quant-ph/0405069

On the principles of quantum mechanics

Eijiro Sakai∗
Department of Physics and Earth Sciences,
University of the Ryukyus, Okinawa, 903-0213, Japan
(Dated: October 22, 2018)
arXiv:quant-ph/0405069v4 30 Aug 2005

Abstract
We propose six principles as the fundamental principles of quantum mechanics: principle of
space and time, Galilean principle of relativity, Hamilton’s principle, wave principle, probability
principle, and principle of indestructibility and increatiblity of particles. We deductively develop
the formalism of quantum mechanics on the basis of them: we determine the form of the La-
grangian that satisfies requirements of these principles, and obtain the Schrödinger equation from
the Lagrangian. We also derive the canonical commutation relations. Then we adopt the following
four guide lines. First, we do not premise the relations between dynamical variables in classical
mechanics. Second, we define energy, momentum, and angular momentum as the constants of
motion that are derived from homogeneity and isotropy in space and time on the basis of principle
of space and time. Since energy and momentum are quantitatively defined in classical mechanics,
we define them in quantum mechanics so that the corresponding conservation laws are satisfied in a
coupling system of a quantum particle and a classical particle. Third, we define Planck’s constant
and the mass of a particle as proportionality constants between energy and frequency due to one of
Einstein-de Broglie formulas and between momentum and velocity, respectively. We shall obtain
the canonical commutation relations and the Schrödinger equation for a particle in an external field
in the definitive form. We shall also prove that relations between dynamical variables in quantum
mechanics have the same forms for those in classical mechanics.

PACS numbers: 03.65.Ca

1
I. INTRODUCTION

About 80 years were passed after quantum mechanics was established. Almost all physi-
cists recognize that quantum mechanics is completely right and that it provides the basis
for the modern physics. Quantum mechanics, however, has the flaw that it does not form a
self-contained formalism.
In most treatises on quantum mechanics, to begin with, the Schrödinger equation is intro-
duced heuristically with the help of Einstein-de Broglie formulas, and next, “interpretations”
are ad hoc added to which are required to understand results obtained by applying it to
specific concrete problems. Such developments are useful in order to make readers learn
quantum mechanics quickly. It is, however, very unsatisfactory since quantum mechanics
is the basis of the modern physics. On the one hand, classical mechanics is deductively
established on the basis of three laws of motion.
In the Course of Theoretical Physics, Landau first determine the Lagrangians for the
nonrelativistic1 and relativistic2 particles and the Lagrangian density for the electromagnetic
field in vacuum3 from physical principles and then derive the equations of motion and the
Maxwell’s equations. He, however, renounce deriving the Schrödinger equation in the same
way, because he had believed that it is in priciple impossible to formulate the basic concepts
of quantum mechanics without using classical mechanics.4
Quantum mechanics includes classical mechanics in the limit; classical mechanics is the
approximated theory of quantum mechanics, because quantum mechanics is reduced to
classical mechanics in the limit in which Planck’s constant approaches zero. Therefore, the
author convinces himself that quantum mechanics should be established on its own proper
basis independently of classical mechanics.
In order to set quantum mechanics as the foundation of the modern physics in the true
sense, quantum mechanics must be reconstructed from the “quantized” mechanics into the
true “quantum” mechanics. The true “quantum” mechanics must not be premised on clas-
sical mechanics that is approximate to quantum mechanics. In the present paper we shall
propose six principles as the fundamental principles of the true quantum mechanics: princi-
ple of space and time, Galilean principle of relativity, Hamilton’s principle, wave principle,
probability principle, and principle of indestructibility and increatiblity of particles. We
shall deductively develop the formalism of quantum mechanics on the basis of them: we

2
determine the form of the Lagrangian that satisfies requirements of these principles, and
obtain the Schrödinger equation from the Lagrangian. We also derive the canonical com-
mutation relations. Then we adopt the following four guide lines. First, we do not premise
the relations between dynamical variables in classical mechanics such as E = p2 /2m and
L = r × p. Second, we define energy, momentum, and angular momentum as the constants
of motion that are derived from homogeneity and isotropy in space and time on the basis
of principle of space and time. Since energy and momentum are quantitatively defined in
classical mechanics, we define them in quantum mechanics so that the corresponding conser-
vation laws are satisfied in a coupling system of a quantum particle and a classical particle.
Third, we define Planck’s constant and the mass of a partilce as proportionality constants
between energy and frequency due to one of Einstein-de Broglie formulas and between mo-
mentum and velocity, respectively. In the interests of simplicity, we assume that the particle
has no internal degrees of freedom.
In Sec. II, we shall propose the six principles of quantum mechanics: principle of space and
time, Galilean principle of relativity, Hamilton’s principle, wave principle, and probability
principle, and principle of indestructibility and increatiblity of particles. In Sec. III, we
shall describe basic concepts that are directly derived from the principles: the expectation
values of operator, properties of the position operator, and the position probability density
of a particle. In the following three sections we shall discuss the properties of generators of
three fundamental transformations: translation, rotation and Galilean transformation. In
Sec. VII, we shall obtain the form of the Lagrangian for a free particle, using the principle
of space and time, probability principle, Galilean principle of relativity, and principle of
indestructibility and increatibility of particles. The resultant Lagrangian gives the essentially
true form of the Schrödinger equation for a free particle except for that it includes three
undetermined constants. In Sec. VIII, we shall show that two of the undetermined constants
are Planck’s constant and the mass of a particle on the basis of conservation laws of energy
and momentum. In Sec. IX, we shall obtain the Lagrangian for two interacting particles,
and from it we shall find the Lagrangian for a particle in an external field as that for the
relative motion. Furthermore we shall determine the last undetermined constant by the
conservation law of momentum for a system consisting of a quantum particle and a classical
particle. This results that we obtain the general form of the Schrödinger equation for a
particle in an external field and verify the commutation relations between the position and

3
momentum operators. In the last section we shall give conclusion.
We are interested in the logical structure of quantum mechanics, so we shall not pursue
mathematical rigorousness. We will not dwell on the theory of measurement.

II. PRINCIPLES OF QUANTUM MECHANICS

In this section, we shall propose the six fundamental principles of nonrelativistic quantum
mechanics, which deals with a microscopic particle that is regarded as a particle from the
point of view of the classical theory. We shall not consider the quantum theory of such as
the electromagnetic field that is regarded classically as wave.

1. Principle of Space and Time

Principle of space and time is stated as

Space is homogeneous, isotropic and flat, which means that space is Euclidean.
Time is homogeneous.

This is the geometric principle with regard to space and time in which physical phenomena
occur. In classical mechanics, this principle is implicitly included in the first law of motion,
“the law of inertia”.

2. Galilean Principle of Relativity

Galilean principle of relativity is stated as

The laws of physics are covariant under Galilean transformations.

Since our purpose is to establish nonrelativistic quantum mechanics, we should naturally

adopt Galilean principle of relativity.

3. Hamilton’s Principle
Hamilton’s principle is stated as

The system makes a motion so that its action has the minimum value. The
action is defined as the integral over time of the Lagrangian. The Lagrangian is
a real additive scalar function of variables that represent states of a system of
interest. To gurantee the covariance of physical laws, the Lagrangian should be

4
 invariant with respect to symmetrical transformations that come from the first
two principles.

From this principle we obtain equation of motion.

4. Wave Principle
Wave principle is stated as

States of a microscopic particle are represented by state vectors that satisfy

the principle of superposition. Especially each state of a microscopic particle is
completely determined except for internal degrees of freedom if the position of
the particle is specified. All the state vectors, therefore, that represent a particle
located at a particular position constitute a complete set. Any dynamical variable
is represented by an operator which operates on state vectors.

A state vector is usually expressed as |ψi, and its adjoint (|ψi)† is expressed as hψ|.
When we need to distinguish an original vector and its adjoint vector, we call an original
vector a ket and its adjoint a bra after Dirac.5 In the present paper we assume that state
vectors are always normalized. The principle of superposition means that a superposition
(a linear combination) of two state vectors, each of which represents a possible microscopic
state, represents also another possible microscopic state. Dynamical variable operators must
be linear in order to satisfy the principle of superposition. The properties of them are
derived from the principles of quantum mechanics, as shall be shown below. The operator
corresponding to a dynamical variable A is expressed by Â.
In the interests of simplicity, we assume that the particle has no internal degrees of
freedom.

5. Probability Principle
Probability principle is stated as

If a state of a particle is given by a normalized state vector |ψi that is a

superposition of |ki:
X
|ψi = ck |ki,
k

where |ki is the eigenvector corresponding to an eigenvalue ak of an operator

Â, the result of measurement of the dynamical variable A is equal to one of

5
 eigenvalues of Â, ak , and the probability that we obtain ak is given by |ck |2 .

The probability principle is in close connection with the theory of measurement. Since
there exists no theory of measurement that is generally accepted, we have inaccurately stated
the principle here. For example, we have not referred to the “reduction of wave packet”.
We shall revise the principle, when the theory of measurement have been established.
The complete theory of measurement, however, is not needed to develop the formalism of
quantum mechanics on the basis of principles. For it only three theorems are needed. First
of them is that dynamical variable operators are observable. Second is that the expectation
value of measurement on a dynamical variable can be calculated from the probability princi-
ple. The last is that the square of the magnitude of the wave function gives the probability
density of existence of the particle. These three theorems will be derived in the next section.

6. Principle of Indestructibility and Increatibility of particles

Indestructibility and Increatibility is stated as

Particles that quantum mechanics deals with are indestructible and increatible.

We require that particles are indestructible and increatible, since quantum mechanics is
devoted to low-energy phenomena. This principle is less general than the other principles
and auxiliary.

We shall deductively develop the formalism of quantum mechanics on the basis of these
principles in the following sections.

III. FUNDAMENTAL CONCEPTS OF QUNATUM MECHANICS

In this section, we shall describe important concepts directly derived from the principles
of quantum mechanics.

A. Property of operators

From the probability principle, we shall derive the property of the operator corresponding
to a dynamical variable.

6
 The result of measurement of a dynamical variable A must always be a real number.
Therefore the operator, Â, corresponding to A must be self-adjoint. Whenever we measure
A, we should get an eigenvalue of  as a result. This requires that the eigenvectors |ki of Â
must constitute a complete set:
X
|kihk| = 1.
k

Therefore  is an observable.
Assume that a normalized state vector |ψi is expanded as
X
|ψi = ck |ki
k

in terms of eigenvectors |ki of a dynamical variable operatorÂ, where the expansion coeffi-
cients are given by
ck = hk|ψi.

From the probability principle, if the state of the system is given by this state vector, the
expectation value hÂi is expressed as
X
hÂi = |ck |2 ak ,
k

where ak is the eigenvalue corresponding to the vector |ki. This expectation value can be
expressed as
X X X
hÂi = |ck |2 ak = hψ|kiak hk|ψi = hψ|Â|kihk|ψi = hψ|Â|ψi,
k k k

where use has been made of the completeness of the eigenvectors |ki.

B. Position operator

We express the position operator of a microscopic particle as r̂ and its eigenvector as |ri;

r̂|ri = r|ri. (1)

We assume that eigenvectors of the position operator are not degenerate, because we only
deal with particles without internal degrees of freedom in the present paper. This follows
that
hr|r′ i = 0 for r 6= r′ .

7
 From the wave principle, all the vectors |ri should constitute a complete set, so that the
equation Z
dr|rihr| = 1

is valid. The relation

Z 2 Z Z
dr|rihr| = dr dr′ |rihr|r′ihr′ | = 1

gives
hr|r′i = δ(r − r′ ),

which is the orthonormal condition for the eigenvectors of the position operator.
From Eq. (1) we obtain the relation

x̂j x̂k |ri = xj xk |ri = x̂k x̂j |ri ,

where x̂j denotes the j-th Cartesian component of r̂. The above equation leads that

[x̂j , x̂k ] = 0, (2)

since |ri constitute a complete set.

C. Position probability density of the particle

In accordance with the general expression for the expectation value, the expectation value
of position is given by
Z
hr̂i = hψ|r̂|ψi = drhψ|r̂|rihr|ψi
Z Z
= drhψ|rirhr|ψi = dr r|hr|ψi|2,

which means that |hr|ψi|2 represents the position probability density of the particle. The
quantity hr|ψi is the wave function, and is usually designated by ψ(r):

ψ(r) = hr|ψi

Since principle of indestructibility and increatibility of particles, the probability find-

ing the particle somewhere is equal to unity independently of time. Thus the equation of
continuity
∂
|ψ(r)|2 + divJ = 0 (3)
∂t

8
must be valid, where J is called the probability current density and is a real vector. We
require that this equation of continuity derives directly from the equation of motion for the
state vector. Thus J should be expressed with ψ(r) and ψ ∗ (r), since the wave function
should satisfy a linear equation of motion from the wave principle.

IV. TRANSLATION

In this and the next two sections we shall discuss the basic theories of translation, rotation
and the Galilean transformation, to prepare for deriving the equation of motion for the state
vector.

A. Translation and its generator

The translation operator Û(a) is defined by

Û(a)|ri = |r + ai. (4)

The translation operator must evidently be unitary. From principle of space and time, all
translations commute with each other and their translation vectors are additive. Translation
operators therefore satisfy the following relation

Û (a)Û (b) = Û (b)Û (a) = Û (a + b). (5)

From the definition of the translation operator we obtain the expression

Û † (a)r̂Û (a)|ri = (r + a)|ri,

which means from the completeness of the kets |ri that

Û † (a)r̂Û (a) = r̂ + a. (6)

From Eq. (5), the translation operator is given by

 
a · p̂
Û(a) = exp −i , (7)
θ
where p̂ is a self-adjoint operator, which is called the translation generator here after, and
must satisfy the commutation relations

[p̂j , p̂k ] = 0. (8)

9
The quantity θ is a real constant. Usually, the translation generator is self-evidently identi-
fied with momentum operator and θ is set to be equal to Planck’s constant divided by 2π.
We do not regard, however, them as self-evident. We shall verify them in Secs. VIII and IX,
respectively.
From Eqs. (6) and (7), we obtain the commutation relations between the position operator
and the translation generator
[x̂j , p̂k ] = iθδjk . (9)

In similar fashion we can obtain the commutation relation

[p̂, f (r̂)] = −iθ ∇f (r)|r=r̂ . (10)

B. Properties of the translation generator and its eigenvectors

From Eqs. (4) and (7), we obtain

p̂|ri = iθ∇|ri (11)

and
hr|p̂ = −iθ∇hr|. (12)

We denote eigenkets of the translation generator p̂ as |pi:

p̂|pi = p|pi

where p is a real vector. Multiplying the above equation by an eigenbra of the position
operator hr| from the left, we obtain, with the help of Eq. (12), a differential equation

−iθ∇hr|pi = phr|pi.

We can easily solve this equation to obtain

hr|pi = Ceip·r/θ .

We therefore obtain
Z Z
′ 2 ′
dphr|pihp|r i = |C| dpeip·(r−r )/θ = (2πθ)3 |C|2 δ(r − r′ ).

10
It follows from this that Z
dp|pihp| = 1 (13)

if we take the normalization constant as

1
C= .
(2πθ)3/2
The equation (13) means that all the eigenvectors for the translation generator constitute a
complete set. Thus the normalized eigenfunctions of the translation generator are
1
hr|pi = 3/2
eip·r/θ .
(2πθ)
The above discussion shows that the translation generator is an observable, which implies
that the translation generator can correspond to a certain dynamical variable.

C. General property of operators corresponding to translational-invariant dynam-

ical variables

Now we consider a dynamical variable  that is invariant with respect to space translation:

Û † (a)ÂÛ (a) = Â

By differentiating both sides of the above equation with respect to a and putting a = 0, we
obtain
h i
Â, p̂ = 0.

This means that any translational-invariant dynamical variable commutes with the transla-
tion generator. The operator  is in general represented as
Z Z Z Z
 = dp dp |pihp|Â|p ihp | = dp dp′ |piA(p, p′ )hp′ |
′ ′ ′

in terms of the eigenvectors of the translation generator p̂ that constitute a complete or-
thonormal set. The off-diagonal elements of A(p, p′ ) must vanish since  commutes with p̂.
Thus, Z Z
 = dp|piA(p)hp| = dpA(p̂)|pihp| = A(p̂),

which shows that  is a function only of the translation generator. Therefore any
translational-invariant dynamical variable is in general represented by a function only of
the translation generator.

11
 Similarly, a dynamical variable that commutes with the position operator is represented
by a function only of the position operator; if
h i
Â, r̂ = 0,

then
 = A(r̂).

V. ROTATION

A. Rotation and its generator

Suppose that a position of a particle, r, is transformed into r′ by a rotation ϕ

r → r′ = R(ϕ)r,

where R(ϕ) is a 3 × 3 orthogonal matrix corresponding to the rotation ϕ. The rotation

operator V̂ (ϕ) is defined by
V̂ (ϕ)|ri = |R(ϕ)ri.

The rotation operator must be unitary. From the definition of the rotation operator we
obtain
V̂ † (ϕ)r̂V̂ (ϕ) = R(ϕ)r̂ (14)

with the help of completeness of all the state vectors |ri.

It is evident that the relation
V̂ n (ϕ) = V̂ (nϕ)

is valid. We can therefore represent the rotation operator V̂ (ϕ), which is unitary, as
 
i
V̂ (ϕ) = exp − ϕ · L̂ , (15)
θ

where the constant θ has been set the same as that in Eq. (7), and L̂ is a self-adjoint
operator, which we call the rotation generator. In Sec. VIII we shall verify that the rotation
generator is identified with angular momentum operator. Usually the rotation generator
is self-evidently identified with angular momentum operator, but we do not regard it as
self-evident and shall verify it in Sec. VIII.

12
 Position r is transformed by an infinitesimal rotation δϕ as

R(δϕ)r ≈ r − r × δϕ,

which follows, with the help of Eqs. (14) and (15), that
h i
L̂j , x̂k = iθεjkl x̂l , (16)

where εjkl is the fully antisymmetric tensor with ε123 = 1 and if the same letter appears
twice as a subscript, the summation over that subscript is meant.

B. Vector and scalar operators

We consider how the translation generator is transformed by rotation. It follows from

the definition of the rotation and translation operators that

V̂ † (ϕ)Û (a)V̂ (ϕ)|ri = |r + R−1 (ϕ)ai = Û (R−1 (ϕ)a)|ri.

This gives
V̂ † (ϕ)Û (a)V̂ (ϕ) = Û (R−1 (ϕ)a)

or      
† i i −1 i
V̂ (ϕ) exp − p̂ · a V̂ (ϕ) = exp − p̂ · (R a) = exp − (Rp̂) · a ,
θ θ θ
which shows that the translation generator is transformed by the rotation as

V̂ † (ϕ)p̂V̂ (ϕ) = R(ϕ)p̂, (17)

in the same way as the position operator(see Eq. (14)). This means that the translation
generator be a vector operator, since a quantity composed of three components that is
transformed in the same way as the position operator by rotation is in general called a
vector. Equation (17) yields the commutation relations
h i
L̂j , p̂k = iθεjkl p̂l . (18)

Any vector operator  must also satisfy the commutation relations

h i
L̂j , Âk = iθεjkl Âl . (19)

13
 Since a quantity that is unchanged under rotation is in general called a scalar, a scalar
operator should commute with the rotation generator. It is easily seen that the inner product
of two arbitrary vector operators  and B̂ commutes with the rotation generator
h i
L̂, Â · B̂ = 0,

which means that the inner product is a scalar, as should be.

C. Explicit expression for the rotation generator

It follows from the commutation relations (2),(8) and (9) that the rotation generator
satisfies the commutation relations (16) and (18) if

L̂j = εjkl x̂k p̂l . (20)

We would assume that there exists another operator L̂′j that also satisfies the relations (16)
and (18) but is different from Eq. (20):
h i h i
L̂j − L̂′j , x̂k = 0, L̂j − L̂′j , p̂k = 0.

From the discussion at the end of Sec. IV, any operator that commutes with the translation
generator must be a function only of the translation generator. Similarly, any operator
that commutes with the position operator must be a function only of the position operator.
Therefore the operator L̂j − L̂′j should be a constant. This leads to the conclusion that
the rotation generator is given by Eq. (20), since a constant term adds only a physically
meaningless phase to the rotation operator. The explicit expression for the rotation generator
(20) gives the commutation relations between the components of the rotation generator
h i
L̂j , L̂k = iθεjkl L̂l , (21)

which show that the rotation generator is also a vector operator.

VI. GALILEAN TRANSFORMATION

We denote the Galilean transformation operator as ÛG (V, t). The position operator r̂ of
a particle is transformed into r̂′ by a Galilean transformation:

r̂ → r̂′ = ÛG† (V, t)r̂ÛG (V, t) = r̂ + Vt. (22)

14
The Galilean transformation operator must be unitary.
It is easily seen from the property of the Galilean transformation that the expression

ÛG (V, t)ÛG (V′ , t) = ÛG (V′ , t)ÛG (V, t) = ÛG (V + V′, t) (23)

is valid. Like the translation operator, therefore, the Galilean transformation operator can
be expressed as  
i
ÛG (V, t) = exp − V · Ĝ(t) (24)
θ
where Ĝ(t) is a self-adjoint vector operator and is called the Galilean transformation gen-
erator. It is also required from Eq. (23) that the relations
h i
Ĝj , Ĝk = 0 (25)

hold. It follows from Eqs. (22), (24), and (25) that

h i
x̂j , Ĝk = iθtδjk .

We therefore obtain, with the help of the commutation relation (9),

h i
x̂j , Ĝk − tp̂k = 0.

This equation gives that

Ĝj = tp̂j + fˆj (r̂), (26)

where f̂(r̂) is a vector function of r̂, since any operator that commutes with the position
operator should be a function only of the position operator. The Galilean transformation
for the translation generator can then be expanded as
∞
X (−i)n hh hh i i i i
ÛG† p̂j ÛG = · · · p̂j , Ĝ , Ĝ , · · · , Ĝ
n=0
n!
∂ fˆk t ∂ 2 fˆk
= p̂j − Vk − Vk Vl +··· , (27)
∂ x̂j 2 ∂ x̂j ∂ x̂l

where we have put

1 1 
Ĝ ≡ V · Ĝ = tV · p̂ + V · f̂ .
θ θ
The transformed translation generator should depend neither on the position operator
nor on time, because of principle of space and time requiring that space and time are

15
homogeneous. Terms of higher than first order in the velocity in the right side of Eq. (27)
should therefore vanish, so that f̂ must a function linear in the position operator:

f̂ (r̂) = −mr̂, (28)

where m is a constant. The operator f̂ can include a constant term, but it can be omitted,
since it only adds a physically meaningless phase to the Galilean transformation operator.
Thus the translation generator is transformed by the Galilean transformation as

ÛG† p̂ÛG = p̂ + mV. (29)

This result might be expected if the translation generator is identified with momentum
operator, and then the constant m can be identified with mass. We, however, have not
identified the translation generator with momentum operator as yet, but shall do in the
next section. It will be shown in Sec. VIII that the constant m be mass of the particle.
Substituting (28) into (26), we obtain the explicit expression for the Galilean transfor-
mation generator:
Ĝ = tp̂ − mr̂ (30)

Usually the Galilean transformation generator is obtained from the transformation rules for
both position and velocity. We cannot adopt such the way since we have not yet defined ve-
locity operator. It is important that the Galilean transformation generator has been obtained
without the help of the transformation rule for velocity. Thus the Galilean transformation
operator is expressed as
   
i i
ÛG (V, t) = exp − V · Ĝ = exp − V · (tp̂ − mr̂) .
θ θ

The operator ÛG can also be rewritten as

mV 2
     
mV Vt
ÛG (V, t) = exp −i t exp i · r̂ exp −i · p̂
2θ θ θ
mV 2
     
Vt mV
= exp i t exp −i · p̂ exp i · r̂ . (31)
2θ θ θ

VII. LAGRANGIAN AND EQUATION OF MOTION FOR A FREE PARTICLE

In this section we shall obtain the explicit expression for the Lagrangian for a free particle
on the basis of the principles introduced in Sec. II. The Lagrangian must be represented by

16
state vectors in the present case, since it should be in general represented by variables that
represent states of a system of interest and states of a microscopic particle are represented
by state vectors. Thus the Lagrangian must be represented in terms of inner products in the
state vector space, since it should be a real c-number. According to the wave principle, the
Lagrangian must be linear in state kets and antilinear in state bras. If it would not be the
case, the equation of motion derived from it would not satisfy the principle of superposition.

A. Requirements of the principle of space and time

The Lagrangian must be independent of time and invariant with respect to space trans-
lation and rotation due to the principle of space and time. First, since the Lagrangian must
be independent of time, we obtain a general expression for the Lagrangian as
 
i d|ψi dhψ|
L = hψ|Ĉ0 |ψi + hψ|Ĉ1 − Ĉ1 |ψi
2 dt dt
dhψ| d|ψi i dhψ| d2 |ψi d2 hψ| d|ψi
 
+ Ĉ2 + Ĉ3 − Ĉ3 +··· (32)
dt dt 2 dt dt2 dt2 dt

where all Ĉn are self-adjoint operators which are independent of time. The above equation
does not involve such terms as, for example,
d|ψi dhψ| d
hψ|Â + Â|ψi = hψ|Â|ψi,
dt dt dt
since the terms that are exact differentials with respect to time do not contribute the equation
of motion.6
Next, we consider the invariance with respect to space translation. Suppose that a state
vector is transformed by a space translation as

|ψi → |ψ ′ i = Û (a)|ψi.

Since the Lagrangian is invariant with respect to this transformation, all the operators that
are involved in the Lagrangian must be unchanged by the transformation:

Û † (a)Ĉn Û (a) = Ĉn ,

which follows that

h i
Ĉn , p̂ = 0.

17
This means that Ĉn are functions only of the translation generator. In similar fashion, it is
shown from space-rotation invariance that Ĉn must commute with the rotation generator:
h i
Ĉn , L̂ = 0.

Thus Ĉn are scalars that are dependent only on the translation generator, so that Ĉn can in
general be expressed as
Ĉn = cn0 + cn2 p̂2 + cn4 (p̂2 )2 + · · · (33)

From Hamilton’s principle we obtain the Lagrangian equations of motion

δL d δL d2 δL
−  E+  E −··· = 0
δ |ψi dt δ ψ̇ dt2 δ ψ̈

δL d δL d2 δL
− D  + 2 D  −··· = 0
δ hψ| dt δ ψ̇  dt δ ψ̈ 

where we have adopted the abbreviation

d d2
|ψ̇i ≡ |ψi, |ψ̈i ≡ |ψi, etc.
dt dt2
Higher order time derivatives can appear in the equations of motion, since the Lagrangian
may involve higher order time derivatives of state vectors.
From the Lagrangian (32) we can obtain the equation of motion
d d2 d3
iĈ1 |ψi − Ĉ2 2 |ψi − iĈ3 3 |ψi − · · · + Ĉ0 |ψi = 0.
dt dt dt
In the interests of simplicity, we rewrite the above equation as
d d|ψi d2 |ψi d3 |ψi
ic10 |ψi = −Ĉ0 |ψi − iĈ1′ + Ĉ2 + iĈ 3 ··· (34)
dt dt dt2 dt3
where c10 is a real constant and we have put

Ĉ1′ ≡ Ĉ1 − c10 = c12 p̂2 + c14 (p̂2 )2 + · · · . (35)

B. Requirements of the probability principle — Derivation of Schrödinger equa-

tion —

In Sec. III it has been shown from the probability principle and the principle of inde-
structibility and increatibility of particles that the wave function must satisfy the equation

18
of continuity (3). We therefore require that the equation of continuity is derived from the
equation of motion for the state vector. The time derivative of the position probability
density of the particle can be rewritten as
∂ ∂ d|ψi dhψ|
|ψ(r)|2 = hψ|rihr|ψi = hψ|rihr| + |rihr|ψi.
∂t ∂t dt dt
Substitution of Eq. (34) into this equation gives
d2 |ψi
 
∂ 2 ′ d|ψi
ic10 |ψ(r)| = hψ|rihr| −Ĉ0 |ψi − iĈ1 + Ĉ2 2 + · · ·
∂t dt dt

dhψ| ′ d2 hψ|
 
− −hψ|Ĉ0 + i Ĉ + Ĉ2 + · · · |rihr|ψi. (36)
dt 1 dt2
This reduces to the equation of continuity, if and only if

Ĉ1′ = 0, and Ĉn = 0 for n ≥ 2

(see Appendix). Because of the requirement that the equation of continuity derives from
the equation of motion for the state vector, we conclude that the equation of motion must
have the form
d
ic10 |ψi = −Ĉ0 |ψi
dt
The Lagrangian, therefore, must be given by
 
ih d|ψi dhψ|
L= hψ| − |ψi − hψ|Ĥ|ψi, (37)
4π dt dt
where we have put
h
Ĥ = −Ĉ0 , = c10
2π
The operator Ĥ is the Hamiltonian. It shall be shown in Sec. VIII that the Hamiltonian is
the operator corresponding to energy. Its explicit form shall be derived below. The quantity
h is now an undetermined constant, which shall be shown to be equal to Planck’s constant
in the next section. As an alternative expression for the Lagrangian, we can take
h d
L = hψ|i − Ĥ|ψi, (38)
2π dt
ih d
which is obtained by adding a exact differential with respect to time, 4π dt
hψ|ψi, to the
expression (37) and is complex rather than real. This expression for the Lagrangian is
simple and convenient, from which we can always derive the same results as those derived
from Eq. (37), so that it will be used henceforth.

19
 Since the Lagrangian involves only the first order time derivatives of the state vector, the
Lagrangian equations of motion reduce to

d δL δL d δL δL
 E= , D = . (39)
dt δ ψ̇ δ |ψi dt δ ψ̇  δ hψ|

The latter gives the equation of motion for the ket in the explicit expression:

h d
i |ψi = Ĥ|ψi. (40)
2π dt

Thus we have derived Schrödinger equation, but have not determined the value of the
constant h and the form and meaning of Ĥ as yet.

C. Galilean invariance — Determination of the more explicit form of Hamilto-

nian —

In this subsection we shall obtain the more explicit form of the Hamiltonian Ĥ from
Galilean invariance.
The state vector is transformed by the Galilean transformation as
 
′ i
|ψi → |ψ i = ÛG (V, t)|ψi = exp − V · (tp̂ − mr̂) |ψi.
θ

The principle of Galilean relativity requires that the Lagrangian be invariant with respect
to Galilean transformation:

ih d
L′ = hψ ′ | − Ĥ|ψ ′ i
2π dt
ih d
= hψ| − Ĥ|ψi = L,
2π dt

so that the following equation must hold:

ih † dÛG (V, t)
ÛG (V, t) − ÛG† (V, t) Ĥ ÛG (V, t) = −Ĥ.
2π dt

Since Eqs. (31) and (29) give the equations

 
ih † dÛG (V, t) h 1 2
ÛG (V, t) = mV + V · p̂
2π dt 2πθ 2

ÛG† (V, t)Ĥ(p̂)ÛG (V, t) = Ĥ(p̂ + mV),

20
the Lagrangian is invariant with respect to Galilean transformation if
 
h 1 2
Ĥ(p̂ + mV) = Ĥ(p̂) + mV + V · p̂ .
2πθ 2

This means that the Hamiltonian must have the form

h
Ĥ = p̂2 . (41)
4πθm

The Hamiltonian can include a real constant term, but it is neglected since it is not physically
meaningful.
The Lagrangian and the Hamiltonian include three undetermined constants, h, θ, and m.
They will be determined bellow.

VIII. SYMMETRIES AND CONSERVATIVE QUANTITIES

In this section, on the basis of conservation laws of energy, momentum, and angular
momentum that are derived from principle of space and time, we shall show that the Hamil-
tonian is the operator corresponding to energy and that the translation generator can be
identified with momentum operator. We shall also prove that h is Planck’s constant and m
is the mass of a particle.

A. Homogeneity of time and conservation law of energy

Now we discuss the conservation law of energy that is derived from homogeneity of time,
which is required from principle of space and time. The time derivative of the Lagrangian
can be expressed as  
dL ∂L d δL δL
= + |ψ̇i + hψ̇| ,
dt ∂t dt δ|ψ̇i δhψ̇|
with the help of the Lagrangian equations of motion (39). It follows from the above equation
that  
∂L d δL δL
=− |ψ̇i + hψ̇| −L =0
∂t dt δ|ψ̇i δhψ̇|
since the Lagrangian does not depend on time explicitly because of homogeneity of time.
This equation means that the quantity in the parentheses on the right side is a constant of
the motion. This conserved quantity, which is energy since it is derived from homogeneity

21
of time, is called the Hamiltonian in classical mechanics and is denoted by H. Then the
classical Hamiltonian is expressed explicitly from Eq. (38) as
δL δL
H= |ψ̇i + hψ̇| − L = hψ|Ĥ|ψi.
δ|ψ̇i δhψ̇|

This shows that the Hamiltonian Ĥ is the operator corresponding to energy, as it should be.
If a state vector |ψ(t)i is an eigenket of Ĥ with an eigenvalue E, then from Eq. (40) we
obtain
h d
i |ψ(t)i = E|ψ(t)i,
2π dt
the solution of which is
|ψ(t)i = e−i2πEt/h |ψ(0)i.

The state vector is a periodic function of time and the relation between its frequency ν and
the energy E is given by
E = hν,

which is one of Einstein-de Broglie formulas. This shows that the constant h should be
Planck’s constant.

B. Homogeneity and isotropy of space and conservation laws of momentum and

angular momentum

Now we discuss the conservation law of momentum that is derived from homogeneity
of space, which is required from principle of space and time. The conservation law of
momentum that is valid for any branch of physics is one of fundamental laws in physics.
Momentum must therefore be defined on the common basis of physics. We thus define
momentum as the conserved additive vector that is derived from homogeneity of space or
translational symmetry.
We suppose that a translation by a transforms |ψi into |ψ ′ i:

|ψi → |ψ ′ i = Û (a) |ψi = e−ia·p̂/θ |ψi ,

where Û (a) is given by Eq. (7). Since the Lagrangian L′ , which is the Lagrangian given by
|ψ ′ i, is independent of a because of homogeneity of space:
∂L′ 

= 0. (42)
∂a a=0

22
On the other hand we obtain

∂L′  δL ∂|ψ ′ i  δL ∂|ψ̇ ′ i 
 
= +
∂a a=0 δ|ψi ∂a a=0 δ|ψ̇i ∂a 

a=0

∂hψ ′ |  ∂hψ̇ ′ | 

δL δL
+ +
∂a a=0 δhψ| ∂a 

a=0
δhψ̇|

d δL ∂ |ψ ′ i  δL ∂ hψ ′ | 
   
= +
dt δ|ψ̇i ∂a a=0 δhψ̇| ∂a a=0

h d
= hψ|p̂|ψi, (43)
2πθ dt

where use has been made of Eqs. (39) and (38). It follows from Eqs. (42) and (43) that the
translation generator is conserved under translational symmetry. Therefore the translation
generator should be proportional to momentum operator.
Next, we consider a system consisting of many particles which need not be identical. It
is clear that in this system the translation generator for each particle can be defined as

[x̂αj , p̂βk ] = iθδαβ δjk , [p̂αj , p̂βk ] = 0,

where Greek subscripts denote particles and alphabetic subscripts denote the components of
vectors. The Hamiltonian for this system should be given as a function of translation gen-
erators and position operators of those particles: Ĥ ({p̂α } ; {r̂α }). The translation operator
that translates the system by a is given by
" #
a X
Û (a) = exp −i · p̂α .
θ α

This shows that translation generators are additive, which is also required for momenta, if
and only if the constant θ involved in the translation operator is a universal constant. We
require from the principle of space and time that the system has translational symmetry;

Û † (a) Ĥ ({p̂α } ; {r̂α }) Û (a) = Ĥ ({p̂α } ; {r̂α + a}) = Ĥ ({p̂α } ; {r̂α }) .

Then we can obtain the equation

d X
hψ| p̂α |ψi = 0,
dt α

23
like Eq. (43) for a free particle. Therefore even if there exist interactions between particles,
the sum of translation generators is conserved. Since the conserved quantity that is derived
from translational symmetry is (total) momentum, the sum of all translation generators
must be proportional to the total momentum operator. We therefore identify translation
generators with momentum operators from now on, the condition for which will be shown
to be θ = h/2π in Sec. IX.
In a similar way, it is shown that the rotation generator is proportional to the angular
momentum operator that is the conserved quantity derived from the isotropy of space,
which is also required from the principle of space and time. It is apparent that the rotation
generator is identical with the angular momentum operator if the translation generator is
identical with the momentum operator.

C. Quantum-mechanical definition of a conserved quantity

The above discussion has shown that the conservation law in quantum mechanics is
expressed such that the expectation value of the operator corresponding to a conserved
dynamical variable is independent of time for any time-dependent state vector which is a
solution of Schrödinger equation; if
d d
hÂi = hψ|Â|ψi = 0, (44)
dt dt
where |ψi is any time-dependent state vector that describes the considered system, then the
dynamical variable A is conserved. This is important because it should be characteristic
of a physical system and independent of its state that a certain dynamical variable of the
system is conserved.
With the help of Schrödinger equation (40), the time derivative of the expectation value
of any operator  is given by
d d ∂ Â 2π h i
hÂi = hψ|Â|ψi = hψ| |ψi + hψ| Â, Ĥ |ψi. (45)
dt dt ∂t ih
Thus the definition of a conserved quantity (44) can be rewritten as
h i
Â, Ĥ = 0,

if  does not depend explicitly on time. This is an alternative expression of the definition of
the conserved quantity in quantum mechanics. Invariance of the Hamiltonian with respect

24
to translation and rotation gives the relations
h i
p̂, Ĥ = 0,
h i
L̂, Ĥ = 0,

which indicate that the momentum and angular momentum are constants of the motion.

D. Velocity operator and mass

Since we deal with an object that is classically considered to be a particle, we naturally

introduce mass as its attribute. In classical mechanics, mass is usually introduced as a
proportionality constant between force and acceleration. However, we cannot adopt this
definition of mass, since the concepts of acceleration and force do not have definite meanings
in quantum mechanics. It is reasonable to define the mass of a particle as the proportionality
constant between its momentum and velocity operators7 , since the momentum operator is
proportional to the velocity operator in quantum mechanics as well as in classical mechanics,
as shall be shown sooner.
It is reasonable that we define the velocity operator as the operator that has the expec-
tation value equal to the time derivative of the expectation value of the position operator:
d
hv̂i ≡ hr̂i.
dt
The definition of the velocity operator can be rewritten as
2π h i
v̂ ≡ r̂, Ĥ ,
ih
since the time derivative of the expectation value is generally given by Eq. (45). Substitution
of Eq. (41) into the definition of the velocity operator gives

p̂ = mv̂

with the help of the commutation relations (9). We have shown that the velocity and
momentum operators, which are defined independently of each other, are proportional to
each other. From our definition of mass, it has verified that the constant m, which was
introduced in the discussion of Galilean transformation of Sec. VI, is the mass of the particle.
From Eq. (29), therefore, we obtain the Galilean transformation for the velocity operator:

v̂ → v̂′ = ÛG† v̂ÛG = v̂ + V,

25
as should be.

IX. A SYSTEM CONSISTING OF TWO INTERACTING PARTICLES AND A

PARTICLE IN AN EXTERNAL FIELD

If there exists an external field, Galilean principle of relativity and homogeneity of space
are broken, and further homogeneity of time is also broken if the external field depends on
time. This fact means that it is difficult to obtain the properties of the external field on
the basis of the principles. The motion in an external field, however, can be regarded as
the relative motion of two particles that interact with each other. Thus in this section we
consider a system that consists of two interacting particles; the six principles should naturally
hold for the system, including Galilean principle of relativity and principle of space and time.
We investigate to what extent the properties of the interaction between two particles are
restricted from the principles. Then we obtain the Lagrangian for a particle in an external
field as that for the relative motion in the two-particle system. Furthermore, by applying
the conservation law of momentum to the system that consists of a quantum particle and a
classical particle, we determine the last undetermined constant θ. Thus we shall obtain the
final forms of the Lagrangian and the Hamiltonian for a particle in an external field, and
then we shall finally attain to the true Schrödinger equation.

A. A system consisting of two free particles

The Lagrangian for two free particles, L1+2 , is given from additivity of Lagrangians by

ih1 d ih2 d
L1+2 = L1 + L2 = hψ1 | − Ĥ1 |ψ1 i + hψ2 | − Ĥ2 |ψ2 i ,
2π dt 2π dt

where the Hamiltonian for the k-th particle Ĥk is expressed as

hk
Ĥk = p̂2 ,
4πθmk k

and mk and hk are the mass and Planck’ constant of the k-th particle, respectively. We have
assumed that Planck’s constant might not be a universal constant. A state vector |ψ1+2 i for
the system consisting of two free particles should be given by a direct product of each state
vector for two particles, |ψ1 i ⊗ |ψ2 i. The Lagrangian L1+2 should therefore be expressed in

26
terms of |ψ1 i ⊗ |ψ2 i:

ih1 d|ψ1 i ih2 d|ψ2 i

L1+2 = hψ1 | ⊗ hψ2 | · 2π dt
⊗ |ψ2 i + hψ1 | ⊗ hψ2 | · |ψ1 i ⊗ 2π dt
(46)
− hψ1 | ⊗ hψ2 | Ĥ1 + Ĥ2 |ψ1 i ⊗ |ψ2 i

For a system that consists of two interacting particles, however, it is evident that its state
vector cannot be expressed as a direct product. Thus the Lagrangian for two interacting
particles should be expressed using a state vector |ψ1+2 i, which requires

h1 = h2 .

If it would not be the case, the time derivatives in Eq. (46) could not be expressed by |ψ1+2 i
and hψ1+2 |. It has been verified that Planck’s constant should be a universal constant. Thus
the Lagrangian for two noninteracting particles is given by

ih d
L1+2 = hψ1+2 | − Ĥ1 − Ĥ2 |ψ1+2 i ,
2π dt

where the Hamiltonian Ĥk is expressed as

h
Ĥk = p̂2 .
4πθmk k

B. Property of a two-particle wave function

In order to investigate the physically permitted form of the Lagrangian for two interacting
particles we consider the property of a two-particle wave function that is required to satisfy
the equation of continuity for the position probability density of each particle which is
derived from the two-particle wave function.
The two-particle wave function ψ1+2 (r1 , r2 ) is given by

ψ1+2 (r1 , r2 ) = hr1 , r2 | ψ1+2 i .

The quantity
|ψ1+2 (r1 , r2 )|2 dV1 dV2

gives the probability of finding particle 1 in a volume element dV1 about r1 and particle 2 in
a volume element dV2 about r2 . Integration of it over r2 (r1 ) gives the probability of finding

27
particle 1 (2). Thus if we integrate the equation for the time derivative of the square of the
magnitude of the two-particle wave function

∂
|ψ1+2 (r1 , r2 )|2 = F ({ψ1+2 (r1 , r2 )}) (47)
∂t

over r2 (r1 ), we should obtain the equation of continuity for particle 1 (2). This requires
that the equation (47) is of the form

∂ 2 ∂2
|ψ1+2 (r1, r2 )| = −∇1 · J1 − ∇2 · J2 − Tkl . (48)
∂t ∂x1k ∂x2l

We call this equation the equation of continuity for two particles expedientially.

C. Interacting two particles

The Lagrangian for a system that consists of two interacting particles can be expressed
in general as
ih d
L1+2 = hψ1+2 | − Ĥ1 − Ĥ2 − V̂ |ψ1+2 i .
2π dt
We assume that the interaction V̂ has the form
 
d
V̂ = V̂ , r̂1 − r̂2 , p̂1 , p̂2
dt

so that it satisfies homogeneity of space and time.

We now investigate whether the interaction is permitted to involve time differential op-
erators. If the interaction involves second-order or higher time differential operators, or the
first-order time differential operator with a factor that depends on r̂1 − r̂2 , p̂1 or p̂2 , then
Eq. (48) does not hold. This is easily verified by replacing |ψi with |ψ1+2 i in the discussion
in Appendix. If the interaction involves first-order time derivative with a constant factor,
it effects changing of Planck’s constant. This is not permitted. If it would be the case, a
three-particle system that consists of two-interacting particles and a free particle would not
satisfy the additivity of Lagrangians. Thus the interaction must not involve time differential
operators.
We next consider invariance under Galilean transformation. The Galilean transformation
operator for a two-particle system is given by
 
i
ÛG (v, t) = exp − v · {(tp̂1 − m1 r̂1 ) + (tp̂2 − m2 r̂2 )} .
θ

28
The requirement that the Lagrangian be invariant under the Galilean transformation gives
the equation
ÛG† (v, t) V̂ ÛG (v, t) = V̂ ,

from which the interaction becomes of the form

 
p̂1 p̂2
V̂ = V̂ r̂1 − r̂2 , − .
m1 m2

Then we can write the interaction as in the following:

V̂ = V̂0 (r̂1 − r̂2 )

 2 " 2 #2
p̂1 p̂2 p̂1 p̂2
+ a20 − + c40 − +···
m1 m2 m1 m2
     
p̂1 p̂2 p̂1 p̂2
+ a11 (r̂1 − r̂2 ) · − + − · (r̂1 − r̂2 )
m1 m2 m1 m2
     2
p̂1 p̂2 p̂1 p̂2
+ a22 (r̂1 − r̂2 ) · − + − · (r̂1 − r̂2 ) +··· ,
m1 m2 m1 m2

where V̂0 is a self-adjoint scalar operator that is a function only of r̂1 −r̂2 . Explicit calculation
shows that the terms in the second and third lines in the above equation are compatible
with the equation of continuity for two particles. Dependence of interaction on momenta
is generally permitted. Thus it has been verified that the Hamiltonian for a two-particle
system is generally given by
 
h h p̂1 p̂2
Ĥ1+2 = Ĥ1 + Ĥ2 + V̂ = p̂21 + p̂2 + V̂ r̂1 − r̂2 , − . (49)
4πθm1 4πθm2 2 m1 m2

D. Lagrangian for an external field

We now derive the Lagrangian for a particle in an external field as the Lagrangian for the
relative motion of two interacting particles. For that purpose we divide six internal degrees
of freedom into two parts, those for the relative motion and those for the center-of-mass
motion.
The degrees of freedom with respect to the relative motion are specified by the relative
position operator
r̂ = r̂1 − r̂1

29
and the momentum operator p̂ canonically conjugate to r̂. The momentum operator p̂
should satisfy the equation
   
i i
exp a · p̂ r̂ exp − a · p̂ = r̂ + a. (50)
θ θ

We require also that p̂ is invariant under Galilean transformation. From this requirement
and with the help of Eq. (50) we obtain

m2 m1
p̂ = p̂1 − p̂2 .
m1 + m2 m1 + m2

The degrees of freedom with respect to the center-of-mass motion are specified by the
sum of momenta for two particles P̂ = p̂1 + p̂2 and the position operator R̂ canonically
conjugate to P̂. The position operator R̂ should satisfy the equations
   
i i
exp a · P̂ R̂ exp − a · P̂ = R̂ + a
θ θ

and
h i
R̂, p̂ = 0.

The latter is necessary for the center-of-mass motion to be independent of the relative
motion. These lead to
m1 m2
R̂ = r̂1 + r̂2 .
m1 + m2 m1 + m2
These operators r̂, p̂, R̂, and P̂ satisfy the following commutation relations:
h i
X̂j , P̂k = iθδjk , [x̂j , p̂k ] = iθδjk
h i h i
X̂j , p̂k = 0, x̂j , P̂k = 0

The original operators are expressed in terms of the new operators as

m2 m1
r̂1 = R̂ + r̂, r̂2 = R̂ − r̂ (51)
m1 + m2 m1 + m2

and
m1 m2
p̂1 = P̂ + p̂, p̂2 = P̂ − p̂. (52)
m1 + m2 m1 + m2
Substitution of Eqs. (51) and (52) into Eq. (49) gives

h h
Ĥ1+2 = Ĥc + Ĥr = P̂2 + p̂2 + V̂ (r̂, p̂) ,
4πθM 4πθm

30
where we have put
m1 m2
M = m1 + m2 , m= ,
m1 + m2
and Ĥc and Ĥr are the Hamiltonians for the center-of-mass motion and the relative motion,
respectively. It is important that the interaction depends only on the relative motion. We
can express a two-particle state vector |ψ1+2 i as a direct product of a state vector for the
center-of-mass motion |Ψi and that for the relative motion |ψi :

|ψ1+2 i = |Ψi ⊗ |ψi .

Then the Lagrangian can be rewritten as

 
ih d
L1+2 = hΨ| ⊗ hψ| − Ĥ1+2 |Ψi ⊗ |ψi
2π dt
ih d ih d
= hΨ| − Ĥc |Ψi + hψ| − Ĥr |ψi
2π dt 2π dt
= Lc + Lr .

The two-particle Lagrangian has been divided into the Lagrangians for the relative motion
and the center-of-mass motion. When we regard the Lagrangian for the relative motion as
the Lagrangian for a particle in an external field, the latter is given by
ih d
L = hψ| − Ĥ |ψi ,
2π dt
where Ĥ is the Hamiltonian for a particle in an external field
h
Ĥ = p̂2 + V̂ (r̂, p̂) . (53)
4πθm
Thus an external field can generally depend on the momentum. This is reasonable, since the
Hamiltonian for a charged particle in an electromagnetic field has such a term. The Hamil-
tonian for a free particle is proportional to p̂2 because the Lagrangian should be invariant
under Galilean transformation. However, it is not required for a particle in an external field
that its Lagrangian is invariant under Galilean transformation, so that dependence of the
interaction on momentum is not restricted strictly.

E. Condition that translation generator is identical with momentum

In Sec. VIII we have verified that the translation generator is proportional to the mo-
mentum operator and have identified the former with the latter. Now we determine the

31
constant θ which is included in the transformation operators so that the translation gen-
erator is identical with the momentum operator, which also means from Eq. (20) that the
rotation generator is identical with the angular momentum operator. Since momentum is
defined quantitatively in classical mechanics, we should define the momentum operator in
quantum mechanics consistently with the momentum in classical mechanics. Therefore we
shall introduce a system that consists of two interacting particles, one of which is a classical
particle and the other a quantum particle. In the system we determine the constant θ so
that the sum of both momenta is conserved.
Suppose that a classical particle of mass M interacts with a quantum particle of mass m.
The interaction between these particles is naturally assumed to be given by

V (R − r̂),

where R is the coordinate of the classical particle and r̂ is the position operator of the
quantum particle8 . Then the Lagrangian for the two-particle system is given by
 
1 2 h d h 2
L = M Ṙ + hψ| i − p̂ − V (R − r̂) |ψi,
2 2π dt 4πθm

from which the Lagrangian equations of motion are derived:

 
h d h 2
i |ψi = p̂ + V (R − r̂) |ψi (54)
2π dt 4πθm

M R̈ = −∇R hψ|V (R − r̂)|ψi. (55)

The analytical-mechanical Hamiltonian H is obtained as

d
H = P · Ṙ + hΠ| |ψi − L
dt
P2
 
2π h 2
= + hΠ| p̂ + V (R − r̂) |ψi,
2M ih 4πθm

where hΠ| and P are the momenta canonically conjugate to |ψi and R, respectively:

δL h
hΠ| ≡ = i hψ|, (56)
δ (d|ψi/dt) 2π

∂L
P≡ = M Ṙ. (57)
∂ Ṙ

32
The time derivative of the analytical-mechanical Hamiltonian is given, with the help of
Eqs. (54), (55), (56) and (57), by
dH P
= · Ṗ + Ṙ · ∇R hψ|V (R − r̂)|ψi = 0,
dt M
which shows that the energy is conserved, as is expected. This also confirms that the
quantum Hamiltonian Ĥ is the operator corresponding to energy.
Next we consider the conservation law of momentum. In discussing the conservation law
for a quantum particle, we must investigate the expectation value of its momentum. The
conservation law of momentum is therefore expressed as
 
d 2π h 2
(P + hp̂i) = −∇R hV (R − r̂)i + p̂, p̂ + V (R − r̂)
dt ih 4πθm
 
2πθ
= − 1 ∇R hV (R − r̂)i
h
= 0,

with the help of Eqs. (55),(45), and (10). This equation shows that the translation generator
is identical with the momentum operator and the total momentum is conserved, if and only
if we put
h
θ= . (58)
2π
The constant θ must be a universal constant in order that momenta are additive. Therefore
Eq. (58) shows again that Planck’s constant must also be a universal constant.
From Eq. (58), the commutation relations (9) between the position and momentum op-
erators become the canonical commutation relations

[x̂j , p̂k ] = i~δjk , (59)

as should be, where we have put

h
~≡ .
2π
Thus the full set of the canonical commutation relations (2), (8), and (59) have derived from
the principles of quantum mechanics. Furthermore the commutation relations between the
components of angular momentum and between angular momentum and any vector operator
 become
h i h i
L̂j , L̂k = i~εjklL̂l , L̂j , Âk = i~εjklÂl ,

33
from Eqs. (19) and (21).
Thus substitution of Eq. (58) into Eq. (53) gives the final form of the Hamiltonian

p̂2
Ĥ = + V (r̂, p̂),
2m

which has the same form as that in classical mechanics. We have finally obtained the
Schrödinger equation:
 2 
d p̂
i~ |ψi = Ĥ|ψi = + V (r̂, p̂) |ψi.
dt 2m

It is crucially important to note that we have derived the Hamiltonian and the Schrödinger
equation without being premised on classical mechanics, except for the definitions of dynam-
ical variables.

X. CONCLUSION

We have deductively established quantum mechanics on the basis of the six fundamental
principles. We now summarize several essential points.
First, it is most important to note that the canonical commutation relations

[x̂j , p̂k ] = i~δjk , [x̂j , x̂k ] = 0, [p̂j , p̂k ] = 0

have been deduced from the principles.

We have defined momentum and angular momentum as the constants of motion that are
derived from homogeneity and isotropy of space, respectively. The properties of momentum
and angular momentum operators, which include the commutation relations between them,
are necessarily established from this definition. It is therefore necessary that the canonical
commutation relations hold.
Second, the Schrödinger equation that describes the time evolution of a state vector has
necessarily been derived in the definitive form from the principles. Moreover, it has been
verified that the operator that appears on the right side of the Schrödinger equation be
the Hamiltonian or the operator corresponding to energy, since energy is the constant of
motion that is derived from homogeneity of time. Furthermore, it has been shown that
Planck’s constant should appear on the left side of the equation by corresponding it to one
of Einstein-de Broglie formulas.

34
 We have shown that it is also necessary in the nonrelativistic theory that both of the
Schrödinger equation and the canonical commutation relations include Planck’s constant9 .
Planck’s constant has been introduced into the Schrödinger equation in accordance with one
of Einstein-de Broglie formulas, E = hν, whereas Planck’s constant in the canonical com-
mutation relations has been introduced from the requirement that the translation generator
is identical with momentum. It has also been derived both from the additivity of momenta
and from that of Lagrangians that Planck’s constant should be a universal constant.
Third, it have been shown that relations between dynamical variables in quantum me-
chanics have the same forms for those in classical mechanics such as the expressions for
angular momentum L̂ = r̂ × p̂, Hamiltonian Ĥ = p̂2 /2m + V̂ , and the relation between
momentum and velocity operators p̂ = mv̂. It is important to note that we have derived
them independently of classical mechanics. It is usually assumed that relations between
dynamical variables in classical mechanics are also preserved in quantum mechanics. This,
however, is not self-evident. For example, the momentum operator has no direct relation to
the velocity operator in relativistic quantum mechanics.
As a consequence, the process of construction of quantum mechanics on the basis of the
fundamental principles shows that quantum mechanics is the subtle and inevitable theory.
For example, quantum mechanics does not permit any other commutation relation except
for the canonical commutation relations. The natural law is not selected arbitrarily but
determined necessarily.
There are several quantization methods such as canonical, path integral, and random
process. All of them are premised on the classical theory and are methods for deriving
the quantum theory from the classical theory by quantizing it. The present formalism is
not premised on the classical theory but is based on the fundamental principles that are
independent of the classical theory. The rigorous quantum theory must not be premised on
the classical theory that is approximate to the quantum theory.

APPENDIX: THE CONDITION THAT THE EQUATION OF CONTINUITY IS

DERIVED FROM THE EQUATION OF MOTION

We now find the condition such that the right side of Eq. (36) can be expressed as the
divergence of a vector. The terms that do not include time derivatives on the right side of

35
Eq. (36) can be rewritten as

−hψ|rihr|Ĉ0|ψi + hψ|Ĉ0|rihr|ψi
h i
2
= −hψ|rihr| c00 + c02 p̂2 + c04 (p̂2 ) + · · · |ψi
h i
2 2 2
+hψ| c00 + c02 p̂ + c04 (p̂ ) + · · · |rihr|ψi
h n o i
2 2 4 2 2
= −hψ|ri −c02 θ {∇ hr|} + c04 θ (∇ ) hr| + · · · |ψi
h n o i
2
+hψ| −c02 θ2 {∇2 |ri} + c04 θ4 (∇2 ) |ri + · · · hr|ψi

= c02 θ2 [ψ ∗ (r) {∇2 ψ(r)} − {∇2 ψ ∗ (r)} ψ(r)]

h n o n o i
2 2
−c04 θ4 ψ ∗ (r) (∇2 ) ψ(r) − (∇2 ) ψ ∗ (r) ψ(r) − · · · ,
where use has been made of Eqs. (33), (12), and (11).
It can easily be seen that the equation

ψ ∗ (∇2 )n ψ − (∇2 )n ψ ∗ ψ = ∇ · J2n

 

is generally valid, where

h n o n o i
n−1 n−1 ∗
J2n = ψ ∗ ∇ (∇2 ) ψ − ∇ (∇2 ) ψ ψ
h n o n o i
∗ 2 n−1 2 n−1 ∗
− (∇ψ ) (∇ ) ψ − (∇ ) ψ (∇ψ)
h n o n o i
n−2 n−2 ∗
+ (∇2 ψ ∗ ) ∇ (∇2 ) ψ − ∇ (∇2 ) ψ (∇2 ψ)

−···
hn on o n on oi
n/2−1 ∗ n/2 n/2 n/2−1
− ∇ (∇2 ) ψ (∇2 ) ψ − (∇2 ) ψ ∗ ∇ (∇2 ) ψ

for even n, and

h n o n o i
∗ 2 n−1 2 n−1 ∗
J2n = ψ ∇ (∇ ) ψ − ∇ (∇ ) ψ ψ
h n o n o i
n−1 n−1 ∗
− (∇ψ ∗ ) ∇ (∇2 ) ψ − ∇ (∇2 ) ψ (∇ψ)
h n o n o i
n−2 n−2 ∗
+ (∇2 ψ ∗ ) ∇ (∇2 ) ψ − ∇ (∇2 ) ψ (∇2 ψ)

−···
hn n−1
on n−1
o n n−1
on n−1
oi
2 2 ∗ 2 2 2 2 ∗ 2 2
+ (∇ ) ψ ∇ (∇ ) ψ − ∇ (∇ ) ψ (∇ ) ψ

for odd n.

36
 Thus we obtain

− hψ | ri hr| Ĉ0 |ψi + hψ| Ĉ0 |ri hr | ψi = ∇ · c02 θ2 J2 − c04 θ4 J4 + · · · ,

 

which can be expressed as the divergence of a vector.

The terms with a first-order derivative with respect to time on the right side of Eq. (36)
are explicitly expressed, with the help of Eqs. (35), (12), and (11), as
d|ψi dhψ| ′
−ihψ|rihr|Ĉ1′ −i Ĉ |rihr|ψi
dt dt 1
d|ψi
= −ihψ|ri [−c12 θ2 {∇2 hr|} + c14 θ4 {∇4 hr|} + · · · ]
dt
dhψ| 
−c12 θ2 ∇2 |ri + c14 θ4 ∇4 |ri + · · · hr|ψi
  
−i
dt
∗
     
2 ∗ 2 ∂ψ 2 ∂ψ
= ic12 θ ψ ∇ + ∇ ψ +··· ,
∂t ∂t
which cannot be expressed as the divergence of a vector. Therefore

Ĉ1′ = 0,

in order that Eq. (36) reduces to the equation of continuity.

In the similar way,
Ĉn = 0, n ≥ 2.

∗ Electronic address: esakai@sci.u-ryukyu.ac.jp

1 See L.D. Landau and E.M. Lifshitz, “Mechanics” (Course of Theoretical Physics, Vol. 1) 3rd ed.
(Pergamon Press, New York, 1976), Chap.1.
2 See L.D. Landau and E.M. Lifshitz, “The Classical Theory of Field” (Course of Theoretical
Physics, Vol. 2) 4th ed. (Pergamon Press, New York, 1975), Chap.2.
3 See L.D. Landau and E.M. Lifshitz, “The Classical Theory of Field” (Course of Theoretical
Physics, Vol. 2) 4th ed. (Pergamon Press, New York, 1975), Chap.4.
4 See L.D. Landau and E.M. Lifshitz, “Quantum Mechanics” (Course of Theoretical Physics, Vol.
3) 3rd ed. (Pergamon Press, New York, 1975), Sec.1.
5 See P.A.M. Dirac, “The Principles of Quantum Mechanics” 4th ed. (Oxford University Press,
Oxford, 1958), Sec.6.

37
6 See L.D. Landau and E.M. Lifshitz, “Mechanics” (Course of Theoretical Physics, Vol. 1) 3rd ed.
(Pergamon Press, New York, 1976), Sec.2.
7 Both of mass and force are defined by equation of motion in classical mechanics, so there is an
ambiguity. Instead, it is valid to define mass as a proportionality between the momentum that
is a constant of motion derived from homogeneity of space and the velocity that is obviously
defined as the time derivative of the position.
8 The interaction can be arbitrarily chosen so long as it does not break translational symmetry
and is physically reasonable, since it is introduced only to determine the constant θ.
9 This is trivial in the relativistic theory, but not in the nonrelativistic theory.

38