that since the two particles are well separated in space, the above two velocities can
readily satisfy the restricting conditions v 0 < c and v 00 < c when the time interval |ta − tb | is
very short.
In fact, since the two particles in the above entangled state are separated in space and
their motion is essentially random, in any inertial frame different from S, the instantaneous

81
 5.1. THE PICTURE OF MOTION DISTORTED BY THE LORENTZ
TRANSFORMATION

correlation between the motion of the two particles in S can only keep half the time, and
the correlation will be reversed for another half of time, during which the two particles
√ will
√
be in state ψu ϕd or ψd ϕu at each instant. For a general entangled state aψu ϕu + bψd ϕd ,
the proportion of correlation-reversed time will be 2ab, and the proportion of correlation-
kept time will be a2 + b2 . Moreover, the instants at which the original correlation is kept
or reversed are discontinuous and random. This means that the synchronicity between the
jumps of the two particles is destroyed too.
To sum up, the above analysis indicates that the instantaneous correlation and syn-
chronicity between the motion of two entangled particles in one inertial frame is destroyed
in other frames due to the Lorentz transformation3 . As we will see below, however, this
distorted picture of quantum entanglement cannot be measured either.

5.1.3 Picture of wavefunction collapse

We have shown that the picture of the instantaneous motion of particles is distorted by the
Lorentz transformation due to the nonlocality and randomness of motion. In the following,
we will further show that the nonlocal and random collapse evolution of the state of motion
(defined during an infinitesimal time interval) will be influenced more seriously by the Lorentz
transformation.
Consider a particle being in a superposition of two Gaussian wavepackets √12 ψ1 + √12 ψ2 in
an inertial frame S. The centers of the two wavepackets are located in x1 and x2 (x1 < x2 ),
respectively, and the width of each wavepacket is much smaller than the distance between
them. After being measured, this superposition state will randomly collapse to ψ1 or ψ2 with
the same probability 1/2. Suppose the collapse happens at different locations at the same
time in frame S. This means that when the superposition state collapses to the branch ψ1
near position x1 , the other branch ψ2 near position x2 will disappear simultaneously. The
simultaneity of wavefunction collapse ensures that the sum of the probabilities of the particle
being in all branches is 1 at every instant.
According to the picture of random discontinuous motion of particles, the above collapse
process can be described as follows. Before the collapse of the superposition state √12 ψ1 +
√1 ψ2 , the particle jumps between the two branches ψ1 and ψ2 or the two regions near x1
2
and x2 in a discontinuous and random way4 . At each instant, the particle is either in a
position near x1 or in a position near x2 , and its probability of being in each region is the
same 1/2. This means that at every instant there is always one particle, which spends half
the time near x1 and half the time near x2 . After the superposition state collapses to one
of its branches, e.g. ψ1 , the particle only jumps in the region near x1 in a discontinuous
and random way, and its probability of being in this region is 1. This means that at every
instant there is always one particle in a position inside the region.
Now let’s see the picture of the above collapse process in another inertial frame S 0 with
velocity v relative to S. Suppose the superposition state √12 ψ1 + √12 ψ2 collapses to the branch
ψ1 near position x1 at instant t in frame S. This process contains two events happening
simultaneously in two spatially separated regions. One event is the disappearance of the
branch √12 ψ2 near position x2 at instant t, and the other is the change from √12 ψ1 to ψ1
happening near position x1 at instant t5 . According to the Lorentz transformation, the times
3
Certainly, in these frames there are still correlations and synchronicity between the jumps of the two
particles at different instants. As noted above, however, these instants are discontinuous and random, and
thus the correlation and synchronicity can hardly be identified.
4
In other words, each branch exists in a set of discontinuous and random instants, whose measure is 1/2,
and the two instant sets constitute the whole continuous time flow.
5
Strictly speaking, since the collapse time is always finite, these events happen not at a precise instant but

82
 5.1. THE PICTURE OF MOTION DISTORTED BY THE LORENTZ
TRANSFORMATION

of occurrence of these two events in S 0 are

t − x1 v/c2
t01 = p , (5.10)
1 − v 2 /c2
t − x2 v/c2
t02 = p . (5.11)
1 − v 2 /c2
It can be seen that x1 < x2 leads to t01 > t02 . Then during the period between t01 and t02 ,
the branch √12 ψ20 near position x02 already disappeared, but the branch √12 ψ10 near position
x01 has not changed to ψ10 . This means that at any instant between t01 and t02 , there is only
a non-normalized state √12 ψ10 . According to the picture of random discontinuous motion
of particles, for a particle in the state √12 ψ10 , the probability of the particle being in the
branch ψ10 is 1/2, and the particle is in the branch ψ10 or in the region near x1 only at some
discontinuous and random instants, whose total measure is 1/2. At other instants, whose
measure is also 1/2, the particle does not exist anywhere. In other words, at each instant the
particle either exists in a position near x1 or disappears in the whole space with the same
probability, 1/2. This result indicates that in the inertial frame S 0 , the particle only exists
half the time during the period between t01 and t02 . By contrast, the particle always exists in
certain position in space at any time in the inertial frame S.
Similarly, if the superposition state √12 ψ1 + √12 ψ2 collapses to the branch ψ2 near position
x2 at instant t in frame S, then in frame S 0 , during the period between t01 and t02 , the branch
√1 ψ 0 near position x0 already turns to ψ 0 , while the branch √1 ψ 0 near position x0 has not
2 2 2 2 2 1 1
disappeared and is still there. Therefore, there is only a non-normalized state √12 ψ10 + ψ20
at any instant between t01 and t02 . According to the picture of random discontinuous motion
of particles, this means that during the period between t01 and t02 , there is more than one
particle in S 0 : the first particle is in the branch ψ20 all the time, and the second identical
particle exists half the time in the branch ψ10 (and it exists nowhere in space for another half
of time).
However, although the state of the particle in S 0 is not normalized, the total probability
of detecting the particle in the whole space is still 1, not 1/2 or 3/2, in the frame6 . In
other words, although the collapse process is seriously distorted in S 0 , the distortion cannot
be measured. The reason is that in S 0 the collapse resulting from measurement happens
at different instants in different locations7 , and the superposition of the branches in these
locations and at these instants are always normalized. In the following, we will give a more
detailed explanation.
As noted above, in frame S 0 the collapse first happens at t02 for the branch √12 ψ20 near
position x02 , and then happens at t01 for the branch √1 ψ 0
2 1
near position x01 after a delay. If
we measure the branch √1 ψ 0 , then the resulting collapse will influence the other branch
2 2
|x1 −x2 |v/c2
√ ψ only after a delay of ∆t0 = √
1 0 , while if we measure the branch √12 ψ10 , then the
2 1 1−v 2 /c2
resulting collapse will influence the other branch √12 ψ20 in advance by the same time interval
∆t0 , and the influence is backward in time. Now suppose we make a measurement on the
branch √1 ψ 0 near position x02 and detect the particle there (i.e. the collapse state is ψ20 ).
2 2

during a very short time, which may be much shorter than the time of light propagating between x1 and x2 .
6
This does not contradict the usual Born rule, which only applies to the situations where collapse happens
simultaneously at different locations in space.
7
Concretely speaking, the time order of the collapses happening at different locations in S 0 is connected
with that in S by the Lorentz transformation.

83
 5.2. ON THE ABSOLUTENESS OF SIMULTANEITY

Then before the other branch √12 ψ10 disappears, which happens after a delay of ∆t0 , we can
make a second measurement on this branch near position x01 . It seems that the probability
of detecting the particle there is not zero but 1/2, and thus the total probability of finding
the particle in the whole space is larger than one and it is possible that we can detect two
particles. However, this is not the case. Although the second measurement on the branch
√1 ψ 0 near position x0 is made later than the first measurement, it is the second measurement
2 1 1
that collapses the superposition state √1 ψ 0 + √1 ψ 0 to ψ20 near position x02 ; the local branch
2 1 2 2
√1 ψ 0 near position x01 disappears immediately after the measurement, while the influence
2 1
of the resulting collapse on the other branch √12 ψ20 near position x02 is backward in time and
happens before the first measurement on this branch. Therefore, the second measurement
near position x01 must obtain a null result, and why the first measurement detects the particle
near position x02 is because the superposition state already collapses to ψ20 near position x02
before the measurement due to the second measurement.
By a similar analysis, we can also demonstrate that the measurements on an entangled
state of two particles, e.g. ψu ϕu + ψd ϕd , can only obtain correlated results in every inertial
frame. If a measurement on particle 1 obtains the result u or d, indicating the state of the
particle collapses to the state ψu or ψd after the measurement, then a second measurement on
particle 2 can only obtain the result u or d, indicating the state of particle 2 collapses to the
state ϕu or ϕd after the measurement. Accordingly, although the instantaneous correlation
and synchronicity between the motion of two entangled particles is destroyed in all but one
inertial frame, the distorted picture of quantum entanglement cannot be measured.

5.2 On the absoluteness of simultaneity

The above analysis clearly demonstrates the apparent conflict between the random dis-
continuous motion of particles and the Lorentz transformation in special relativity. The crux
of the matter lies in the relativity of simultaneity. If simultaneity is relative as manifested by
the Lorentz transformation, then the picture of random discontinuous motion of particles will
be seriously distorted except in one preferred frame, though the distortion is unobservable
in principle. Only when simultaneity is absolute, can the picture of random discontinuous
motion of particles be kept perfect in every inertial frame. In the following, we will show that
absolute simultaneity is not only possible, but also necessitated by the existence of random
discontinuous motion of particles and its collapse evolution.
Although the relativity of simultaneity has been often regarded as one of the essential
concepts of special relativity, it is not necessitated by experimental facts but a result of the
choice of standard synchrony (see, e.g. Reichenbach 1958; Grünbaum 1973)8 . As Einstein
(1905) already pointed out in his first paper on special relativity, whether or not two spatially
separated events are simultaneous depends on the adoption of a convention in the framework
of special relativity. In particular, the choice of standard synchrony, which is based on the
constancy of one-way speed of light and results in the relativity of simultaneity, is only a
convenient convention. Strictly speaking, the speed constant c in special relativity is two-way
speed, not one-way speed, and as a result, the general spacetime transformation required by
the constancy of two-way speed of light is not the Lorentz transformation but the Edwards-
Winnie transformation (Edwards 1963; Winnie 1970):

x0 = η(x − vt), (5.12)

8
For more discussions about this issue see Janis (2010) and references therein.

84
 5.2. ON THE ABSOLUTENESS OF SIMULTANEITY

t0 = η[1 + β(k + k 0 )]t + η[β(k 2 − 1) + k − k 0 ]x/c, (5.13)

where x, t and x0 , t0 are the coordinates of inertial frames S and S0,
respectively, v is the
velocity 0
of S relative to S, c is the invariant two-way speed of light, β = v/c, and η =
1/ (1 + βk)2 − β 2 . k and k 0 represent the directionality of one-way speed of light in S and
p

S 0 , respectively, and they satisfy −1 6 k, k 0 6 1. Concretely speaking, the one-way speeds

c c
of light along x and −x directions in S are cx = 1−k and c−x = 1+k , respectively, and the
0 0 0
one-way speeds of light along x and −x directions in S are cx0 = 1−k c c
0 and c−x0 = 1+k 0 ,

respectively.
If adopting the standard synchrony convention, namely assuming the one-way speed of
light is isotropic and constant in every inertial frame, then k, k 0 = 0 and the Edwards-Winnie
transformation will reduce to the Lorentz transformation, which leads to the relativity of
simultaneity. Alternatively, one can also adopt the nonstandard synchrony convention that
makes simultaneity absolute. In order to do this, one may first synchronize the clocks at
different locations in an arbitrary inertial frame by Einstein’s standard synchrony, that is,
one assumes the one-way speed of light is isotropic in this frame, and then let the clocks in
other frames directly regulated by the clocks in this frame when they coincide in space. The
corresponding spacetime transformation can be derived as follows. Let S be the preferred
Lorentz frame in which the one-way speed of light is isotropic, namely let k = 0. Then we
get

k 0 = β(k 2 − 1) + k = −β. (5.14)

Besides, since the synchrony convention leads to the absoluteness of simultaneity, we also
have in the Edwards-Winnie transformation:

β(k 2 − 1) + k − k 0 = 0. (5.15)
Thus the spacetime transformation that restores absolute simultaneity is:
1
x0 = p · (x − vt), (5.16)
1 − v 2 /c2

t0 = 1 − v 2 /c2 · t.
p
(5.17)
where x, t are the coordinates of the preferred Lorentz frame, x0 , t0 are the coordinates of
another inertial frame, and v is the velocity of this frame relative to the preferred frame.
c2
In this frame, the one-way speed of light along x0 and −x0 direction are cx0 = c−v and
2
c
c−x0 = c+v , respectively.
The above analysis demonstrates the possibility of keeping simultaneity absolute within
the framework of special relativity. One can adopts the standard synchrony that leads to the
relativity of simultaneity, and one can also adopts the nonstandard synchrony that restores
the absoluteness of simultaneity. This is permitted because there is no causal connection
between two spacelike separated events in special relativity. However, if there is a causal
influence connecting two distinct events, then the claim that they are not simultaneous
will have a nonconventional basis (Reichenbach 1958; Grünbaum 1973; Janis 2010). In
particular, if there is an arbitrarily fast causal influence connecting two spacelike separated
events, then these two events will be simultaneous. In the following, we will show that
random discontinuous motion and its collapse evolution just provide a nonconventional basis
for the absoluteness of simultaneity.
Consider a particle being in a superposition of two well separated spatial branches. Ac-
cording to the picture of random discontinuous motion, the particle jumps between these two

85
 5.3. COLLAPSE DYNAMICS AND PREFERRED LORENTZ FRAME

branches in a random and discontinuous way. At an instant the particle is in one branch, and
at the next instant it may be in the other spatially-separated branch. The disappearance
of the particle in the first branch can be regarded as one event, and the appearance of the
particle in the second branch can be regarded as another event. Obviously there is an instan-
taneous causal connection between these two spacelike separated events; if the particle did
not disappear in the first branch, it could not appear in the second branch. Therefore, these
two events should be regarded as simultaneous. Note that this conclusion is irrelevant to
whether the two events and their causal connection are observable. Besides, the conclusion
is also irrelevant to the reference frame, which further means that simultaneity is absolute9 .
Let’s further consider the collapse evolution of random discontinuous motion during a
measurement. It can be seen that the measurement on one branch of the superposition has
a causal influence on the other branch (as well as on the measured branch) via the collapse
process, and this nonlocal influence is irrelevant to the distance between the two branches.
Accordingly, the time order of the measurement and the collapse of the superposition hap-
pening in the two separated regions cannot be conventional but must be unique. Since the
collapse time can be arbitrarily short, the measurement and the collapse of the superposition
can be regarded as simultaneous. Moreover, the collapses of the superposition in the two
regions, which are spacelike separated events, are also simultaneous10 . The simultaneity is
irrelevant to the selection of inertial frames, which again means that simultaneity is absolute.
Certainly, the collapse of an individual superposition cannot be measured within the
framework of the existing quantum mechanics. However, on the one hand, the above con-
clusion is irrelevant to whether the collapse events can be measured or not, and on the other
hand, the collapse of an individual superposition may be observable when the quantum dy-
namics is deterministic nonlinear (Gisin 1990), e.g. when the measuring device is replaced
with a conscious observer (Squires 1992; Gao 2004).

5.3 Collapse dynamics and preferred Lorentz frame

The random discontinuous motion of particles and its collapse evolution requires that
simultaneity is absolute. If the collapse of the wave function happens simultaneously at
different locations in space in every inertial frame, then the one-way speed of light will be
not isotropic in all but one inertial frame. In other words, if the absolute simultaneity is
restored, then the non-invariance of the one-way speed of light will single out a preferred
Lorentz frame, in which the one-way speed of light is isotropic11 . The detectability of this
frame seems to depend on the measurability of individual collapse. Once the collapse of
an individual wave function can be measured, the clocks at different locations in space can
be synchronized with the help of the instantaneous wavefunction collapse in every inertial
frame, and the preferred Lorentz frame can then be determined by measuring the one-way
9
Why does each instantaneous jump of a particle in one inertial frame last much long time in another
inertial frame? The lapse of time cannot be explained in physics, and it can only result from the inappropriate
synchrony of clocks at different locations in the later frame.
10
Note that there exists no causal influence between these two events, and they both result from the
measurement of the local measuring device, which is the common cause.
11
Similarly, if the invariance of the one-way speed of light or standard synchrony is assumed as by the
Lorentz transformation, then the collapse evolution of random discontinuous motion will also single out a
preferred Lorentz frame, in which the collapse of the wave function happens simultaneously at different
locations in space, whether the frame can be actually determined. In the final analysis, the emergence of
a preferred Lorentz frame is the inevitable result of the combination of the constancy of two-way speed of
light and the existence of random discontinuous motion and its collapse evolution. Thus, no matter which
assumption is adopted, the preferred Lorentz frame can always be defined as the inertial frame in which the
one-way speed of light is isotropic and the collapse of the wave function happens simultaneously in the whole
space.

86
 5.3. COLLAPSE DYNAMICS AND PREFERRED LORENTZ FRAME

speed of light, which is isotropic in the frame.

However, even if the collapse of an individual wave function cannot be measured, the
preferred Lorentz frame may also be determined by measuring the (average) collapse time
of the wave functions of identical systems in an ensemble according to our energy-conserved
collapse model12 . The reason is that the collapse dynamics, like the time order of the collapses
happening in different positions, is not relativistically invariant in our model. Let’s give a
more detailed analysis below.
According to the energy-conserved collapse model, the (average) collapse time formula
for an energy superposition state, denoted by Eq. (4.15), can be rewritten as

~2
τc ≈ , (5.18)
tP (∆E)2
where tP is the Planck time, ∆E is the energy uncertainty of the state. We assume this
collapse time formula is still valid in an inertial frame in the relativistic domain. This
assumption seems reasonable, as the collapse time formula already contains the speed of light
c via the Planck time tp 13 . Since the formula is not relativistically invariant, its relativistically
invariant form must contain a term relating to the velocity of the experimental frame relative
to a preferred Lorentz frame. In other words, there must exist a preferred Lorentz frame
according to the collapse model. We define the preferred Lorentz frame, denoted by S0 ,
as the inertial frame where the above formula is valid. Then in another inertial frame the
collapse time will depend on the velocity of the frame relative to S0 . According to the
Lorentz transformation14 , in an inertial frame S 0 with velocity v relative to the frame S0 we
have:
1
τc0 = p · τc , (5.19)
1 − v 2 /c2
1
t0P = p · tP , (5.20)
1 − v 2 /c2
1 − v/c
∆E 0 ≈ p · ∆E. (5.21)
1 − v 2 /c2
Here we only consider the situation where the particle has very high energy, namely E ≈ pc,
and thus Eq. (5.21) holds. Besides, we assume the Planck time tP is the minimum time in
the preferred Lorentz frame, and in another frame the minimum time (i.e. the duration of a
discrete instant) is connected with the Planck time tP by the time dilation formula required
by special relativity. Then by inputting these equations into Eq. (5.22), we can obtain the
relativistic collapse time formula for an arbitrary experimental frame with velocity v relative
to the frame S0 :

~2
τc ≈ (1 + v/c)−2 . (5.22)
tP (∆E)2
This formula contains a term relating to the velocity of the experimental frame relative to the
preferred Lorentz frame. It can be expected that this velocity-dependent term originates from
12
Although it has been argued that quantum nonlocality and special relativity are incompatible, and a
consistent description of wavefunction collapse demands the existence of a preferred Lorentz frame (see, e.g.
Bell 1986a; Percival 1998b), it is widely thought that the preferred Lorentz frame cannot be measured even
within the framework of dynamical collapse theories.
13
By contrast, the dynamical collapse theories in which the collapse time formula does not contain c are
not directly applicable in the relativistic domain.
14
Here we still use the standard synchrony for the convenience of practical realization.

87
 5.3. COLLAPSE DYNAMICS AND PREFERRED LORENTZ FRAME

the relativistic equation of collapse dynamics. Indeed, the equation of collapse dynamics,
whose non-relativistic form is denoted by Eq. (4.17), does contain a velocity term in order
to be relativistic invariant15 :

∆E
Pi (t + tP ) = Pi (t) + f (v) [δEs Ei − Pi (t)]. (5.23)
EP
where f (v) ≈ 1 + v/c when E ≈ pc, and v is the velocity of the experimental frame relative
to the preferred Lorentz frame. From this equation we can also derive the above relativistic
collapse time formula.
Therefore, according to our energy-conserved collapse model, the collapse time of a given
wave function will differ in different inertial frames. For example, when considering the
maximum difference of the revolution speed of the Earth with respect to the Sun is ∆v ≈
60km/s, the maximum difference of the collapse time measured in different times (e.g. spring
and fall respectively) on the Earth will be ∆τc ≈ 4×10−4 τc . As a result, the collapse dynamics
will single out a preferred Lorentz frame in which the collapse time of a given wave function
is longest, and the frame can also be determined by comparing the collapse times of a given
wave function in different frames16 . It may be expected that this preferred Lorentz frame
is the CMB-frame in which the cosmic background radiation is isotropic, and the one-way
speed of light is also isotropic in this frame.

15
This seems to be an inevitable consequence of the requirement of energy conservation for wavefunction
collapse.
16
In general, we can measure the collapse time of a wave function through measuring the change of the
interference between the corresponding collapse branches for an ensemble of identical systems. The main
technical difficulty of realizing such a measurement is to exclude the influence of environmental decoherence
(cf. Marshall et al 2003).

88
 Bibliography

[1] Adler, S. L. (2002). Environmental influence on the measurement process in stochastic

reduction models, J. Phys. A: Math. Gen. 35, 841-858.

[2] Adler, S. L. (2007). Comments on proposed gravitational modifications of Schrödinger

dynamics and their experimental implications. J. Phys. A: Math. Gen. 40, 755-764.

[3] Aharonov, Y., Albert, D. Z. and Vaidman, L. (1988). How the result of a measurement
of a component of the spin of a spin-1/2 particle can turn out to be 100. Phys. Rev. Lett.
60, 1351.

[4] Aharonov, Y., Anandan, J. and Vaidman, L. (1993). Meaning of the wave function, Phys.
Rev. A 47, 4616.

[5] Aharonov, Y., Anandan, J. and Vaidman, L. (1996). The meaning of protective measure-
ments, Found. Phys. 26, 117.

[6] Aharonov, Y., Englert, B. G. and Scully M. O. (1999). Protective measurements and
Bohm trajectories, Phys. Lett. A 263, 137.

[7] Aharonov, Y., Erez, N. and Scully M. O. (2004). Time and Ensemble Averages in
Bohmian Mechanics. Physica Scripta 69, 81-83.

[8] Aharonov, Y. and Vaidman, L. (1993). Measurement of the Schrödinger wave of a single
particle, Phys. Lett. A 178, 38.

[9] Aharonov, Y. and Vaidman, L. (1996). About position measurements which do not show
the Bohmian particle position, in J. T. Cushing, A. Fine, and S. Goldstein (eds.), Bohmian
Mechanics and Quantum Theory: An Appraisal, Dordrecht: Kluwer Academic, 141-154.

[10] Albert, D. (1992). Quantum Mechanics and Experience. Cambridge, MA: Harvard Uni-
versity Press.

[11] Albert, D. (1996). Elementary Quantum Metaphysics, in J. T. Cushing, A. Fine, and

S. Goldstein (eds.), Bohmian Mechanics and Quantum Theory: An Appraisal, Dordrecht:
Kluwer Academic, 277-284.

[12] Albert, D. Z. and Loewer, B. (1988). Interpreting the Many Worlds Interpretation,
Synthese, 77, 195-213.

[13] Allori, V., Goldstein, S., Tumulka, R., and Zanghi, N. (2008), On the Common Structure
of Bohmian Mechanics and the Ghirardi-Rimini-Weber Theory, British Journal for the
Philosophy of Science 59 (3), 353-389.

[14] Anandan, J. (1993). Protective Measurement and Quantum Reality. Found. Phys. Lett.,
6, 503-532.

[15] Bacciagaluppi, G. (2008). The role of decoherence in quantum mechanics. The

Stanford Encyclopedia of Philosophy (Fall 2008 Edition), Edward N. Zalta (eds.),
http://plato.stanford.edu/archives/fall2008/entries/qm-decoherence/.

[16] Bacciagaluppi G. and Valentini, A. (2009). Quantum Theory at the Crossroads: Recon-
sidering the 1927 Solvay Conference. Cambridge: Cambridge University Press.

[17] Barrett, J., Leifer, M. and Tumulka, R. (2005). Bells jump process in discrete time.
Europhys. Lett. 72, 685.

89
 BIBLIOGRAPHY

[18] Barrett, J. A. (1999). The Quantum Mechanics of Minds and Worlds. Oxford: Oxford
University Press.

[19] Barrett, J. A. (2011). Everett’s Relative-State Formulation of Quantum Mechanics, The

Stanford Encyclopedia of Philosophy (Spring 2011 Edition), Edward N. Zalta (ed.), URL
= http://plato.stanford.edu/archives/spr2011 /entries/qm-everett/.

[20] Bassi, A., Ippoliti, E., and Vacchini, B. (2005). On the energy increase in space-collapse
models. J. Phys. A : Math. Gen. 38, 8017.

[21] Bassi, A. (2007). Dynamical reduction models: present status and future developments.
J. Phys.: Conf. Series 67, 012013.

[22] Bedingham, D. J. (2011). Relativistic state reduction dynamics. Found. Phys. 41, 686-
704.

[23] Bell, J. S. (1986a). in The Ghost in the Atom, Davies, P. C. W.and Brown, J. R. eds.
Cambridge: Cambridge University Press. pp. 45-57.

[24] Bell, J. S. (1986b). Beables for quantum field theory. Phys. Rep. 137, 49-54.

[25] Bell, J. S. (1990). Against measurement, in A. I. Miller (eds.), Sixty-Two Years of Uncer-
tainty: Historical Philosophical and Physics Enquiries into the Foundations of Quantum
Mechanics. Berlin: Springer, 17-33.

[26] Belot, G. (2011). Quantum states for primitive ontologists: A case study. http://philsci-
archive.pitt.edu/8468/. Forthcoming in European Journal for Philosophy of Science.

[27] Belousek, D. (2003). Formalism, ontology and methodology in Bohmian mechanics.

Foundations of Science 8, 109-172.

[28] Bohm, D. (1952). A suggested interpretation of quantum theory in terms of “hidden”

variables, I and II. Phys. Rev. 85, 166-193.

[29] Bohm D. and Hiley, B. J. (1993). The Undivided Universe: An Ontological Interpreta-
tion of Quantum Theory. London: Routledge.

[30] Born, M. (1926). Quantenmechanik der Stoβvorgänge. Z. Phys. 38, 803; English trans-
lation in Ludwig, G., eds., 1968, Wave Mechanics, Oxford: Pergamon Press: 206.

[31] Boughn, S. (2009). Nonquantum Gravity. Found. Phys. 39, 331.

[32] Brown, H. R., Dewdney, C. and Horton, G. (1995). Bohm particles and their detection
in the light of neutron interferometry. Found. Phys. 25, 329.

[33] Carlip, S. (2008). Is quantum gravity necessary? Class. Quant. Grav., 25, 154010-1.

[34] Christian, J. (2001). Why the quantum must yield to gravity. In: Physics Meets Phi-
losophy at the Planck Scale, C. Callender and N. Huggett (eds.). Cambridge: Cambridge
University Press. pp.305.

[35] Czachor, M. (1996). Nonlinear Schrödinger equation and two-level atoms. Phys. Rev. A
53, 1310.

[36] Dass, N. D. H. and Qureshi, T. (1999). Critique of protective measurements. Phys. Rev.
A 59, 2590.

90
 BIBLIOGRAPHY

[37] Daumer, M., Dürr, D., Goldstein, S., and Zangh, N. (1997). Naive Realism About
Operators, Erkenntnis 45, 379-397.

[38] de Broglie, L. (1928). in: Electrons et photons: Rapports et discussions du cinquime

Conseil de Physique Solvay, eds. J. Bordet. Paris: Gauthier-Villars. pp.105. English trans-
lation: Bacciagaluppi and Valentini (2009).

[39] DeWitt, B. S. and Graham, N. (eds.) (1973). The Many-Worlds Interpretation of Quan-
tum Mechanics, Princeton: Princeton University Press.

[40] DeWitt, C. and Rickles, D. (eds.) (2011). The Role of Gravitation in Physics: Report
from the 1957 Chapel Hill Conference. Max Planck Research Library for the History and
Development of Knowledge, Volume 5.

[41] Dickson, M. (1995). An Empirical Reply to Empiricism: Protective Measurement Opens

the Door for Quantum Realism, Phil. Sci. 62, 122-126.

[42] Diósi, L. (1984). Gravitation and the quantum-mechanical localization of macro-objects.

Phys. Lett. A 105, 199-202.

[43] Diósi, L. (1989). Models for universal reduction of macroscopic quantum fluctuations.
Phys. Rev. A 40, 1165-1173.

[44] Diósi, L. (2007). Notes on certain Newton gravity mechanisms of wave function locali-
sation and decoherence. J. Phys. A: Math. Gen. 40, 2989-2995.

[45] Dirac, P. A. M. (1930). The Principles of Quantum Mechanics, Oxford: Clarendon

Press.

[46] Drezet, A. (2006). Comment on “Protective measurements and Bohm trajectories”,

Phys. Lett. A 350, 416.

[47] Duff, M. J., Okun, L. B., and Veneziano, G. (2002). Trialogue on the number of funda-
mental constants. Journal of High Energy Physics, 0203, 23.

[48] Dürr, D., Goldstein, S., and Zanghi, N. (1992). Quantum equilibrium and the origin of
absolute uncertainty. Journal of Statistical Physics 67, 843-907.

[49] Dürr, D., Goldstein, S., and Zanghı̀, N. (1997). Bohmian mechanics and the meaning
of the wave function, in Cohen, R. S., Horne, M., and Stachel, J., eds., Experimental
Metaphysics - Quantum Mechanical Studies for Abner Shimony, Volume One; Boston
Studies in the Philosophy of Science 193, Boston: Kluwer Academic Publishers.

[50] Edwards, W. F. (1963). Special relativity in anisotropic space, Am. J. Phys. 31, 482-489.

[51] Einstein, A. (1905). Zur Elektrodynamik bewegter Körper, Ann. d. Phys. 17, 891-921.
English translations in Stachel, J. (eds.), Einstein’s Miraculous Year, Princeton: Princeton
University Press, 1998, pp. 123-160.

[52] Englert, B. G., Scully, M. O., Süssmann, G., Walther, H. (1992). Z. Naturforsch. Sur-
realistic Bohm Trajectories. 47a, 1175.

[53] Everett, H. (1957). ‘Relative state’ formulation of quantum mechanics. Rev. Mod. Phys.
29, 454-462.

[54] Feynman, R. (1995). Feynman Lectures on Gravitation. B. Hatfield (eds.), Reading,

Massachusetts: Addison-Wesley.

91
 BIBLIOGRAPHY

[55] Feynman, R. (2001). The Character of Physical Law (Messenger Lectures, 1964). Cam-
bridge: The MIT Press.

[56] Friedman J. R., Patel, V., Chen, W., Tolpygo, S. K. and Lukens, J. E. (2000). Nature
406, 43.

[57] Frigg, R and Hoefer, C. (2007). Probability in GRW Theory. Studies in the History and
Philosophy of Modern Physics 38, 371-389.

[58] Gao, S. (1993). A suggested interpretation of quantum mechanics in terms of discontin-

uous motion (unpublished manuscript).

[59] Gao, S. (1999). The interpretation of quantum mechanics (I) and (II). physics/9907001,
physics/9907002.

[60] Gao, S. (2000). Quantum Motion and Superluminal Communication, Beijing: Chinese
Broadcasting and Television Publishing House. (in Chinese)

[61] Gao, S. (2003). Quantum: A Historical and Logical Journey. Beijing: Tsinghua Univer-
sity Press. (in Chinese)

[62] Gao, S. (2004). Quantum collapse, consciousness and superluminal communication,

Found. Phys. Lett, 17, 167-182.

[63] Gao, S. (2005). A conjecture on the origin of dark energy. Chin. Phys. Lett. 22, 783.

[64] Gao, S. (2006a). A model of wavefunction collapse in discrete space-time. Int. J. Theor.
Phys. 45, 1943-1957.

[65] Gao, S. (2006b). Quantum Motion: Unveiling the Mysterious Quantum World. Bury St
Edmunds, Suffolk U.K.: Arima Publishing.

[66] Gao, S. (2008). God Does Play Dice with the Universe. Bury St Edmunds, Suffolk U.K.:
Arima Publishing.

[67] Gao, S. (2010). On Diósi-Penrose criterion of gravity-induced quantum collapse. Int. J.

Theor. Phys. 49, 849-853.

[68] Gao, S. (2011a). The wave function and quantum reality, in Proceedings of the In-
ternational Conference on Advances in Quantum Theory, A. Khrennikov, G. Jaeger, M.
Schlosshauer, G. Weihs (eds), AIP Conference Proceedings 1327, 334-338.

[69] Gao, S. (2011b). Meaning of the wave function, International Journal of Quantum Chem-
istry. 111, 4124-4138.

[70] Gao, S. (2011c). Is gravity an entropic force? Entropy special issue “Black Hole Ther-
modynamics”, Jacob D. Bekenstein (eds), 13, 936-948.

[71] Gao, S. (2011d). Comment on “How to protect the interpretation of the wave function
against protective measurements” by Jos Uffink. http://philsci-archive.pitt.edu/8942.

[72] Garay, L. J. (1995). Quantum gravity and minimum length. International Journal of
Modern Physics A 10, 145.

[73] Ghirardi, G. C. (1997), Quantum Dynamical Reduction and Reality: Replacing Prob-
ability Densities with Densities in Real Space, Erkenntnis, 45, 349.

92
 BIBLIOGRAPHY

[74] Ghirardi, G. C. (2008). Collapse Theories, The Stanford Encyclo-

pedia of Philosophy (Fall 2008 Edition), Edward N. Zalta (eds.),
http://plato.stanford.edu/archives/fall2008/entries/qm-collapse/.

[75] Ghirardi, G. C., Grassi, R. and Benatti, F. (1995). Describing the macroscopic world:
Closing the circle within the dynamical reduction program. Found. Phys., 25, 313-328.

[76] Ghirardi, G. C., Pearle, P. and Rimini, A. (1990). Markov-processes in Hilbert-space

and continuous spontaneous localization of systems of identical particles. Phys. Rev. A
42, 78.

[77] Ghirardi, G.C., Rimini, A., and Weber, T. (1986). Unified dynamics for microscopic
and macroscopic systems. Phys. Rev. D 34, 470.

[78] Gisin, N. (1990). Weinberg’s non-linear quantum mechanics and superluminal commu-
nications, Phys. Lett. A 143, 1-2.

[79] Goldstein, S. (2009). Bohmian Mechanics, The Stanford Encyclo-

pedia of Philosophy (Spring 2009 Edition), Edward N. Zalta (eds.),
http://plato.stanford.edu/archives/spr2009/entries/qm-bohm/.

[80] Goldstein, S. and Teufel, S. (2001). Quantum spacetime without observers: Ontological
clarity and the conceptual foundations of quantum gravity, in Callender, C. and Huggett,
N., eds., Physics meets Philosophy at the Planck Scale, Cambridge: Cambridge University
Press.

[81] Grabert, H., Hänggi, P. and Talkner, P. (1979). Is quantum mechanics equivalent to a
classical stochastic process? Phys. Rev. A 19, 24402445.

[82] Grünbaum, A. (1973). Philosophical Problems of Space and Time (Boston Studies in the
Philosophy of Science, Volume 12), 2nd enlarged edition, Dordrecht/Boston: D. Reidel.

[83] Hanson, H. and Thoma, F. (2011) Bohmian Mechanics, https://wiki.physik.uni-

muenchen.de/TMP/images/b/b4/BohmianMechnics.pdf. Accessed on: April 21, 2011.

[84] Holland, P. (1993). The Quantum Theory of Motion: An Account of the de Broglie-
Bohm Causal Interpretation of Quantum Mechanics. Cambridge: Cambridge University
Press.

[85] Holland, P. and Philippidis, C. (2003). Implications of Lorentz covariance for the guid-
ance equation in two-slit quantum interference. Phys. Rev. A 67, 062105.

[86] Holman M. (2006) On arguments for linear quantum dynamics. quant-ph/0612209.

[87] Hosten, O. and Kwiat, P. G. (2008). Observation of the Spin Hall Effect of Light via
Weak Measurements. Science 319, 787-790.

[88] Hughston, L. P. (1996). Geometry of stochastic state vector reduction. Proc. Roy. Soc.
A 452, 953.

[89] Janis, A. (2010). Conventionality of simultaneity, The Stanford Encyclope-

dia of Philosophy (Fall 2010 Edition), Edward N. Zalta (eds.), URL =
http://plato.stanford.edu/archives/fall2010/entries/spacetime-convensimul/.

[90] Joos, E. and Zeh, H. D. (1985). The emergence of classical properties through interaction
with the environment. Zeitschrift fr Physik B 59, 223-243.

93
 BIBLIOGRAPHY

[91] Kochen, S. and Specker, E. (1967). The Problem of Hidden Variables in Quantum
Mechanics, J. Math. Mech. 17, 59-87.

[92] Lewis, P. (2004). Life in configuration space, British Journal for the Philosophy of
Science 55, 713-729.

[93] Marshall, W., Simon, C., Penrose, R., and Bouwmeester, D. (2003). Towards quantum
superpositions of a mirror, Phys. Rev. Lett. 91, 130401.

[94] Maudlin, T. (2002). Quantum Non-Locality and Relativity: Metaphysical Intimations

of Modern Physics. Oxford: Blackwell.

[95] Miller, D. J. (2011). State-dependent rotations of spins by weak measurements, Phys.

Rev. A 83, 032121.

[96] Monton, B. (2002). Wave function ontology. Synthese 130, 265-277.

[97] Monton, B. (2006). Quantum mechanics and 3N-dimensional space. Philosophy of Sci-
ence, 73(5), 778-789.

[98] Moore, W. J. (1994). Schrödinger: Life and Thought. Cambridge: Cambridge University
Press.

[99] Mott, N. F. (1929). The Wave Mechanics of α-ray Tracks, Proceedings of the Royal
Society of London A 126, 79-84.

[100] Nakamura, K. et al (Particle Data Group) (2010). J. Phys. G: Nucl. Part. Phys. 37,
075021.

[101] Nelson, E. (1966). Derivation of the Schrödinger equation from Newtonian mechanics.
Phys. Rev. 150, 1079-1085.

[102] Nelson, E. (2005). The mystery of stochastic mechanics, manuscript 2005-11-22.

[103] Okun, L. B. (2009). Energy and Mass in Relativity Theory. New Jersey: World Scien-
tific.

[104] Pearle, P. (1989). Combining stochastic dynamical state-vector reduction with spon-
taneous localization. Phys. Rev. A 39, 2277.

[105] Pearle, P. (1999). Collapse models. In: Petruccione, F. and Breuer, H. P. (eds.), Open
Systems and Measurement in Relativistic Quantum Theory. Springer Verlag, New York.

[106] Pearle, P. (2000). Wavefunction Collapse and Conservation Laws. Found. Phys. 30,
1145-1160.

[107] Pearle, P. (2004). Problems and aspects of energy-driven wavefunction collapse models.
Phys. Rev. A 69, 42106.

[108] Pearle, P. (2007). How stands collapse I. J. Phys. A: Math. Theor., 40, 3189-3204.

[109] Pearle, P. (2009). How stands collapse II. in Myrvold, W. C. and Christian, J. eds.,
Quantum Reality, Relativistic Causality, and Closing the Epistemic Circle: Essays in
Honour of Abner Shimony. The University of Western Ontario Series in Philosophy of
Science, 73(IV), 257-292.

94
 BIBLIOGRAPHY

[110] Pearle, P. and Squires, E. (1996). Gravity, energy conservation and parameter values
in collapse models. Found. Phys. 26, 291.

[111] Penrose, R. (1996). On gravitys role in quantum state reduction. Gen. Rel. Grav. 28,
581.

[112] Penrose, R. (1998). Quantum computation, entanglement and state reduction. Phil.
Trans. R. Soc. Lond. A 356, 1927.

[113] Percival, I. C. (1995). Quantum space-time fluctuations and primary state diffusion.
Proc. Roy. Soc. A 451, 503.

[114] Percival, I. C. (1998a). Quantum State Diffusion. Cambridge: Cambridge University

Press.

[115] Percival, I. C. (1998b). Quantum transfer function, weak nonlocality and relativity,
Phys. Lett. A 244, 495-501.

[116] Reichenbach, H. (1958). The Philosophy of Space and Time, New York: Dover.

[117] Rosenfeld, L. (1963). On quantization of fields. Nucl. Phys. 40, 353-356.

[118] Rovelli, C. (2011). “Forget time”: Essay written for the FQXi contest on the Nature
of Time. Found. Phys. 41, 1475-1490.

[119] Salzman, P. J. (2005). Investigation of the Time Dependent Schrödinger-Newton Equa-

tion, Ph.D. Dissertation, University of California at Davis.

[120] Salzman, P. J. and Carlip, S. (2006). A possible experimental test of quantized gravity.
arXiv: gr-qc/0606120.

[121] Saunders, S., Barrett, J. A., Kent, A. and Wallace, D. (eds.) (2010) Many Worlds?
Everett, Quantum Theory, and Reality. Oxford: Oxford University Press.

[122] Schrödinger, E. (1926). Quantizierung als Eigenwertproblem (Vierte Mitteilung). Ann.

d. Phys. (4) 81, 109-139. English translation: Quantisation as a Problem of Proper Values.
Part IV, Reprint in Schrödinger, E. (1982). Collected Papers on Wave Mechanics. New
York: Chelsea Publishing Company, pp. 102-123.

[123] Shankar, R. (1994). Principles of Quantum Mechanics, 2nd ed. New York: Plenum.

[124] Sonego, S. and Pin, M. (2005). Deriving relativistic momentum and energy. Eur. J.
Phys. 26, 33-45.

[125] Squires, E. J. (1992). Explicit collapse and superluminal signaling, Phys. Lett. A 163,
356-358.

[126] Suárez, M. (2004). Quantum selections, propensities and the problem of measurement,
British Journal for the Philosophy of Science, 55(2), 219-55.

[127] Suárez, M. (2007). Quantum Propensities, Studies in the History and Philosophy of
Modern Physics 38, 418-438.

[128] Tooley, M. (1988). In defence of the existence of states of motion. Philosophical Topics
16, 225-254.

95
 BIBLIOGRAPHY

[129] Uffink, J. (1999). How to protect the interpretation of the wave function against pro-
tective measurements. Phys. Rev. A 60, 3474-3481. arXiv: quant-ph/9903007.

[130] Vaidman, L. (2009) Protective measurements, in Greenberger, D., Hentschel, K., and
Weinert, F. (eds.), Compendium of Quantum Physics: Concepts, Experiments, History
and Philosophy. Springer-Verlag, Berlin. pp.505-507.

[131] Valentini, A. (1997). On Galilean and Lorentz invariance in pilot-wave dynamics. Phys.
Lett. A 228, 215-222.

[132] Valentini, A. (2010). De Broglie-Bohm Pilot-Wave Theory: Many Worlds in Denial?

in Saunders, S., Barrett, J., Kent, A. and Wallace, D. (eds.). Many Worlds? Everett,
Quantum Theory, and Reality. Oxford: Oxford University Press.

[133] Vink, J. C. (1993). Quantum mechanics in terms of discrete beables. Phys. Rev. A 48,
1808.

[134] von Neumann, J. (1955). Mathematical Foundations of Quantum Mechanics, Prince-

ton: Princeton University Press. (Translated by R. Beyer from Mathematische Grundlagen
der Quantenmechanik, Springer: Berlin, 1932)

[135] Wallace, D. and Timpson, C. G. (2010). Quantum mechanics on spacetime I: Spacetime

state realism. British Journal for the Philosophy of Science, 61 (4): 697-727.

[136] Wallstrom, T. (1994). Inequivalence between the Schrdinger equation and the
Madelung hydrodynamic equations. Phys. Rev. A 49, 16131617.

[137] Wheeler, J. A. and W. H. Zurek (eds.) (1983). Quantum Theory and Measurement,
Princeton: Princeton University Press.

[138] Winnie, J. (1970). Special relativity without one-way velocity assumptions: I and II,
Philosophy of Science 37, 81-99, 223-238.

[139] Yablonovitch, E. (1987). Inhibited spontaneous emission in solid-state physics and

electronics. Phys. Rev. Lett. 58, 2059.

96