space

arXiv:2003.08657v1 [quant-ph] 19 Mar 2020

DISTRIBUTION OF QUANTUM ENTANGLEMENT:

PRINCIPLES AND APPLICATIONS

I WAYAN GEDE TANJUNG KRISNANDA

SCHOOL OF PHYSICAL AND MATHEMATICAL

SCIENCES

2020
space

DISTRIBUTION OF QUANTUM ENTANGLEMENT:

PRINCIPLES AND APPLICATIONS

I WAYAN GEDE TANJUNG KRISNANDA

SCHOOL OF PHYSICAL AND MATHEMATICAL SCIENCES

A thesis submitted to the Nanyang Technological

University in partial fulfilment of the requirement for the
degree of Doctor of Philosophy

2020
Abstract

Quantum entanglement is a form of correlation between quantum particles that

has now become a crucial part in quantum information and communication science.
For example, it has been shown to enable or enhance quantum processing tasks
such as quantum cryptography, quantum teleportation, and quantum computing.
However, quantum entanglement is prone to decoherence as a result of interactions
with environmental scatterers, making it an expensive resource. Therefore, it is
crucial to understand its creation.
We centre our attention to a situation where one would like to distribute quantum
entanglement between principal particles that are apart. In this case, it is necessary
to use ancillary systems that are communicated between them or interact with
them continuously. Cubitt et al. showed that the ancillary systems need not be
entangled with the principal particles in order to distribute entanglement. This has
been demonstrated experimentally in the case of communicated ancillary particles
and it is now known that the bound on the distributed entanglement is given by a
communicated quantum discord. On the other hand, little is understood about the
setting with continuous interactions, despite its abundant occurrence in nature.
The main focus of this thesis is to study the distribution of quantum entangle-
ment via continuous interactions with ancillary particles, which I will call mediators.
First, basic concepts and tools that are helpful for this thesis will be introduced.
This includes the description of quantum objects within the framework of quantum
mechanics, their dynamics, and important properties. Next, I will present my work
regarding the necessary conditions for entanglement distribution, the factors that are
relevant for the distributed amount, and the speed limit to achieving maximum en-
tanglement gain. Finally, I present some notable applications that can benefit from
our work. This includes, among others, indirect probing of the quantum nature of
optomechanical mirrors, photosynthetic organisms, and gravitational interactions.

1
Dedication

To my mother and stepfather

Acknowledgements

I would like to give a gazillion thanks to my supervisor A/Prof. Paterek for all his
thoughts and guidance. I really appreciate the time we spent discussing and arguing
throughout my PhD programme. I also thank my co-supervisor Dr. Zuppardo for
helping me to get started in this research. Special thanks to Prof. Paternostro for
his advice and collaboration for most parts of our projects.
I thank my current supervisor Prof. Dumke and also Prof. Brukner for taking
interest in our projects and approving my transitions.
Many thanks to our colleagues Mr. Ganardi, Dr. Lee, Prof. Kim, Dr. Mar-
letto, Prof. Vedral, Dr. Noh, A/Prof. Streltsov, Prof. Bandyopadhyay, Dr. Banerjee,
Dr. Deb, Dr. Halder, Dr. Modi, Prof. Sengupta, Mr. Pal, Ms. Batra, and Prof. Ma-
hesh for their collaboration.
I thank Prof. Winter, Dr. Coles, and Prof. Gröblacher for stimulating discus-
sions, which improved our articles.
I wish to express my gratitude to Prof. Laskowski for hospitality at the Univer-
sity of Gdańsk, Dr. Dutta for hospitality and collaboration at the Korea Institute
for Advanced Study, and Prof. Liew for hospitality at Nanyang Technological Uni-
versity.
I had the opportunity to co-supervise undergraduate students who were keen in
doing research. I would like to thank Guo Yao Tham, Wen Yu Kon, Koo Sui Ho
Edmund, Parth Patel, Swetha Sridhar, Jeremy Goh Swee Kang, Chee Mun Yin, and
Tan Xue Yi for their contributions to our projects.
I thank my colleagues Dr. Miller, Dr. Lake, Minh Tran, Kai Sheng, Zhao Zhuo,
Liu Zheng, Lamia Varawala, and Ankit Kumar for the time we spent together.
Thank you to my sister, my mother, my stepfather, and my friends Andhita,
David, Hendra, Wiswa, and Zhonglin for their company. I also thank Mary and
Steven for their hospitality. Towards the end of my PhD, there is a special person
who added new colours in my life. I would like to express my gratefulness to my
girlfriend Qiannan Zhang.
I wish to acknowledge the funding support for our projects by the National
Research Foundation Singapore and Singapore Ministry of Education Academic Re-
search Fund Tier 2 Project No. MOE2015-T2-2-034.

3
Contents

Abstract 1

Acknowledgements 3

Table of contents 5

Abbreviations 9

List of symbols 11

1 Introduction 17
1.1 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
1.1.1 Direct interactions . . . . . . . . . . . . . . . . . . . . . . . . 18
1.1.2 Indirect interactions . . . . . . . . . . . . . . . . . . . . . . . 19
1.2 Objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
1.3 Thesis organisation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20

2 Quantum objects, their dynamics, and properties 23

2.1 Quantum objects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
2.1.1 Discrete systems . . . . . . . . . . . . . . . . . . . . . . . . . 24
2.1.2 Continuous variable systems . . . . . . . . . . . . . . . . . . . 25
2.1.3 Quantum dynamics . . . . . . . . . . . . . . . . . . . . . . . . 27
2.2 Properties of quantum states . . . . . . . . . . . . . . . . . . . . . . . 28
2.2.1 von Neumann entropy . . . . . . . . . . . . . . . . . . . . . . 28
2.2.2 Purity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
2.2.3 Fidelity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
2.3 Correlations between quantum states . . . . . . . . . . . . . . . . . . 30
2.3.1 Quantum entanglement . . . . . . . . . . . . . . . . . . . . . . 30
2.3.2 Quantum discord . . . . . . . . . . . . . . . . . . . . . . . . . 32
2.3.3 Mutual information . . . . . . . . . . . . . . . . . . . . . . . . 35

5
3 Revealing non-classicality of inaccessible objects 37
3.1 Motivation and objectives . . . . . . . . . . . . . . . . . . . . . . . . 38
3.2 Instrumental discord . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
3.3 Quantum discord is a resource for entanglement gain . . . . . . . . . 42
3.3.1 The role of correlated mediator . . . . . . . . . . . . . . . . . 46
3.4 Entanglement localisation via classical mediators . . . . . . . . . . . . 47
3.5 Non-classicality detection protocols . . . . . . . . . . . . . . . . . . . 49
3.5.1 Via entanglement breaking channel . . . . . . . . . . . . . . . 49
3.5.2 Via initial entropies . . . . . . . . . . . . . . . . . . . . . . . . 50
3.6 Minimalistic assumptions and applications . . . . . . . . . . . . . . . 51
3.7 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52

4 Non-decomposability of evolution & extreme gain of correlations 53

4.1 Motivation and objectives . . . . . . . . . . . . . . . . . . . . . . . . 54
4.2 Correlation capacity bound . . . . . . . . . . . . . . . . . . . . . . . 55
4.2.1 Arbitrary initial states . . . . . . . . . . . . . . . . . . . . . . 56
4.2.2 Uncorrelated initial states . . . . . . . . . . . . . . . . . . . . 59
4.3 Correlation capacity of particular quantifiers . . . . . . . . . . . . . . 60
4.4 Local environment & non-commutativity of Hamiltonians . . . . . . . 61
4.5 Intuition behind excessive gain of correlations . . . . . . . . . . . . . 62
4.6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62

5 Speed of distribution of quantum entanglement 65

5.1 Motivation and objectives . . . . . . . . . . . . . . . . . . . . . . . . 66
5.2 Preliminaries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
5.3 Entangling speed limit: Direct interactions . . . . . . . . . . . . . . . 67
5.4 Entangling speed limit: Indirect interactions . . . . . . . . . . . . . . 69
5.4.1 Saturating the limit . . . . . . . . . . . . . . . . . . . . . . . . 72
5.5 Initial correlation with mediators bounds localised entanglement . . . 74
5.6 Charging power . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
5.6.1 Charge quantifier . . . . . . . . . . . . . . . . . . . . . . . . . 76
5.7 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78

6 Observable quantum entanglement due to gravity 79

6.1 Motivation and objectives . . . . . . . . . . . . . . . . . . . . . . . . 80
6.2 Proposed experimental setups . . . . . . . . . . . . . . . . . . . . . . 81
6.3 Dynamics of gravitationally induced entanglement: Oscillators . . . . 82
6.3.1 Langevin equations . . . . . . . . . . . . . . . . . . . . . . . . 83
6.3.2 Dynamics of covariance matrix . . . . . . . . . . . . . . . . . 84
6.3.3 Noiseless dynamics: Analytical solution . . . . . . . . . . . . . 86

6
 6.3.4 Noisy dynamics: Numerical simulation . . . . . . . . . . . . . 88
6.3.5 Details of entanglement dynamics . . . . . . . . . . . . . . . . 89
6.4 Dynamics of gravitationally induced entanglement: Released masses . 90
6.4.1 Langevin equations and covariance matrix . . . . . . . . . . . 90
6.4.2 Noiseless dynamics: Analytical solution . . . . . . . . . . . . . 91
6.4.3 Quantum and classical trajectories . . . . . . . . . . . . . . . 92
6.5 Other shapes of masses . . . . . . . . . . . . . . . . . . . . . . . . . . 93
6.6 Casimir interactions . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
6.7 Standard decoherence . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
6.8 Comparison with recent experimental achievements . . . . . . . . . . 97
6.9 Gravity: Direct or mediated interaction? . . . . . . . . . . . . . . . . 97
6.10 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98

7 Probing quantum features of photosynthetic organisms 101

7.1 Motivation and objectives . . . . . . . . . . . . . . . . . . . . . . . . 102
7.2 Proposed setup: Bacteria in a Fabry-Perot cavity . . . . . . . . . . . 103
7.3 Model of optical coupling . . . . . . . . . . . . . . . . . . . . . . . . . 105
7.4 Dynamics of light-bacteria system . . . . . . . . . . . . . . . . . . . . 106
7.4.1 Langevin equations . . . . . . . . . . . . . . . . . . . . . . . . 106
7.4.2 Covariance matrix . . . . . . . . . . . . . . . . . . . . . . . . 108
7.4.3 Quantum entanglement . . . . . . . . . . . . . . . . . . . . . . 109
7.4.4 Excitation number . . . . . . . . . . . . . . . . . . . . . . . . 113
7.5 Entanglement as a witness of ultra-strong coupling . . . . . . . . . . 114
7.6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115

8 Other applications 117

8.1 Membrane-in-the-middle optomechanics . . . . . . . . . . . . . . . . . 118
8.1.1 Experimental setup . . . . . . . . . . . . . . . . . . . . . . . . 118
8.1.2 Dynamics of the system . . . . . . . . . . . . . . . . . . . . . 119
8.2 Detecting system-environment correlations . . . . . . . . . . . . . . . 123
8.2.1 Previous detection schemes . . . . . . . . . . . . . . . . . . . . 123
8.2.2 Our protocol . . . . . . . . . . . . . . . . . . . . . . . . . . . 124
8.3 Two cavity fields coupled via a two-level atom . . . . . . . . . . . . . 124
8.3.1 Witnessing non-commutativity and discord . . . . . . . . . . . 125
8.3.2 Strong interactions and bounded entanglement . . . . . . . . . 127
8.3.3 Estimating dimension of mediators . . . . . . . . . . . . . . . 128

9 Conclusion and future work 129

9.1 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
9.2 Future work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131

7
 9.2.1 Entanglement bound for indirect continuous interactions . . . 131
9.2.2 The strength of non-decomposability and correlation gain . . . 132
9.2.3 Protocols witnessing the presence or absence of quantum objects134
9.2.4 Complete link between initial correlations with mediators and
entangling time limit . . . . . . . . . . . . . . . . . . . . . . . 135
9.2.5 Gravity-mediated entanglement: Full treatment . . . . . . . . 135

A Trotter expansion 137

B Entanglement localisation: Prescription 139

Bibliography 141

8
Abbreviations

am Air molecules
BChl Bacteriochlorophyll
cg Casimir-gravity
CM Covariance matrix
CV Continuous variable
det Determinant
di Direct interaction
FWHM Full-width at half-maximum
gnd Ground
inf Infimum
int Interaction
LIGO Laser Interferometer Gravitational-Wave Observatory
loc Local
LOCC Local operations and classical communication
max Maximum
min Minimum
ph Photon
PPT Positive under partial trasposition
QC, qc Quantum-classical
QSL Quantum speed limit
RED Relative entropy of discord
REE Relative entropy of entanglement
RHS Right hand side
SEC System-environment correlation
sep Separable
SDE Standard deviation of energy
sq Squeezed
sup Supremum
th Thermal
tr Trace

9
List of notations

A list of general symbols used throughout this thesis is provided below. Some related
notations are defined in groups for better clarity. Special usage in some Chapters
will be mentioned. Note that trivial notations are not included.

A, B, C Generally used to indicate principal quantum systems (A & B)

and a mediating system (C). Also, multiple systems together are
denoted simply as AB, ABC, etc.
CX:Y , C̃X:Y Classical correlation between systems X & Y and its lower
bound.
D Used to denote a diagonal matrix associated with Brownian
noises and decay rates.
DX|Y Quantum discord between systems X and Y , given the
measurement outcomes on system Y .
D(ρ, σ) General distance measure between quantum states ρ and σ.
Dtr (ρ, σ) denotes trace distance measure.
d, dX Dimension of the Hilbert space of a quantum system and of
a quantum system X. Not to be confused with an object’s length.
EX:Y Quantum entanglement (entropy of entanglement or relative
entropy of entanglement) between systems X and Y . It is also
used to denote logarithmic negativity for the applications part of
the thesis, for simplicity. Not to be confused with another usage
of E which denotes energy or expectation value of energy.
Em Used to denote strength of driving lasers (has frequency unit).
Fm , Qn Zero-mean Gaussian noise operators for CV mode m and n,
respectively. Also, Xm and Ym are used to denote noise
quadratures. These are only used in Chapter 7.
gmn , Gmn Individual and collective coupling between mode m and n. We
also use G̃n to indicate base coupling. These quantities are only
defined in Chapter 7.
G0J Coupling constant defined in Chapter 8. Note also its effective
coupling GJ .

11
 j
H, HX , HX General Hamiltonian, Hamiltonian applies only on system X,
and its jth variant. A lowercase letter can also indicate variant,
e.g., Hj . Also, Hg is used to indicate gravitational term.
hHi, ∆H Expectation value of energy with respect to the ground state
level, i.e., tr(ρH) − Eg and energy variance. The dimensionless
quantities are written as hM i and ∆M , respectively.
IX:Y Mutual information between systems X and Y . Not to be
confused with IX:Y , see below.
j, j † Annihilation and creation operator, respectively,
for a general d-dimensional quantum system.
jin Used to denote input noise in Chapter 8. The corresponding
noise quadratures are denoted by xin,J and yin,J .
K A drift matrix. Note also the notation W± (t) = exp (±Kt) or
M (t) = exp (Kt).
LX:Y Logarithmic negativity between systems X and Y . Not to be
confused with L which denotes distance.
Lρ, LX ρ Lindblad operation on a quantum state ρ and with the operation
only on system X.
m, mX Mass of an object. Not to be confused with dummy index.
nr Refractive index.
n(t) Noise vector used in Chapter 8.
n̄, N̄ Mean excitation numbers. For example, n̄ = (eβ − 1)−1 is
mean thermal phonon with β = ~ω/kB T .
NX:Y Negativity between systems X and Y .
P Used to indicate pressure. Not to be confused with power P .
Q, QX , QjX A quantum operation, operation applied only on system X,
and its jth variant.
QX:Y General correlation quantifier between systems X and Y .
r1 , r2 , r3 Interaction rates for different configurations in Chapter 6.
rcg The ratio between Casimir and gravitational energy terms,
which are relevant for entanglement gain in Chapter 6.
R, RX Radius, radius of object X. Not to be confused with reflectivity
of mirrors, R1,2 , in Chapter 7.
S, S † Squeezing operators with squeezing strength s. Subscripts can
be used to indicate the application on a quantum system,
e.g., sA , sB .

12
S(ρ), S(ρX ) von Neumann entropy of state ρ and its subsystem ρX ,
also can be simply written as SX .
SX|Y Conditional von Neumann entropy.
S(ρ||σ) Relative entropy distance between states ρ and σ.
T Dimensionless time, T = ωt or T = Ωt, where ω and Ω
being frequencies. Not to be confused with temperature. As a
power, it denotes transposition. Also TX denotes partial
transposition with respect to system X.
u Used to define quadrature vector, e.g., u = (XA , PA , XB , PB )T .
Not to be confused with dummy index.
U, UX Unitary operator in general and only on system X.
V, Vij Covariance matrix and its elements. The elements can also be
denoted by 2 × 2 matrices. For example, for a two-mode system,
the components are IA , IB , L, and LT . A useful quantity is
defined in this regard, i.e., Σ = detIA + detIB − 2 detL. After
partial transposition, a covariance matrix is written as Ṽ .
Vm Mode volume, only defined in Chapter 7.
x, p, X, P Position and momentum. The bold uppercase notations are used
to indicate the corresponding dimensionless quantities (or
quadratures). Also, subscripts are used to indicate their systems,
e.g., xA , pB , XA . For field operators, quadratures are denoted
by X and Y .
X, Y Generally used to indicate quantum systems, but they can
indicate parties each consisting of one or more quantum systems.
X:Y Used to indicate symmetrical partition of systems X and Y .
X|Y Indicates asymmetrical partition of systems X and Y . In
particular, it denotes a situation given the knowledge of system
Y , e.g., measurement outcomes.
C A non-convex set containing quantum-classical states.
Ec Casimir energy between two objects.
F(ρ, σ) Fidelity between quantum states ρ and σ.
Fi Finesse.
IX:Y A distance from a state to the closest product state of the form
σX ⊗ σY as quantified by a general correlation quantifier D(·, ·).
L, LX General map, applies only on system X. It is used in association
with Lindblad operators.
M A non-convex set containing product states.
P(ρ) Purity of a quantum state ρ.

13
Q Mechanical quality factor.
S A convex set containing separable states. Also used to indicate
a set in general for general illustration.
1, 1X Identity operator, identity operator applied only on system X.
|ψi, |ψiX A pure quantum state, state of system X.
A lowercase subscript indicates its variant, e.g., |ψij or |ψj i.
Λ, ΛX A general completely positive trace-preserving map,
applies only on system X. Not to be confused with frequency of
lasers Λm in Chapter 7.
Λph , Λam Decoherence coefficients as a result of scattering photons and air
molecules.
ηg Dimensionless figure of merit for gravity-mediated
entanglement gain between two oscillators. σ is the
corresponding figure of merit of entanglement between two
released masses.
Θ(ρ, σ) Bures angle defined as Θ(ρ, σ) = arccos(F(ρ, σ)).
ΞX Charge of quantum battery for system X.
ω, Ω Used to denote frequencies. In Chapter 8 Section 8.3, we also
use g.
ΩN Symplectic form composed of N CV modes. Not to be confused
with the frequency Ω.
λ Wavelength. Not to be confused with linear mass density or
complex coefficients.
∆0J Cavity-laser detuning. Note also its effective detuning ∆j .
δmn Kronecker delta.
δ(t − t0 ) Direct delta.
κν , υ(t), l(t) Constant vector, noise vector, and a vector defined as
l(t) = υ(t) + κν .
ξ, ξX Brownian noise, affecting system X. Also used to denote a
combined annihilation or creation operators in Chapter 8.
Section 8.3.
γ, γX Decay rate, affecting system X. Note that a notation κ is also
used to indicate decay in Chapter 7 and 8.
τe , τph , τam Entangling time, coherence time due to photon scattering, and
due to air molecules.

14
τ (ρ, σ) The time it takes for a state ρ to evolve to another state σ.
Its dimensionless quantity is denoted as Γ(ρ, σ) or simply Γ.
Note, for example, Γdi stands for dimensionless time bound for
direct interactions scenario.
Γn FWHM of bacterial spectrum in Chapter 7.
µ Dipole moment of two-level transition. Not to be confused with
dummy index.
νk , ν̃k The kth symplectic eigenvalue of a covariance matrix,
and its value after partial transposition. Not to be confused
with the frequency constant ν in Chapter 6
ρ, ρX , ρjX General density matrix, density matrix of system X, and its jth
variant. A lowercase subscript also denotes its variant, e.g., ρj .
Not to be confused with mass density in Chapter 6.
Note that σ is also often used to denote a density matrix.
j
σ j , σX j-Pauli matrix (j = x, y, z). The subscript indicates its
application on system X in the context of multiple two-level
systems.
+(−)
σ +(−) , σX Two-level system raising (lowering) operator in general and for
system X.
ΠjX The jth projection operator of system X.

15
Chapter 1

Introduction

This chapter presents the motivation, objectives, and organisation of this thesis.1 It
begins with the importance of quantum entanglement, leading to the need for under-
standing the process of its creation. In particular, we focus on mediated interaction
scenarios where the aim is to create or distribute entanglement between two quantum
objects by utilising an ancillary system. This sets the stage for the thesis. Finally,
I will describe the objectives of our study and how the thesis is organised.

1
Note that this chapter serves as a brief introduction. Some terminologies used here will be
explained more thoroughly in Chapter 2.

17
1.1 Motivation
Entanglement is a form of quantum correlation first recognised by Einstein, Podol-
sky, and Rosen [1], and Schrödinger [2] in 1935. This correlation has proven crucial
and resulted in key proposals for enabling tasks that are not possible in the classical
regime. This includes, to name a few, quantum cryptography [3], quantum dense
coding [4], and quantum teleportation [5]. The rapid utilisation and manipulation
of quantum entanglement for the purpose of quantum information tasks have made
it a crucial resource, some argued, as real as energy [6]. Therefore, the study of its
creation (or distribution) is a necessity. In this section, we will briefly review proto-
cols for creating quantum entanglement between two principal objects, here denoted
as, A and B. In particular, we describe a scenario where A interacts directly with
B and the one where the interactions are mediated by an ancillary system C. For
the latter case, the nature of the interactions can be discrete or continuous, which
will be the main focus of this thesis.

1.1.1 Direct interactions

A straightforward way to create quantum correlations is to engineer direct inter-
actions between the principal objects, see Fig. 1.1. This dynamics can either be
closed (unitary) or open (e.g., following the Lindblad master equation). The general
Hamiltonian considered in this case is of the form HAB = j HAj ⊗ HBj , where the
P

subscripts indicate the corresponding objects. Note also that the Hamiltonian com-
ponents can be local, i.e., HA ⊗ 1B or 1A ⊗ HB , which we will simply write as HA
and HB hereafter. As an example, let us consider two qubits that are interacting
with Hamiltonian H = ~ω σAx ⊗ σBx , where ~ω is the energy unit and σA(B) x
is the
x-Pauli matrix with the subscript denoting the corresponding object. For the initial
condition, let us consider the state |ψ(0)i = |0iA |0iB , where |0iA(B) is the eigen-
state (ground state) of the z-Pauli matrix. One can see that the dynamics from the
Hamiltonian will create excitations in both objects A and B, i.e., the state at time t
is given by |ψ(t)i = α(t) |0iA |0iB + β(t) |1iA |1iB , where α(t) and β(t) are complex
coefficients. This is a form of state where A is said to be entangled with B, as the
state of one object cannot be written separately from the other.2

Figure 1.1: Direct interaction setting. Two quantum particles A and B are inter-
acting with Hamiltonian HAB .

2
A detailed analysis will be given later in this thesis.

18
1.1.2 Indirect interactions
In some cases, the principal objects are not situated together, e.g., A is in Alice’s
lab and B is in Bob’s lab (see Fig. 1.2). Therefore, they require an ancillary system
to mediate interactions between them. We will refer to this as indirect interaction
setting. Two scenarios can be applied here based on the nature of the interactions.
A discrete interaction scenario (Fig. 1.2a) involves performing operations (or discrete
interactions) on objects A and C in Alice’s lab, then C is sent to Bob’s lab, after
which Bob can perform operations on B and C. On the other hand, a continuous
interaction scenario (Fig. 1.2b) considers continuous (in time) interactions between
A and C as well as between B and C. We note that both scenarios have been
presented by Cubitt et al. [7].

Figure 1.2: Indirect interaction setting. Quantum particles A and B are in different
locations. In discrete interaction scenario (a), an ancillary system C first interacts
with A, then it is sent to Bob, after which it also interacts with B. In scenario (b),
the ancillary system is interacting continuously with A and B. Note that in both
scenarios, A and B are not interacting directly, i.e., there is no Hamiltonian HAB .

Discrete interactions

In the discrete interaction scenario, a simple way to create entanglement between

A and B is to first entangle A and C, after which one can perform entanglement
swapping (e.g., see Ref. [8] and its implementations [9, 10]) on B and C, resulting in
entanglement between the principal objects. This scenario requires C to be entan-
gled in the process. Interestingly, the distribution of entanglement is also possible
via a separable ancillary system [7]. The results of Cubitt et al. have attracted at-
tention, resulting in entanglement distribution proposals [11–14]. The distribution

19
via separable mediators has also been verified experimentally [15–18]. Additionally,
Refs. [19, 20] have shown that the entanglement gain in this scenario is bounded by
another form of non-classical correlation, known as quantum discord, between the
principal particles AB and the mediator C. It demonstrates that the change of quan-
tum entanglement as a result of relocating the object C from Alice’s lab to Bob’s lab
is bounded by the quantum discord carried by C. This way, some form of quantum
correlation is required to distribute entanglement. We also note the distribution in
the presence of noise with the use of different entanglement quantifiers [21, 22].

Continuous interactions

On the other hand, the continuous interaction scenario assumes that the total energy
of the objects is described by Hamiltonian of the form H = HAC + HBC . We note
A and B do not interact directly, and therefore there is no HAB component. More
explicitly, the total Hamiltonian can be written as H = j HAj ⊗HCj 1 + k HBk ⊗HCk 2
P P

(local Hamiltonian terms are included). Note that the subscripts 1 and 2 are used
to indicate that the terms HCj 1 and HCk 2 can be different in general. For most parts
of this thesis, we allow each object to interact with its own local environment, with
the corresponding dynamics following the Lindblad master equation. This way, the
continuous interaction scenario corresponds to ample situations in nature, some of
which will be considered in this thesis. Yet, this entanglement distribution method
is not as well understood as its counterpart, the discrete interaction scenario.

1.2 Objectives
As first shown by Cubitt et al., the continuous scenario allows entanglement distri-
bution via a separable mediator [7]. Investigating further questions in this direction
is therefore the goal of this thesis. In particular, one seeks answers to the following:

1. The requirements for entanglement distribution (if not entangled mediator).

2. The amount of distributed entanglement and the relevant factors.

3. The speed limit of distributing quantum entanglement.

In addition to answering these questions, we will also present the implications of our
study and explore applications that can benefit from it.

1.3 Thesis organisation

This thesis is organised as follows. Firstly, Chapter 2 covers preliminaries that are
useful for the core of the thesis. Readers who are interested mainly in new results can

20
proceed to Chapters 3–8. Chapter 3 presents our results on the resource necessary
for distribution of quantum entanglement via continuous interactions. An impor-
tant and general application will be constructed that is aimed at indirect revelation
of non-classical properties of unknown objects. Another resource for entanglement
distribution will be presented in Chapter 4. More specifically, we link the amount
of distributed correlations to one of the most elementary non-classical features of
quantum observables, that is non-commutativity. Chapter 5 considers the speed of
entanglement distribution for direct and indirect interaction cases. We present con-
ditions for the indirect interaction setting to saturate the maximum entangling speed
that is derived from the quantum speed limit. Major applications of our theorems
will be presented in Chapter 6 and 7. Chapter 6 addresses one of the most unsettled
quests in physics, that is showcasing quantum nature of gravitational interactions.
In particular, we propose experimental setups where gravity can mediate observable
quantum entanglement between two otherwise non-interacting masses. Chapter 7
covers the problem of detecting quantum features in living organisms by proposing
an experimental setup where one can probe non-classical properties of photosyn-
thetic organisms, such as green sulphur bacteria. Finally, other applications will be
presented in Chapter 8.

21
Chapter 2

Quantum objects, their dynamics,

and properties

This chapter provides preliminaries that will be useful for the remainder of this
thesis. First, I describe quantum objects and their states, both for discrete systems
and continuous variable systems. Additionally, quantum systems evolve in time due
to interactions between them or simply as a result of local energy. In this regard,
I provide the dynamics of quantum objects both for closed and open (interactions
with environment) systems. Finally, I proceed to review important properties that
quantum objects can possess. This includes quantities such as von Neumann entropy,
purity, and fidelity, as well as correlations such as quantum entanglement, quantum
discord, and mutual information. More in-depth analysis can be found in Refs. [6,
23–25]. I will also introduce the notion of classicality and non-classicality that will
be used throughout the thesis.

23
2.1 Quantum objects
Within the framework of quantum mechanics, a physical object is fully described
by its wave function (or, more abstractly a state |ψi). Knowing the state of the
object, one can then calculate expectation values of observables such as position,
momentum, and others. In this section, we focus on describing the state of objects
for two cases. One is when an object “lives” in a finite dimensional Hilbert space,
such as the spin state of an electron, while the other is when the description requires
infinite dimension, such as an atom in a harmonic potential. Indeed, the latter has
infinitely many energy levels. Finally, I will also cover the evolution of these objects
both where they are isolated and having interactions with environment.

2.1.1 Discrete systems

For quantum systems that only require finite-dimensional space, their pure state is
superposition
Xd−1
|ψi = λn |ni , (2.1)
n=0

where, e.g., {|ni} correspond to discrete energy levels,1 d is the dimension of the
object’s Hilbert space, and λn is a complex coefficient. As a simple example, consider
a two-dimensional system such as the spin state of an electron |ψs i = λ0 |0i + λ1 |1i,
where |0i (|1i) denotes spin up (down). Note that normalisation of the state requires
|λ0 |2 + |λ1 |2 = 1, which can be interpreted as the electron having spin up with
probability |λ0 |2 and spin down with probability |λ1 |2 .
A quantum system can also be in a mixture of states (simply, mixed state or
density matrix)
X
ρ= pm |ψm i hψm | , (2.2)
m
P
where |ψm i is any pure state. The normalisation condition requires m pm = 1
and pm ≥ 0 gives the probability of observing the state |ψm i. This requirement
is referred to as positive semidefiniteness, i.e., a physical state ρ has to have real
non-negative eigenvalues with the sum equal to unity. It turns out that any such ρ
has a spectral decomposition

d−1
X
ρ= pn |nihn|, (2.3)
n=0

where {|ni} form a basis in which ρ is diagonal. Note that in a matrix representation,
the mixed state in Eq. (2.3) has dimension d×d, while the pure state |ψi in Eq. (2.1)
1
We will also call them eigenstates and they form an orthonormal basis, i.e., hn|n0 i = δnn0 .

24
is a vector with dimension d. In this thesis, we will use |ψi or ρ to describe the
state of quantum objects, from which one can calculate important quantities. Some
examples will be given later in this chapter.

Two-level systems

A special case where d = 2 (simply, a qubit) will be used often in this thesis. Now
we provide useful notations for the states and operators.
! ! ! !
1 0 1 1 1 1
|0i = , |1i = , |±i = √ , |y± i = √ . (2.4)
0 1 2 ±1 2 ±i

Note that {|0i , |1i}, {|±i}, and {|y± i} each form an eigenbasis of the Pauli matrix
! ! !
1 0 0 1 0 −i
σz = , σx = , σy = , (2.5)
0 −1 1 0 i 0

respectively. Additionally, we also use the raising and lowering operators, respec-
tively written as σ + = |1i h0| and σ − = |0i h1|.

2.1.2 Continuous variable systems

Technically, infinite dimensional (simply, continuous variable) systems can be de-
scribed using the state in (2.1) or (2.3) with infinite sums. The calculations of
observables follow accordingly. However, for continuous variable (CV) Gaussian
systems, which belong to a special class of states mostly considered for applications
later in this thesis, the description of the corresponding object can be made simpler.
Indeed, a complete specification of the object can be incorporated in the covariance
matrix (CM), denoted by V , whose components are defined as

hui uj + uj ui i
Vij = − hui ihuj i, (2.6)
2

where u = (X, P )T is a quadrature vector2 with X and P being dimensionless

position and momentum quadratures.3
In this thesis, we use the following relations

a + a† a − a†
X= √ , P = √ , (2.7)
2 2i
2
Note that we adopt the term quadrature vector from literature. However, one should note that
u is not a vector in the normal algebraic sense. For instance, the cross product u × u 6= 0, which
is a consequence of the non-commuting relation between X and P .
3
In this example, V is a single-mode covariance matrix. For an N -mode CM, one writes the
quadrature vector u = (X1 , P1 , X2 , P2 , · · · , XN , PN )T .

25
where a (a† ) is the annihilation (creation) operator of the corresponding infinite
dimensional object. We note that an N -mode covariance matrix has N symplectic
eigenvalues {νk }N
k=1 , which can be calculated from the spectrum of a matrix |iΩN V |,
where !
N
M 0 1
ΩN = , (2.8)
k=1
−1 0
is known as symplectic form. A physical covariance matrix has

νk ≥ 1/2 (2.9)

for all k = 1, 2, · · · , N . This is essentially a consequence of the uncertainty rela-

tion [25].4
We now provide some examples of states and the corresponding CM that will be
useful in this thesis. This includes coherent, thermal, and squeezed thermal states
of a harmonic oscillator
∞ ∞  j
−
|α|2 X αj 1 X n̄
|αi = e 2 √ |ji , ρth = |ji hj| , ρsq = Sρth S † , (2.10)
j=0
j! n̄ + 1 j=0 n̄ + 1

where the mean phonon number n̄ = (exp(β) − 1)−1 with β = ~ω/kB T , ~ω the
energy unit of the oscillator, and T the temperature. Also, |ji denotes the jth
eigenstate of the operator a† a. The single-mode squeezing operator is given by

η∗
 
η † †
S = exp aa − a a , (2.11)
2 2

where η = s exp (iθ) is a complex number with s and θ being the squeezing strength
and angle respectively. The corresponding CM is obtained by computing the expec-
tation value of observables in Eq. (2.6) with respect to the state of the object. One
can confirm
!
2s
1 2n̄ + 1 2n̄ + 1 e 0
Vα = 1, Vth = 1, Vsq = −2s
, (2.12)
2 2 2 0 e

where 1 here denotes a 2 × 2 identity matrix and we have assumed the case θ = 0.5
Moreover, the symplectic eigenvalues are given by 1/2, (2n̄ + 1)/2, and (2n̄ + 1)/2
respectively. Note that all the covariance matrices above are physical, i.e., they
satisfy the condition in Eq. (2.9).
4
√
Note that some literature uses the relations in (2.7) without the factor 2, and therefore the
corresponding condition in (2.9) becomes νk ≥ 1.
5
This corresponds to squeezing the momentum (position) quadrature for s > 0 (s < 0).

26
2.1.3 Quantum dynamics
Here we describe the evolution of quantum objects, i.e., how their state |ψi (or, in
general ρ) changes in time.

Closed systems

For an isolated system, the dynamical equation reads

|ψ(t)i = U |ψ(0)i , ρ(t) = U ρ(0)U † , (2.13)

for pure and mixed states respectively, where U = exp(−iHt/~) is a unitary operator
and H is the Hamiltonian that represents the corresponding total energy of the
object.6 In general, the Hamiltonian can be composed of many terms, i.e., H =
†
P
j Hj . Note also that during the evolution, we have tr(ρ(t)) = tr(U ρ(0)U ) =
tr(ρ(0)U † U ) = tr(ρ(0)) = 1, where we used the cyclic property of trace and unitary
property U † U = U U † = 1.
Let us now provide a tool for the expression of the unitary operator that will be
crucial for some parts of this thesis. In particular, we consider a unitary operator
with Hamiltonian H = H1 + H2 . We utilise the Trotter expansion (see Appendix A
for a simple derivation) and write the operator as
      n
i(H1 + H2 )t iH1 ∆t iH2 ∆t
U = exp − = lim exp − exp − , (2.14)
~ n→∞ ~ ~

where ∆t = t/n. Hence, one can think of the evolution as if it consisted of sequences
of two unitary operations with Hamiltonian H2 and H1 (or, in reverse), each for a
time ∆t. For a special case, in which the Hamiltonians commute, [H1 , H2 ] = 0, the
corresponding operator reads
   
iH1 t iH2 t
U = exp − exp − . (2.15)
~ ~

Open systems

For quantum systems that are open, i.e., having interactions with their environment,
we will consider the evolution of the state following the Lindblad equation

dρ(t) i
= − [H, ρ(t)] + Lρ(t), (2.16)
dt ~
6
We have also assumed that the Hamiltonian here is time-independent which may not apply
in general. In this thesis, we always try to move into a frame where the Hamiltonian is time-
independent (e.g., by changing to a rotating frame) and, wherever convenient, solve the problem
in Heisenberg picture.

27
where Lρ(t) ≡ k Qk ρ(t)Qk† − 12 {Qk† Qk , ρ(t)}, H is the Hamiltonian of the object,
P

and Qk describes the operation on the object as a result of interactions with its
environment.7 Note that, for simplicity, the strength of the interaction is absorbed
in Qk such that the term Lρ(t) has the unit frequency. For the case where the
strength goes to zero, the 2nd term on the RHS of Eq. (2.16) vanishes and the
dynamics becomes unitary as in Eq. (2.13). Also, by using the cyclic property of
trace, it is clear that tr([H, ρ(t)]) and tr(Lρ(t)) both equal zero, and consequently,
the Lindblad equation is trace-preserving.

2.2 Properties of quantum states

Having known the state of quantum objects and its evolution in time, we now proceed
to some quantities that can be computed from it. This includes von Neumann
entropy, purity, and fidelity between quantum states. We will focus on definitions
and list of properties that will be utilised later in this thesis.

2.2.1 von Neumann entropy

The amount of disorder in a quantum state can be quantified using

S(ρ) = −tr(ρ log2 (ρ)).8 (2.17)

For a general mixed state, e.g., the one in Eq. (2.3), the von Neumann entropy
P
(simply, entropy) is given by − n pn log2 (pn ). As {pn } form a probability distribu-
tion, the entropy of a quantum object is non-negative. Some other useful properties
of von Neumann entropy are listed below.

1. S(ρ) = 0 if and only if the state ρ is pure.

2. For a d-dimensional object, the maximum entropy is log2 (d).

3. S(U ρU † ) = S(ρ) (invariant under unitary operations).

4. For two systems, S(ρX ⊗ ρY ) = S(ρX ) + S(ρY ) (additive).

5. For three systems, S(ρXY Z )+S(ρY ) ≤ S(ρXY )+S(ρY Z ) (strongly subadditive).

7
For consistency, we have used the superscript k to denote the variant of the operator Q and
not the power.
8
In this thesis, we shall call this the von Neumann entropy as it is commonly done in literature.

28
2.2.2 Purity
Some quantum states are more mixed than the others. In order to quantify the
purity of a state ρ, we use
P(ρ) = tr(ρ2 ). (2.18)

For a general state with dimension d, it acquires values 1/d ≤ P ≤ 1. Indeed, P = 1

for a pure state ρ = |ψi hψ| and P = 1/d for a maximally mixed state ρ = 1/d.
We note that in a unitary dynamics, one has

tr(ρ(t)2 ) = tr(U ρ(0)U † U ρ(0)U † ) = tr(ρ(0)2 ), (2.19)

with the help of the cyclic property of trace and the unitary property. Therefore,
unitary dynamics is purity-preserving. While the purity of the state ρ is invariant,
the purity of its subsystems can change. We will see below some examples showing
this case.
As the von Neumann entropy characterises the disorder in the system, one can
also quantify the purity as 1 − S(ρ)/ log2 (d). In this case, its value ranges from a
minimum of 0 to a maximum of unity.

2.2.3 Fidelity
In order to assess how close a quantum state is from another, one of the quantifiers
we use is the Uhlmann root fidelity
q 
√ √
F(ρ, σ) = tr ρσ ρ , (2.20)

where ρ and σ are two density matrices having the same dimension [26, 27]. In
a way, the fidelity above is related to a probability of identifying one state as the
other. Some properties of fidelity are listed below

1. F(ρ, σ) = F(σ, ρ) (symmetrical).

2. Its value follows 0 ≤ F(ρ, σ) ≤ 1 and F(ρ, ρ) = 1.

3. For pure states ρ = |ψi hψ| and σ = |φi hφ|, F(ρ, σ) = |hψ|φi|.

4. F(Λ[ρ], Λ[σ]) ≥ F(ρ, σ) for any trace-preserving completely positive map Λ.

One might also consider a measure introduced by Helstrom that shows the best
success probability to distinguish two quantum states, i.e.,

1
pbest = (1 + Dtr (ρ, σ)) , (2.21)
2

29
where Dtr (ρ, σ) = ||ρ − σ||1 /2 is known as the trace distance and || · ||1 is trace
norm [28]. The probability in Eq. (2.21) is then related to the fidelity in Eq. (2.20)
through the following relation [23]
p
1 − F(ρ, σ) ≤ Dtr (ρ, σ) ≤ (1 − F(ρ, σ)2 ). (2.22)

2.3 Correlations between quantum states

Two or more quantum objects can have correlations between them that are beneficial
for quantum information tasks. These correlations can be in many forms ranging
from purely classical to having quantum nature. This section reviews types of cor-
relations used in this thesis. In particular, this includes quantum entanglement,
quantum discord, and mutual information. Important properties of the quantifiers
that will be useful for this thesis will be mentioned. We will also introduce the
notion of classicality and non-classicality.

2.3.1 Quantum entanglement

In this thesis, we will only consider bipartite entanglement, i.e., between party X
and Y .9 For pure states, separable (not entangled) states are those which can be
written in a product form
|ψi = |ψX i ⊗ |ψY i , (2.23)

where the subscripts denote the corresponding party. Therefore, if the description of
the whole system XY does not follow Eq. (2.23), the parties X and Y are entangled.
For mixed states, separability is defined as a mixture of product states
X
ρsep = pj ρjX ⊗ ρjY , (2.24)
j

where {pj } form a probability distribution. States which are not of this form there-
fore possess quantum entanglement. We note an important property of quantum
entanglement – monotonicity, i.e., it cannot be created or increased on average un-
der local operations and classical communication (LOCC). It is clear that both the
states in (2.23) and (2.24) can be prepared via LOCC.
As examples, we list below the Bell states (maximally entangled two-qubit states)

1 1
|ψ± i = √ (|01i ± |10i) , |φ± i = √ (|00i ± |11i) . (2.25)
2 2

Note the simplified notation, e.g., |0iX ⊗ |1iY = |01i, will be used throughout this
9
In general, each party can consist of more than one physical object.

30
thesis. In general, some states are more entangled than the others, and therefore are
more useful for performing tasks. Below we provide some exemplary entanglement
quantifiers.

Entropy of entanglement

For pure states, one can quantify the amount of entanglement using the entropy of
entanglement, which is defined below. Schmidt decomposition allows one to write a
pure state |ψi of system XY as
X
|ψi = λj |xj i |yj i , (2.26)
j

where {|xj i} and {|yj i} are orthonormal on system X and Y respectively. The state
of each party is given by partially tracing out the other. For example,
X
ρX = trY (|ψi hψ|) = |λj |2 |xj i hxj | ,
j
X
ρY = trX (|ψi hψ|) = |λj |2 |yj i hyj | (2.27)
j

The information regarding the correlation between X and Y (λj ) is carried in the
states ρX and ρY . The entropy of entanglement is defined as

EX:Y = S(ρX ) = S(ρY ). (2.28)

Relative entropy of entanglement

By using the so-called distance approach based on relative entropy, one can quantify
the amount of entanglement in a state as its distance from the closest separable state.
In particular, we have the relative entropy of entanglement (REE)

EX:Y = inf S(ρ||σ), (2.29)

σ∈S

where the relative entropy S(ρ||σ) = −tr(ρ log(σ)) − S(ρ) and S is a convex set
which contains separable states of Eq. (2.24) [29]. Note that for pure states, REE
reduces to the entropy of entanglement. In general, REE is hard to compute as it
involves optimisation.

Negativity and logarithmic negativity

For separable states in Eq. (2.24), it has been shown that the eigenvalues are pos-
itive after partial transposition with respect to one party [30]. This is commonly

31
referred to as the PPT criterion. If a state is not PPT, one argues that the corre-
sponding negative eigenvalues quantify the amount of entanglement in the system.
These negative eigenvalues show that the corresponding density matrix after partial
transposition is not physical. In particular, one defines a computable quantifier of
entanglement known as negativity

||ρTX ||1 − 1
NX:Y = , (2.30)
2

where ||ρTX ||1 is the trace norm of the state after partial transposition with respect
to party X [31–34].10
A logarithmic quantifier, which stems from negativity, can be defined as

LX:Y = log2 (||ρTX ||1 ) = log2 (2NX:Y + 1) , (2.31)

and is known as logarithmic negativity. In this thesis, we will mostly used the
logarithmic negativity to quantify quantum entanglement between CV systems. In
this case, the expression is given by

LX:Y = max 0, − log2 (2ν̃min ) , (2.32)

where ν̃min is the minimum symplectic eigenvalue after partial transposition of the
state (or covariance matrix) with respect to one party [34, 35]. Similar to the PPT
criterion, entanglement is shown by the violation of the physicality requirement of
the covariance matrix, i.e., ν̃min < 1/2.

2.3.2 Quantum discord

A quantum state can easily be perturbed by measurements. Classical states are
distinguishable, and therefore should be represented by orthogonal states. For a
bipartite system, the so-called quantum-classical (QC) state is defined as
X
ρqc = pj ρjX ⊗ |yj i hyj | , (2.33)
j

where {|yj i} form an orthonormal basis. We note that there exists a von Neu-
mann measurement on the side of Y that does not perturb the whole system, i.e.,
P j j j P j 11
j ΠY ρΠY = ρ, where ΠY = |yj i hyj | is a projector and j ΠY = 1Y . Other states
will be perturbed by any von Neumann measurement, characterising the quantum-
ness in the correlation. Quantum discord is a non-classical correlation of states that
10
Note that the partial transposition can be done with respect to party Y , i.e., it is symmetrical.
11
To clarify, this is a non-selective von Neumann measurement and will be referred to as simply
von Neumann measurement henceforward.

32
cannot be written as in Eq. (2.33) [36, 37]. We note that entanglement is a stronger
type of quantum correlation in the following sense. As any quantum-classical state
in Eq. (2.33) is a type of separable state in Eq. (2.24), all entangled states contain
quantum discord, but the reverse is not always true. As an example, consider the
state
1

ρ= |0i h0| ⊗ |0i h0| + |1i h1| ⊗ |+i h+| . (2.34)
2
This state is clearly separable, but is discorded since it cannot be written as (2.33)
(with the orthonormal basis on the side of system Y ). One can see that by exchang-
ing the system X ↔ Y in Eq. (2.34), the resulting state is of the form (2.33) and
therefore has zero discord. This emphasises that discord is not symmetrical.

Relative entropy of discord

Based on two expressions of mutual information that are equal in the classical regime
but give different values in the quantum regime, a quantifier of quantum discord was
first presented as the difference [37]. In this thesis, however, we will use another form
of quantifier using the distance approach based on relative entropy. In particular,
quantum discord is a distance from a given state ρ to the closest quantum-classical
state of Eq. (2.33), i.e.,
DX|Y = inf S(ρ||σ), (2.35)
σ∈C

where C is a non-convex set containing the quantum-classical states. This quantifier

is known as the relative entropy of discord (RED) [38] or the one-way quantum
deficit [39]. Note that one can also rewrite the expression of discord as an opti-
misation over von Neumann measurement on the side of one party, i.e., DX|Y =
inf {Πj } S(ρ|| j ΠjY ρΠjY ). Consequently, this means that for the states in (2.33),
P
Y
there is always a measurement {ΠjY = |yj i hyj |} (not perturbing the system) that
minimises the RED. It is also apparent that RED is not a symmetrical quantity, i.e.,
in general, DX|Y 6= DY |X . This can be seen in the state of Eq.(2.34), where DX|Y
is nonzero while DY |X equals zero. For pure states, RED reduces to the entropy of
entanglement.

The flags condition

Here we note a useful property for the states of Eq. (2.33). Let us consider three
quantum objects A, B, and C with the state written as
X
ρ= pj ρjAB ⊗ |cj i hcj | . (2.36)
j

33
The entanglement of this quantum-classical state follows
X
EA:BC (ρ) = pj EA:B (ρjAB ). (2.37)
j

This equation is known as the flags condition [40]. Furthermore, if one uses the REE
as entanglement quantifier, the closest separable state to ρ reads
X j
σ= pj σAB ⊗ |cj i hcj | , (2.38)
j

j
where σAB is the closest separable state to ρjAB [40].

Classicality & non-classicality

This section is dedicated to setting the notion of classicality and non-classicality

that will be used throughout this thesis. These definitions are correlation-based in
the sense that they are related to the type of correlation between the object under
consideration and other objects.
We will use the term classical for an object if there is a von Neumann measure-
ment on it that does not perturb the whole (bigger) system. As an example, we say
that system Y is classical for the QC states in Eq. (2.33), which have zero discord.
Note that this way, classicality is not a property of an object alone, rather it is a
property of a bigger system, with the object as one of its components.
Note that if we were to consider only the object in question, i.e., by tracing
out the remaining others, one can always find a spectral decomposition of the corre-
sponding density matrix such that it is expressed as a sum of weighted orthogonal (a
signature of classicality) projectors. Therefore, a single object (regardless of having
correlations with other objects) is always classical at a particular time in a particular
reference basis. Should this basis change in time, the system is developing coher-
ence, which is quantum in nature [41]. This is simply shown by nonzero off-diagonal
elements in the density matrix, as represented in the reference basis.
In contrast with classicality, non-classicality is related to discorded states. There-
fore, being “quantum” or non-classical is associated with the ability to have non-
classical correlations with other objects. Furthermore, for discorded states, e.g., with
DX|Y 6= 0, one can prepare quantum coherence on the side of Y by only operating
on system X [42]. In this way, our notion of non-classicality can be thought of as
a property of an object alone. Also, as entangled objects possess quantum discord,
they are automatically non-classical.

34
2.3.3 Mutual information
In addition to non-classical correlations mentioned above, quantum systems can
be classically correlated. As a measure of total correlation, we use the mutual
information
IX:Y = S(ρX ) + S(ρY ) − S(ρ), (2.39)

where ρX and ρY are marginals (reduced states) of ρ [43]. One can also phrase the
mutual information using the distance approach. In particular, one writes IX:Y =
inf σ∈M S(ρ||σ), where M is a non-convex set containing product states. As the
closest product state to ρ is simply given by its marginals, the expression simplifies
to Eq. (2.39) [38]. Also note that the mutual information is equal to twice the
entropy of entanglement for pure states.
Finally, all considered correlations are summarised in Fig. 2.1. In this unified
view, one can clearly see that EX:Y ≤ DX|Y ≤ IX:Y . This is a consequence of the
relation for the sets M ∈ C ∈ S.
We should also mention that the presence of quantum discord, e.g., DX|Y 6= 0,
means that the total correlations between systems X and Y , i.e., mutual information,
cannot all be accessed by measuring system Y alone [37]. In the context of this thesis,
we will show how to reveal quantum discord without performing measurements on
system Y , which is one of the main results in Chapter 3.

Figure 2.1: Geometrical illustration of correlations. The correlation quantifiers

shown include relative entropy of entanglement EX:Y , relative entropy of discord
DX|Y , and mutual information IX:Y . The blue shaded area represents the convex
set of separable states and the solid lines represent the non-convex set of quantum-
classical states that also consist of product states, e.g., ρX ⊗ ρY .

Conditional entropy

Conditional entropy is a useful quantity that we will utilise later in this thesis. It is
defined as
SX|Y = S(ρ) − S(ρY ). (2.40)

35
We note that for separable states in Eq. (2.24), the conditional entropy follows
SX|Y , SY |X ≥ 0 [44]. Consequently, the negative value, SX|Y , SY |X < 0, implies
quantum entanglement between party X and party Y . This way, one notices a
difference between the classical regime where the conditional entropy is always non-
negative and the quantum regime where it can be negative. For example, any pair
of maximally entangled qubits in Eq. (2.25) gives SX|Y = SY |X = −1.

36
Chapter 3

Revealing non-classicality of
inaccessible objects

In this chapter, the resource for entanglement gain in a scenario where two principal
objects are continuously interacting via a mediating system will be investigated.1 I
begin with an application-related motivation of showcasing quantum features of un-
known objects that are not accessible by direct experimentation and a construction
of a general scenario that will be considered throughout this chapter. Next, by tak-
ing non-correlated systems as the initial condition, I will show that quantum discord
between the mediator and the principal objects is a necessary requirement. On the
other hand, an interesting scenario will be presented where the entanglement is dis-
tributed via a classical mediator, i.e., zero discord. This scheme, however, needs
some correlation already present in the initial state. By utilising this knowledge, I
will devise a detection method capable of revealing a quantum property of an inac-
cessible object, as quantified by the quantum discord. Our protocols use minimalistic
assumptions: no information is required regarding the dimensionality of the objects
involved, the initial state, and the explicit form of the interactions. Also, all the
objects can be open systems and no measurements are applied on the inaccessible
object.

1
Parts of this chapter are reproduced from our published article of Ref. [45], c [2019] American
Physical Society. Where applicable, changes made will be indicated.

37
3.1 Motivation and objectives
What should be known about an inaccessible object to conclude that it is “not
classical”? In this thesis, inspired by quantum communication scenarios, I show the
verification that such object can be used to increase quantum entanglement between
remote probing particles (that individually interact with it) is a sufficient indication.
Specifically, we prove that such gain in quantum entanglement is only possible
if, during its evolution, the object shares with the probes quantum correlations in
the form of quantum discord [24, 36, 37, 46, 47]. In turn, the presence of quantum
discord between the probes and the object entails a non-classical feature of the
object itself. According to the definition of discord, two or more subsystems share
quantum correlations if there is no von Neumann measurement on one of them
that keeps the total state unchanged. This can only happen when non-orthogonal
(indistinguishable) states are involved in the description of the physical configuration
of the measured subsystem. This indistinguishability is the non-classical feature that
we aim to detect. We formulate analytical criteria revealing such non-classicality
based on operations performed only on the probing objects, and without any detailed
modelling of the inaccessible object in question.
We emphasise that the non-classicality is revealed under a set of minimal as-
sumptions. Namely: (i) The object may remain inaccessible at all times, i.e., it
needs not be directly measured. In particular its quantum state and Hilbert space
dimension can remain unknown throughout the whole assessment. Our method is
thus valid when the object is an elementary system or an arbitrarily complex one;
(ii) The details of the interactions between the object and the probes may also re-
main unspecified; (iii) Every party can be open to its own local environment. These
properties make our method applicable to a large number of experimentally relevant
situations. We demonstrate the revealing power of our criteria in Chapters 6, 7, and
8 for the detection of quantum property of gravitational interaction, photosynthetic
organisms, optomechanical mirror in the membrane-in-the-middle setting, and more.
The general scenario we consider in this chapter is depicted in Fig. 3.1. System
C is assumed to be the inaccessible object and to mediate the interaction between
two remote probes, labeled A and B. Therefore, here, we refer to system C as the
mediator. It is essential for our method that the probes are not directly coupled
and only interact via the mediator. This means that the Hamiltonian for the pro-
cess under scrutiny can be written as HAC + HBC , with HJC being the interaction
Hamiltonian between the mediator C and probe J = A, B.

38
Figure 3.1: General setting for revealing non-classicality of inaccessible objects. An
inaccessible object C is mediating interactions between otherwise non-interacting
probing objects A and B. In general, the state of C can be unknown and no
measurement can be conducted on it. In the text, we present conditions for which
entanglement gain between A and B implies non-classicality of C, in the form of
quantum discord DAB|C . Note that our methods do not assume the dimensionality
of any object and the explicit form of the interacting Hamiltonians HAC + HBC .
Also, all the objects can be open to their own local environment (grey shaded area).

3.2 Instrumental discord

Our work is developed in the context of entanglement distribution with continuous
interactions [7]. We first focus on the partition A : BC and demonstrate a result
which will be instrumental to design our criteria for the inference of non-classicality
of C based on entanglement dynamics in AB only. Recall that previous studies on
the resources allowing for entanglement distribution showed that any three-body
density matrix, i.e., the state of ABC at any time t in the present context, satisfies
the inequality [19, 20]:

|EA:BC (t) − EAC:B (t)| ≤ DAB|C (t). (3.1)

Here EX:Y is the relative entropy of entanglement in the partition X : Y [29], and
DX|Y is the relative entropy of discord [38], also known as the one-way quantum
deficit [39]. Note that relative entropy of discord is in general not symmetric, i.e.,
DX|Y 6= DY |X . Eq. (3.1) shows that the change in entanglement due to the relocation
of C is bounded by the quantum discord carried by it.
Let us start from the simple case where the overall probes-mediator system is
closed (which allows us to ignore for now the grey-colored shadows in Fig. 3.1). If
the interaction Hamiltonians HJC satisfy [HAC , HBC ] = 0, the evolution operator
from the initial time t = 0 to some finite time τ is just U = UBC UAC , where
UJC = exp (−iHJC τ /~). This situation is equivalent to first interacting C with A
and then C with B (or in reversed order). However, note that the density matrix
†
ρ0 = UAC ρ0 UAC obtained by “evolving” the initial state through UAC only does not
describe the state of the system at τ . Nevertheless, we now show the relevance of
the properties of state ρ0 for entanglement gain.
Consider the following forms of Eq. (3.1) written for the initial state ρ0 and the

39
instrumental state ρ0 , respectively

EAC:B (0) − EA:BC (0) ≤ DAB|C (0),

0 0 0
EA:BC − EAC:B ≤ DAB|C . (3.2)

0
Note that EAC:B (0) = EAC:B , because UAC is a local unitary operator in this parti-
†
tion. The state at time τ is given by ρτ = UBC ρ0 UBC 0
, and thus EA:BC (τ ) = EA:BC ,
this time owing to UBC being local. Summing the above inequalities we obtain a
bound on the entanglement gain

0
EA:BC (τ ) − EA:BC (0) ≤ DAB|C (0) + DAB|C . (3.3)

This opens up the possibility to create entanglement at time τ without producing

discord at both t = 0 and τ , but rather by utilising non-classicality in the instru-
mental state. In other words, the gain of entanglement in A : BC could be mediated
by object C, which gets non-classically correlated by UAC and then decorrelated by
UBC . Therefore, C is only classically correlated at times t = 0 and τ . We now give
a concrete example of this type of entanglement creation.
Consider all the objects to be qubits and take the interaction Hamiltonian as

H = (σAx ⊗ 1 ⊗ σCx + 1 ⊗ σBx ⊗ σCx ) ~ω, (3.4)

where σ j (j = x, y, z) is the Pauli-j matrix and ω is the frequency. As initial state

we choose the classically correlated state

1
ρ0 = 2
|011ih011| + 12 |100ih100|, (3.5)

where, e.g., σ z |0i = |0i. One can now readily check that the relative entropy of
entanglement EA:BC grows from 0 to 1 in the timespan from T = 0 to T = π/4,2
whereas discord DAB|C remains zero at these two times. The gain is indeed due to
non-classical correlations of the instrumental state: applying only UAC for a time
0
T = π/4 produces discord DAB|C = 1. A summary of this dynamics is presented in
Fig. 3.2.
For completeness, I provide the detailed calculations below. First, we note that
the Hamiltonian has commuting components, i.e., [HAC , HBC ] = 0 and therefore we
write the evolution operator as U = UBC UAC . Next, entanglement in the partition
2
For simplicity, we define a dimensionless time variable T = ωt.

40
Figure 3.2: Entanglement distribution utilising instrumental discord. The Hamil-
tonian is taken to be H = (σAx ⊗ 1 ⊗ σCx + 1 ⊗ σBx ⊗ σCx ) ~ω with initial state
1
2
|011ih011| + 12 |100ih100|. Note the maximum quantum discord is DAB|C ≈ 0.81 at
ωt = π/8. This shows that the entanglement gain is not bounded by the maximum
discord.

A : BC is calculated as follows

0
EA:BC (T ) = EA:BC
1 † 1 †
= EA:C (UAC |01i h01| UAC ) + EA:C (UAC |10i h10| UAC ), (3.6)
2 2
0 †
where now EA:BC = EA:BC (UAC ρ0 UAC ) for a time T and the steps are justified
as follows. In the first line, we use the fact that UBC is local in the partition
A : BC. Next, by applying UAC operator to the initial state, the state of B is still
classical, i.e., represented by an orthogonal basis {|0i , |1i}. Therefore, we arrive at
the second line with the help of the flags condition [40]. We note that the states
in the second line are pure and one can equate the entanglement EA:C with the
corresponding von Neumann entropy of object A or C. As an example, by using
UAC = cos(T )1 − i sin(T )σAx σCx , one has for the first term
 
†
trB UAC |01i h01| UAC = cos2 (T ) |0i h0| + sin2 (T ) |1i h1| , (3.7)

giving an entropy − sin2 (T ) log2 (sin2 (T )) − cos2 (T ) log2 (cos2 (T )). The calculation
for the second term gives the same result, and therefore the entanglement dynamics
is simply given by

EA:BC (T ) = − sin2 (T ) log2 (sin2 (T )) − cos2 (T ) log2 (cos2 (T )). (3.8)

41
 0
Now I will show that the evolution of DAB|C is equal to EA:BC (T ) as follows.
0 0
First, note that EA:BC (T ) = EA:BC = EAB:C , where the second equality is from
the flags condition as a result of classical B. By showing that the state ρ0 has
0 0
the closest separable state σAB:C that is classical on C, we confirm that DAB|C =
0
EAB:C = EA:BC (T ). In order to do this, we recall that the flags condition also states
that the closest separable state of a quantum–classical density matrix ρ0 (classical
B) is given by
0 1 01 1 02
σAB:C = σA:C ⊗ |1iB h1| + σA:C ⊗ |0iB h0| , (3.9)
2 2
01 02 †
where σA:C and σA:C are the closest separable states respectively to UAC |01i h01| UAC
†
and UAC |10i h10| UAC . Next, we use the fact that the closest separable state of a
pure state is given by the density matrix that is de-phased in the local Schmidt basis,
01
to arrive at σA:C 02
= cos2 (T ) |01i h01|+sin2 (T ) |10i h10| and σA:C = cos2 (T ) |10i h10|+
sin2 (T ) |01i h01|. By putting the expressions together, one realises that σAB:C0
is also
classical on the side of C, hence proving the claimed statement above.
Finally, the dynamics of DAB|C is computed numerically. We note that the von
Neumann measurement on C minimising the discord is given by the basis {|0i , |1i}
for T = [0, π/8] and {|y+ i , |y− i} (the eigenbasis of σ y ) for T = [π/8, π/4].
For general non-commuting interaction Hamiltonians, one can pursue a similar
analysis with the help of the Suzuki-Trotter expansion (see Appendix A for a simple
proof). The evolution operator U is now discretised into short-time interactions of
C with A and then B (or the reversed order) as

n
U = lim e−iHBC ∆t/~ e−iHAC ∆t/~ , (3.10)
n→∞

where ∆t = τ /n → 0. Accordingly, Eq. (3.3) holds with τ replaced by ∆t. It

is now natural to ask if a scenario exists where entanglement could be increased
via interactions with a classical C at all times by exploiting the discord in the
instrumental state. The example given above is not of this sort because, although
we have DAB|C = 0 at t = 0 and τ , it is non-zero for t ∈ (0, τ ). It turns out that,
for short evolution times, the discord of the instrumental state cannot be exploited
as the theorem in the following section demonstrates.

3.3 Quantum discord is a resource for entangle-

ment gain
This section focuses on showing that quantum discord DAB|C is necessary for en-
tanglement gain in the partition A : BC. On the other hand, entanglement in the
partition A : B requires that the mediator is correlated with the principal objects

42
(might be in the form of classical correlation). We begin with the following theorem.

Theorem 3.3.1. Consider two quantum objects A and B that are interacting via a
mediator C, but not with each other. The entanglement follows

EA:BC (τ ) ≤ EA:BC (0) (3.11)

if the state of the system is quantum–classical in the partition AB : C at all times,

i.e., DAB|C (t) = 0 for t ∈ [0, τ ]. Note that all objects are allowed to interact with
their own local environments.

Proof. For simplicity, the Hamiltonian of the system is taken as H = HAC + HBC ,
where HAC = HA ⊗ HC1 and HBC = HB ⊗ HC2 .3 We note that it is straightforward
to extend the proof for more general Hamiltonians HAC = µ HAµ ⊗HCµ1 and HBC =
P
P ν ν
ν HB ⊗ HC2 .
As we assume, at all times, quantum–classical state in the partition AB : C, one
writes the initial density matrix of the system as
X
ρ0 = pc ρAB|c ⊗ |ci hc| , (3.12)
c

and the state after a short time ∆t as

X
ρ∆t = pc (∆t) ρAB|c (∆t) ⊗ |φc i hφc | , (3.13)
c

where {|ci} and {|φc i} both form orthonormal bases of the mediating system C.
Now we incorporate the local environment of all the objects under scrutiny. In
particular, the dynamics of the whole system assumes the following coarse-grained
master equation in Lindblad form

ρ∆t − ρ0 i X
= − [H, ρ0 ] + LX ρ0 , (3.14)
∆t ~ X=A,B,C

where the first term on the RHS is the coherent evolution while the second de-
scribes incoherent interactions with the local environments. We recall that LX ρ0 ≡
P k k† 1 k† k k
k QX ρ0 QX − 2 {QX QX , ρ0 }, with the operator QX only acting on object X. Note
that the strength of the interactions with environments have been absorbed in the
operator QkX .
By utilising the master equation (3.14), one obtains the following conditional
3
Note that the term HC1 can be different from HC2 in general.

43
state after a short time ∆t

pj (∆t) ρAB|j (∆t) = hφj | ρ∆t |φj i

∆t
= hφj | ρ0 |φj i − i hφj | [H, ρ0 ] |φj i
X ~
+∆t hφj | LX ρ0 |φj i . (3.15)
X=A,B,C

We note that the continuous dynamics follows by taking the limit ∆t → 0 and
applying the short time evolution successively. In this proof, terms of the order
O(∆t2 ) will not be taken into account, and we use “'” to denote equality where the
O(∆t2 ) terms are irrelevant and ignored. In this short time, we write the change
in the basis of C, up to ∆t order, as |ci → |φc i = αc |ci + βc |c⊥ i, where |c⊥ i and
|ci are orthogonal, and βc ∝ ∆t. As a result, one gets |hc|φj i|2 ' δcj , making
hφj | ρ0 |φj i ' pj ρAB|j .
The coherent part of the dynamics in Eq. (3.15) is proportional to ∆t, and
therefore, we render ∆t order terms in hφj | [H, ρ0 ] |φj i irrelevant. As a result, we
have
hφj | [H, ρ0 ] |φj i ' pj [ECj 1 HA + ECj 2 HB , ρAB|j ], (3.16)

where we define ECj 1 (C2 ) ≡ hj| HC1 (C2 ) |ji as the expectation value of energy.4 Note
that, in this way, the coherent component is simply equivalent to an effective dy-
namics that are local in A and B.
On the other hand, the incoherent part in Eq. (3.15) consists of three terms. The
first two in the summation are given by

hφj | LA ρ0 + LB ρ0 |φj i ' pj (LA + LB )ρAB|j , (3.17)

and they are local operations with respect to object A and object B. The last one
reads
XX
hφj | LC ρ0 |φj i ' pc | hj| QkC |ci |2 ρAB|c
c k
X
− pj hj| Qk† k
C QC |ji ρAB|j . (3.18)
k

Plugging all these findings to Eq. (3.15), we write the conditional state explicitly
4
For clarity, we note that one can make these expectation values dimensionless such that the
energy unit is absorbed in the terms HA and HB respectively.

44
as

hj| Qk† k
P
pj (1 − k C QC |ji ∆t)
ρAB|j (∆t) ' ρ̃AB|j
pj (∆t)
X pc X
+ | hj| QkC |ci |2 ∆t ρAB|c , (3.19)
c
pj (∆t) k

where we define

∆t
ρ̃AB|j ≡ ρAB|j − i[ECj 1 HA + ECj 2 HB , ρAB|j ]
~
+ (LA + LB )ρAB|j ∆t. (3.20)

Therefore, the density matrix ρ̃AB|j gives the evolution of ρAB|j through an effective
Lindblad master equation, where the interactions are local on the side of A and B.
By using tr(ρAB|j (∆t)) = 1, one obtains the probability
!
X
pj (∆t) = pj 1− hj| Qk† k
C QC |ji ∆t
k
X X
+ pc | hj| QkC |ci |2 ∆t, (3.21)
c k

where we have utilised the cyclic property of trace. Note that the factors for the
states ρ̃AB|j and ρAB|c in Eq. (3.19) are all real, non-negative, and sum up to unity.
Finally, the entanglement in the partition A : BC reads
X
EA:BC (∆t) = pj (∆t) EA:B (ρAB|j (∆t)) (3.22)
j
X
≤ pj EA:B (ρAB|j )
j
X X
− pj hj| Qk† k
C QC |ji ∆t EA:B (ρAB|j )
j k
XX X
+ pc | hj| QkC |ci |2 ∆t EA:B (ρAB|c ), (3.23)
j c k
X
= pj EA:B (ρAB|j ) = EA:BC (0), (3.24)
j

where we explain the steps taken above as follows. The flags condition [40] is applied
in the first line to the quantum–classical state in the partition AB : C. Next, the
inequality is obtained by using the convexity property of entanglement on the state
ρAB|j (∆t) in Eq. (3.19), and EA:B (ρ̃AB|j ) ≤ EA:B (ρAB|j ) from the monotonicity of
entanglement under local interactions. We note that the second and third lines in
Eq. (3.23) cancel out. This is apparent by inserting c |ci hc| = 1 between Qk†
P
C and
k
QC in the second line and exchanging the dummy indices c ↔ j. The last equality

45
is obtained from the flags condition applied to the initial state. The conclusion of
the theorem follows by applying the arguments above successively from t = 0 to τ
with the condition DAB|C (t) = 0 at all times.

Theorem 3.3.1 above has an important special case, which is shown by the corol-
lary below.

Corollary 3.3.1.1. For the premise of Theorem 3.3.1, where all the objects are
closed systems, we have
EA:BC (τ ) = EA:BC (0). (3.25)

Proof. In this case, the dynamics is simply given by a unitary operator with Hamil-
tonian H = HAC + HBC . The conditional state of Eq. (3.19) simplifies to

pj
ρAB|j (∆t) = ρ̃AB|j , (3.26)
pj (∆t)

where ρ̃AB|j now reads ρAB|j −i[ECj 1 HA +ECj 2 HB , ρAB|j ]∆t/~. One arrives at pj (∆t) =
pj and ρjAB (∆t) = ρ̃jAB by taking the trace of Eq. (3.26). Therefore, ρAB|j (∆t) is
simply an evolution from the initial state ρAB|j via an effective unitary operator that
is local in both A and B, making the corresponding entanglement in the partition
A : B invariant. Finally, from the flags condition applied to the initial state and the
state after ∆t one gets EA:BC (∆t) = EA:BC (0). The Corollary follows by considering
successive applications of the arguments above for a time τ .

We emphasise the generality of Theorem 3.3.1, where both the mediator and
probing objects are open to their own local environments. This matches a large
number of experimentally relevant situations – some of the important ones will be
addressed later in this thesis. It is also worth stressing that this Theorem extends the
monotonicity of entanglement under local operations and classical communication
(LOCC) [48] to the case of continuous interactions. In general, zero-discord states
are good models for classical communication as they allow for continuous projective
measurements on C that do not disturb the whole multipartite state.

3.3.1 The role of correlated mediator

Now let us consider entanglement between the principal objects A and B. In par-
ticular, we have the following Lemma.

Lemma 3.3.2. Consider the premise of Theorem 3.3.1. It follows that the gain
of entanglement in the partition A : B implies the existence of correlation in the
partition AB : C.

46
Proof. This is proof by contradiction. Let us write the initial state with an uncor-
related mediator C as ρ(0) = ρAB ⊗ ρC . This gives us EA:BC (0) = EA:B (0) from the
monotonicity of entanglement under tracing out the uncorrelated object C. Fur-
thermore, if one has quantum discord satisfying DAB|C (t) = 0 for t ∈ [0, τ ], one can
utilise the conclusion of Theorem 3.3.1 and obtain

EA:B (τ ) ≤ EA:BC (τ ) ≤ EA:BC (0) = EA:B (0), (3.27)

where the first inequality is also a consequence of the monotonicity property. This
shows that the observation of entanglement gain EA:B (τ ) > EA:B (0) implies that
either the initial state ρ(0) 6= ρAB ⊗ ρC or DAB|C (t) 6= 0 at some time during the
dynamics. In either case, the entanglement gain detects the presence of correlation
between AB and C.

Let us note that in many experimental settings, one usually assumes decoupled
objects as the initial condition, i.e., ρ0 = ρA ⊗ ρB ⊗ ρC . As this type of state is a
special case of ρAB ⊗ ρC considered in Lemma 3.3.2, one concludes that, for initial
decoupled objects, quantum discord is a necessary resource for entanglement gain
between the principal objects.

3.4 Entanglement localisation via classical medi-

ators
Before we proceed with the non-classicality detection protocols, I will show in this
section an exemplary dynamics where quantum entanglement can be mediated using
a classical mediator.
Let us again consider three qubits, where the Hamiltonian is given by Eq. (3.4),
and choose the initial state

ρ0 = 12 |ψ+ i hψ+ | ⊗ |+i h+| + 21 |φ+ i hφ+ | ⊗ |−i h−| , (3.28)

where σ x |±i = ±|±i, and |ψ+ i = √12 (|01i + |10i) and |φ+ i = √12 (|00i + |11i)
are two Bell states between subsystems AB. As the initial state in Eq. (3.28)
contains the eigenstates of HC , the system remains classical, as measured on C,
at all times. Furthermore, the classical basis is the same at all times. Yet, one
can verify that the relative entropy of entanglement between the probes is given
by EA:B (T ) = 1 − SAB (T ), where SAB (T ) is the von Neumann entropy of the AB
state at time T , and oscillates between 0 and 1. Note this means, in general, that
entanglement gain in the partition A : B does not signify the non-classicality of C
(nonzero DAB|C ). The corresponding dynamics is illustrated in Fig. 3.3. One can

47
also see from Fig. 3.3 that there is entanglement already present initially in the bigger
partition A : BC. The subsequent evolution only localises such entanglement to the
A : B partition. Let us therefore call this phenomenon entanglement localisation.

Figure 3.3: Entanglement distribution via classical ancillary system. The Hamil-
tonian is taken to be H = (σAx ⊗ 1 ⊗ σCx + 1 ⊗ σBx ⊗ σCx ) ~ω with initial state
1
2
|ψ+ i hψ+ | ⊗ |+i h+| + 12 |φ+ i hφ+ | ⊗ |−i h−|. The dynamics shows that the entan-
glement between A and B is increasing while the quantum discord DAB|C is zero at
all times.

For completeness I will provide the calculations for the dynamics in Fig. 3.3. Let
us start with the state of AB at time T
 
† †
ρAB (T ) = trC UBC UAC ρ0 UAC UBC
1 1
= ρα + ρβ , (3.29)
2 2

where we define ρα as the density matrix for a pure state |αi = cos(2T ) |ψ+ i −
i sin(2T ) |φ+ i and similarly, ρβ for |βi = cos(2T ) |φ+ i + i sin(2T ) |ψ+ i. As ρα and
ρβ are both pure states, one can calculate the corresponding entanglement in the
partition A : B by von Neumann entropy of either A or B. It is straightforward to
obtain that EA:B (ρα ) = EA:B (ρβ ) = 1, i.e., both states are maximally entangled at
all times. From the definition of REE:

EA:B (ρα ) = min (−tr (ρα log2 (σα )) − S(ρα )) . (3.30)

σα

In this case, EA:B (ρα ) = 1 and S(ρα ) = 0. Therefore, we have −tr (ρα log2 (σα )) ≥ 1
(*), and similarly, −tr (ρβ log2 (σβ )) ≥ 1 (**). Now we show that there exists a single

48
separable state that achieves the minima in both (*) and (**). By choosing

1 1
σ= |ψ+ i hψ+ | + |φ+ i hφ+ | , (3.31)
2 2

one directly verifies that −tr (ρα log2 (σ)) = −tr (ρβ log2 (σ)) = 1. Now, we calculate
the entanglement as follows
  
1 1
EA:B (T ) = min −tr ρα + ρβ ||σ − SAB (T )
σ 2 2
 
1 1
= min (−tr (ρα ||σ)) + (−tr (ρβ ||σ)) − SAB (T )
σ 2 2
= 1 − SAB (T ), (3.32)

where we have used (3.31) as the minimising separable state. See Appendix B for
the dynamics of EA:BC and DAB|C , and for a general prescription of entanglement
localisation. We note that our entanglement localisation proposal has been realised
experimentally [49].
We note that similar considerations have been presented in Ref. [50] to provide
a counter-example to the impossibility of entanglement gain via LOCC. However,
as mentioned above, the partition A : BC is entangled already from the beginning
(in our example we have EA:BC (0) = 1). Therefore, the entanglement localisation
emphasises that the ancillary systems within the framework of LOCC, here C, are
not allowed to be initially correlated with the principal objects A and B, even if the
correlations are classical.

3.5 Non-classicality detection protocols

Based on the acquired knowledge in previous sections, I will now present two meth-
ods for detecting quantum property of an inaccessible object (nonzero discord). Note
that, in this section, we will not assume the form of the initial state such that it is
beneficial for general experimental purposes.

3.5.1 Via entanglement breaking channel

We note that the only way of gaining entanglement in subsystem AB via classical
C is to localise it from the already present entanglement in A : BC. This is a
consequence of Theorem 3.3.1 and reinforces its role as a proper generalisation of
the monotonicity of entanglement to continuous interactions. Namely,

EA:B (τ ) ≤ EA:BC (τ ) ≤ EA:BC (0). (3.33)

49
Now, if we ensure by operating on the probes only that the initial entanglements
coincide, i.e. EA:BC (0) = EA:B (0), entanglement gain in system AB is only possi-
ble due to nonzero discord DAB|C . As we are interested in observing entanglement
gain, it is natural to start with as small entanglement as possible. This leads us
to propose the application of an entanglement-breaking channel to one of the prin-
cipal systems, at time t = 0. Indeed, after application of the channel on A, we
have EA:B (0) = EA:BC (0) = 0. In a more concrete example, the channel is a von
Neumann measurement. An arbitrary measurement is allowed and experimentalist
should choose the one having potential for biggest entanglement gain. Note that
the measurement results need not be known. Our method is symmetrical, i.e., the
entanglement-breaking channel can be applied on B. This is a consequence of the
symmetry in Theorem 3.3.1, leading to EA:B (τ ) ≤ EB:AC (τ ) ≤ EB:AC (0). Together
with the condition after the channel, EA:B (0) = EB:AC (0) = 0, they give the same
conclusion. The non-classicality detection method is illustrated and summarised
in Fig. 3.4a. Note that entanglement estimation in step (iii) can be realised with
entanglement witnesses [6, 51], rendering state tomography unnecessary.

3.5.2 Via initial entropies

Another method for revealing the non-classicality can be derived in terms of the
purity (or entropy) of the probes. For this purpose, we present the following Lemma.

Lemma 3.5.1. Provided the premise of Theorem 3.3.1, the following inequality holds

EA:B (τ ) ≤ SA (0) + SB (0), (3.34)

where the entanglement quantifier here is REE and SX denotes the von Neumann
entropy of object X.

Proof. Let us begin here with Eq. (3.33), that is EA:B (τ ) ≤ EA:BC (τ ) ≤ EA:BC (0).
Now we will provide an upper bound to the initial entanglement EA:BC (0) in terms
of initial von Neumann entropies of the accessible objects A and B. First, note that
REE is bounded by mutual information [38], which in our case reads EA:BC ≤ IA:BC .
We also have IA:BC ≤ SA + SB − SAB|C from the sub-additivity of von Neumann
entropy. As the state of object C is classical, the conditional entropy follows SAB|C ≥
0, and therefore
EA:B (τ ) ≤ EA:BC (0) ≤ SA (0) + SB (0). (3.35)

Any violation of inequality (3.35) then reveals the non-classicality of C. See

Fig. 3.4b for a summary of this detection scheme.

50
Figure 3.4: Summary of protocols for revelation of non-classicality of an inaccessible
object C. (a) Protocol via entanglement breaking channel involves the following
steps: (i) entanglement breaking channel on object A (or object B), e.g., von Neu-
mann measurement; (ii) dynamics of the whole system; (iii) estimation of entan-
glement between A and B. In this scheme, nonzero entanglement implies positive
discord DAB|C . (b) Protocol via initial entropies has similar steps with the excep-
tion of measurement of initial entropies for objects A and B. This scheme reveals
positive DAB|C if the entanglement between A and B is larger than the sum of initial
entropies, i.e., EA:B > SA (0) + SB (0). Note that for both protocols we have mini-
malistic assumptions. All objects can be open systems and no assumption is made
regarding the initial state nor the explicit form of the Hamiltonian HAC + HBC .

3.6 Minimalistic assumptions and applications

We have proposed an entanglement-based criteria for the inference of quantumness
of an inaccessible object. Our protocols are fully non-disruptive of the state of the
system to probe, and rely on only weak assumptions on the nature of the inter-
actions involved. They are also robust against decoherence. These features make
our proposal suitable to address non-classicality at many levels, from technological
platforms such as quantum optomechanics (Chapter 8) to the problem of detect-
ing quantumness in living organisms (Chapter 7) and the fundamental questions
on the nature of gravity (Chapter 6). For example, the gravity scheme puts C as
a gravitational field coupling massive objects A and B, which are mutually non-
interacting. By determining experimentally the entanglement gain between A and
B one would conclude, according to our scheme, the non-classical nature of the
gravitational field between them. That is, if we were to embed into the quantum
formalism description of the masses and the field, the dynamics would be driving
the field through non-orthogonal states as this is required for the quantum discord
DAB|C to be non-zero. A similar analysis can also be done for remote quantum dots
in a solid-state substrate [52] or spin-chain systems like in Ref. [53], as their physics
also naturally distinguishes a mediating object that is inaccessible, e.g., locations of
unpaired spins are unknown in a sample. We have performed detailed analysis of
the latter in Ref. [54].

51
3.7 Summary
In this study we considered two principal objects that are only indirectly interacting,
via a mediator. We have shown that quantum discord (a form of quantum property
of the mediator) is a resource for entanglement gain. This prompted us to propose
schemes for assessing quantumness of an inaccessible object (the mediator). The
schemes are based on monitoring entanglement dynamics between the principal ob-
jects, which serve as probes over which we have control. Our method is robust and
experimentally friendly as it allows the controlled objects and the inaccessible one
to be open systems, makes no assumptions about the initial state, dimensionality of
involved Hilbert spaces, and details of the interaction Hamiltonian.

52
Chapter 4

Non-decomposability of evolution
& extreme gain of correlations

In this chapter, the amount of distributed correlations will be studied.1 I show that
non-commutativity of interaction Hamiltonians (or, in a more general setting, non-
decomposability of time evolution) is a necessary resource for having high gain of
correlations between two objects. From the extreme gain of correlations one can
then detect the non-commutativity, which shows another level of quantum property
that is present in the system. This chapter starts with a motivation and a construc-
tion of a general setting that will be considered herein. I will then proceed to show
that for decomposable dynamics the correlations between two principal objects are
bounded by a correlation capacity of a mediating system. This applies for a plethora
of correlation quantifiers, some of which will be presented in this chapter. Next, I
will discuss a special scenario where all the objects involved are open to their local
environments. Finally, the origin of the possible extreme gain of correlations will be
covered.

1
Parts of this chapter are reproduced from our published article of Ref. [55], c [2019] American
Physical Society. Where applicable, changes made will be indicated.

53
4.1 Motivation and objectives
All classical observables are functions of positions and momenta. Since there is
no fundamental limit on the precision of position and momentum measurement in
classical physics, all classical observables are, in principle, measurable simultane-
ously. Quite differently, the Heisenberg uncertainty principle forbids simultaneous
exact knowledge of quantum observables corresponding to position and momentum.
The underlying non-classical feature is their non-commutativity: Any pair of non-
commuting observables cannot be simultaneously measured to arbitrary precision,
as first demonstrated by Robertson in his famous uncertainty relation [56]. Other
examples of non-classical phenomena with underlying non-commutativity of observ-
ables include violations of Bell inequalities [57, 58] or, more generally, non-contextual
inequalities; e.g., see [59]. Here we describe a method to detect non-commutativity of
interaction Hamiltonians, and generally non-decomposability of temporal evolution,
from the dynamics of correlations.
Consider the situation depicted in Fig. 4.1, where the probing systems A and
B do not interact directly but only via the mediator C; i.e., there is no Hamil-
tonian term HAB . In general, we allow all objects to be open systems and study
whether the evolution operator cannot be represented by a sequence of operations
between each probe and the mediator, i.e., ΛABC = ΛBC ΛAC or in reverse order.
For the special case in which all the systems are closed, non-decomposability im-
plies non-commutativity of interaction Hamiltonians, i.e., [HAC , HBC ] 6= 0. Indeed,
for commuting Hamiltonians, the unitary evolution operator is decomposable into
UBC UAC , where, for example, UAC = exp(−itHAC /~). We show that for decom-
posable evolution, correlations between A and B are bounded. We also show with
concrete dynamics generated by non-commuting Hamiltonians that these bounds
can be violated. The bounds derived depend solely on the dimensionality of C
and not on the actual form of the evolution operators. Hence, these operators can
remain unknown throughout the assessment. This is a desired feature, as experi-
menters usually do not reconstruct the evolution operators via process tomography.
It also allows applications of the method to situations where the physics is not un-
derstood to the extent that reasonable Hamiltonians or Lindblad operators can be
written down. Furthermore, the assessment does not depend on the initial state
of the tripartite system and does not require any operations on the mediator. It
is therefore applicable to a variety of experimental situations; Refs. [52, 53, 60, 61]
provide concrete examples.
Below I shall begin by presenting the general bounds on the amount of corre-
lations one can establish if the evolution is decomposable. It is shown that these
bounds are generic and hold for a large number of correlation quantifiers. We then

54
Figure 4.1: General setting for detecting non-decomposability of time evolution.
High dimensional objects A and B are interacting with a low dimensional object C,
but not with each other. In general, we allow all objects to be open to their local
environments. The interaction Hamiltonians are given by HAC and HBC , which
describe the coherent part of the dynamics. In the text, we show that large gain
of correlation between A and B, exceeding certain thresholds (functions of the di-
mensionality of C), implies non-decomposability of time evolution, which makes the
latter crucial for substantial correlation distribution. Note that a lot of correlation
quantifiers are useful in our method. This setup is different from the one considered
in Fig. 3.1 of Chapter 3. For non-decomposability detection, here the dimension of
C is assumed to be known and, as we will show in the text, small value of such
dimension is preferable.

calculate concrete bounds on exemplary quantifiers. In Chapter 8, I will show how

they can be violated in a system of two fields coupled by a two-level atom. We
discuss the origin of the violation in terms of “Trotterized” evolution, where a vir-
tual particle is exchanged between A and B multiple times if the Hamiltonians do
not commute but only once if they do commute. Finally, we focus on immediate
applications in quantum information and discuss the consequences of our findings
for correlation distribution protocols and dimension witnesses.

4.2 Correlation capacity bound

Consider the setup illustrated in Fig. 4.1. System C, with finite dimension dC , is
mediating interactions between higher-dimensional systems A and B. For simplicity
we take dA = dB > dC . We assume that there is no direct interaction between
A and B, such that the Hamiltonian of the whole tripartite system is of the form
HAC +HBC (local Hamiltonians HA , HB , and HC included). Our bounds follow from
a generalization of the following simple observation. Consider, for the moment, the
relative entropy of entanglement as the correlation quantifier [29]. If the evolution
is decomposable, it can be written as ΛBC ΛAC , or in reverse order. Therefore, it
is as if particle C interacted first with A and then with B, a scenario similar to
that in Refs. [7, 16–20]. The first interaction can generate at most log2 (dC ) ebits of
entanglement, whereas the second, in the best case, can swap all this entanglement.
In the end, particles A and B gain at most log2 (dC ) ebits. The bound is indeed
independent of the form of interactions. Furthermore, it is intuitively clear, as this
is just the “quantum capacity” of the mediator. Below we will generalise this bound

55
for varied initial states and more correlation quantifiers.

4.2.1 Arbitrary initial states

Let us consider correlation quantifiers defined in the so-called “distance” approach [29,
38]. The idea is to quantify correlation QX:Y in a state ρXY as the shortest dis-
tance D(ρXY , σXY ) from ρXY to a set of states σXY ∈ S without the desired cor-
relation property, i.e., QX:Y ≡ inf σXY ∈S D(ρXY , σXY ). For example, the relative
entropy of entanglement is given by the relative entropy of a state to the set of
disentangled states [29]. It turns out that most such quantifiers are useful for the
task introduced here. The conditions we require are that (i) S is closed under
local operations ΛY on Y , (ii) D(Λ[ρ], Λ[σ]) ≤ D(ρ, σ) (monotonicity), and (iii)
D(ρ0 , ρ1 ) ≤ D(ρ0 , ρ2 ) + D(ρ2 , ρ1 ) (triangle inequality). They are sufficient to prove
Theorem 4.2.2 below.
For completeness let us begin with a useful lemma.

Lemma 4.2.1. Consider correlation between two parties X and Y , as measured by

a correlation quantifier QX:Y that is non-increasing under local operations on the
side of Y . It follows that QX:Y stays constant under all reversible operations on Y ,
e.g., tracing-out uncorrelated objects in Y .

Proof. Let us suppose that one performs an operation on party Y . Based on the
monotonicity property, QX:Y cannot increase in value. Furthermore, the value also
cannot decrease for a reversible operation, because this means QX:Y can increase
by reversing the process. Therefore, QX:Y can only be invariant. As tracing-out
uncorrelated objects is a reversible process, the conclusion follows.

The main theorem in this chapter is proven as follows.

Theorem 4.2.2. Consider a distance-based quantifier of correlation between party

X and party Y , defined as QX:Y ≡ inf σXY ∈S D(ρXY , σXY ) with properties

(i) S is closed under local maps ΛY on Y ;

(ii) D(Λ[ρ], Λ[σ]) ≤ D(ρ, σ); and

(iii) D(ρ0 , ρ1 ) ≤ D(ρ0 , ρ2 ) + D(ρ2 , ρ1 ).

It follows that for a decomposable dynamical operator, i.e., ΛABC = ΛBC ΛAC , one
has
QA:B (t) ≤ IAC:B (0) + sup QA:C , (4.1)
|ψi

where we define here IAC:B (0) ≡ inf σAC ⊗σB D(ρ, σAC ⊗ σB ), use ρ as the initial
density matrix, and the supremum in the last term is over pure states {|ψi} of AC.

56
Proof. First, we show that QX:Y is monotone under local operations ΛY . This is
done as follows

QX:Y (ΛY [ρXY ]) = inf D(ΛY [ρXY ], σXY )

σXY ∈S
0
≤ D(ΛY [ρXY ], ΛY [σXY ])
0
≤ D(ρXY , σXY )
= QX:Y (ρXY ), (4.2)

0
where σXY is a state in S closest to ρXY . The second line in (4.2) is due to property
0
(i) and that ΛY [σXY ] might not be the the closest to ΛY [ρXY ], while the third line
follows from property (ii). This allows us to use the property in Lemma 4.2.1.
Next, we have the following arguments

QA:B (t) ≤ QA:BC (ΛBC ΛAC [ρ]) (4.3)

≤ QA:BC (ΛAC [ρ]) (4.4)
≤ D (ΛAC [ρ], µ) (4.5)
0 0


≤ D ΛAC [ρ], ΛAC [σAC ] ⊗ σB
0
] ⊗ σB0 , µ


+ D ΛAC [σAC (4.6)
0
≤ D(ρ, σAC ⊗ σB0 )
0
] ⊗ σB0 , µ


+ D ΛAC [σAC (4.7)
0
= IAC:B (0) + QA:BC (ΛAC [σAC ] ⊗ σB0 ) (4.8)
0
= IAC:B (0) + QA:C (ΛAC [σAC ]) (4.9)
≤ IAC:B (0) + sup QA:C , (4.10)
|ψi

where the steps are explained as follows. We utilise that QX:Y is monotone under
local maps on Y (in this case, tracing out C) in line (4.3). Similarly, in (4.4), QA:BC
is monotone under the operation ΛBC . The inequality (4.5) utilises the definition of
QA:BC as a distance-based quantifier and holds for any state µ ∈ SA:BC . The triangle
inequality (iii) confirms line (4.6). We note that the first term in (4.6) is independent
0
of µ and one can choose any state σAC and σB0 at this point. Property (ii) is utilised
0
in equality (4.7). For line (4.8), the state σAC ⊗ σB0 is chosen as the closest product
0
state to ρ, while µ ∈ SA:BC as the closest state to ΛAC [σAC ] ⊗ σB0 . The equality
(4.9) is obtained as QA:BC is invariant under the partial trace of the decoupled state
σB0 . The final line follows from a property that a bipartite correlation quantifier is
maximal on pure states if it is non-increasing under local operations on at least one
party [62].

Note that although the relative entropy does not satisfy (iii) it still follows The-

57
orem 4.2.2. This is the result of the following lemma.

Lemma 4.2.3. For a distance measure based on the relative entropy the conclusion
in Theorem 4.2.2 follows.

Proof. First, we show the following identity

S(ρ||σX ⊗ σY ) = tr(ρ log ρ − ρ log σX ⊗ σY )

= tr(ρ log ρ − ρ log ρX ⊗ ρY )
+tr(ρ log ρX ⊗ ρY − ρ log σX ⊗ σY )
= S(ρ||ρX ⊗ ρY ) + S(ρX ||σX )
+S(ρY ||σY ), (4.11)

where ρX and ρY are the reduced states of ρ. We have also used, e.g., the relation
tr(ρ log σX ⊗ σY ) = tr(ρX log σX ) + tr(ρY log σY ).
While relative entropy is known to follow property (ii) [63], it does not satisfy
property (iii). In this case, one starts with line (4.4) in Theorem 4.2.2 and obtains

QA:BC (ΛAC [ρ]) = inf S (ΛAC [ρ]||µ) (4.12)

µ∈SA:BC
≤ S(ΛAC [ρ]||µAC ⊗ µB ) (4.13)
= S(ΛAC [ρ]||ρ0AC ⊗ ρ0B )
+S(ρ0AC ||µAC ) + S(ρ0B ||µB ) (4.14)
= IAC:B (ΛAC [ρ]) + QA:C (ρ0AC ) (4.15)
≤ IAC:B (0) + sup QA:C , (4.16)
|ψi

where ρ0AC and ρ0B are the reduced states of ΛAC [ρ]. We justify the steps above as
follows. In line (4.13), we use a state of the form µAC ⊗ µB that is in the set of
SA:BC . The identity of (4.11) confirms line (4.14). From the definition of mutual
information as the distance from a state to its marginals [38], one arrives at equality
(4.15). Note that µAC ∈ SA:C has been chosen as the closest to ρ0AC and µB = ρ0B .
The final line is obtained from the monotonicity of mutual information under the
local map ΛAC and the supremum of QA:C is attained over pure states.

Correlations between probe A and probe B are therefore bounded by the maximal
achievable correlation with the mediator, sup|ψi QA:C . The additional term IAC:B (0)
in Eq. (4.1) reduces to the usual mutual information IAC:B (0) if D(ρXY , σXY ) is the
relative entropy distance [38] and characterizes the amount of total initial correla-
tions between one of the probes and the rest of the system. Note that the bound
is independent of time. This can be seen as a result of the effective description of

58
such dynamics given by ΛBC ΛAC . The particle C is exchanged between A and B
only once, independently of the duration of the dynamics.
In a typical experimental situation the initial state can be prepared as completely
uncorrelated ρ = ρA ⊗ρB ⊗ρC , in which case Theorem 4.2.2 simplifies and the bound
is given solely in terms of the “correlation capacity” of the mediator:

QA:B (t) ≤ sup QA:C . (4.17)

|ψi

Clearly, the same bound holds for initial states of the form ρ = ρAC ⊗ ρB . In
Section 4.2.2 we show that, with this initial state, Eq. (4.17) holds for any correlation
quantifier that is monotonic under local operations ΛBC , not necessarily based on
the distance approach, e.g., any entanglement monotone.

4.2.2 Uncorrelated initial states

Here we prove a bound on correlations for particular initial states where one of the
probing objects is decoupled.

Theorem 4.2.4. Consider the initial density matrix of the system being in the form
ρ = ρAC ⊗ ρB . For a decomposable dynamical operator ΛABC = ΛBC ΛAC , it follows
that
QA:B (t) ≤ sup QA:C , (4.18)
|ψi

where Q is any correlation quantifier that is monotone under local maps ΛBC .

Proof. The proof is done with the following arguments

QA:B (t) ≤ QA:BC (t) ≤ QA:BC (ΛAC [ρ])

= QA:C (ΛAC [ρ]) ≤ sup QA:C , (4.19)
|ψi

where the steps are justified as follows. The first inequality is obtained as tracing
out party C (which might be correlated with AB in general) is a local operation on
the side of BC. The monotonicity of the quantifier under ΛBC confirms the second
argument. Party B stays uncorrelated after the application of ΛAC on the initial
density matrix ρAC ⊗ ρB . As a result, the equality follows by utilising Lemma 4.2.1.
The quantifier QA:C is maximised on pure states, giving the final inequality.

We note that for initial states that are close to ρ = ρAC ⊗ ρB one can utilize the
continuity of the von Neumann entropy [64] and see that IAC:B (0) in Eq. (4.1) is
indeed small. We can also ensure that the initial state is of the form ρ = ρAC ⊗ ρB
by performing a correlation breaking channel on B first. One example of such a

59
channel is a measurement in the computational basis followed by a measurement in
some complementary (say Fourier) basis. This implements the correlation breaking
channel (11AC ⊗ ΛB )(ρABC ) = ρAC ⊗ d11B .2 In this way, our method does not require
any knowledge of the initial state and any operations on the mediator, similar in
spirit to the detection of quantum discord of inaccessible objects in Ref. [45]. We
now move to concrete correlation quantifiers and their correlation capacities.

4.3 Correlation capacity of particular quantifiers

We provide four correlation quantifiers (whose definitions will be repeated here for
convenience), which capture different types of correlations between quantum par-
ticles. They are mutual information, classical correlation, quantum discord, and
negativity. All of them are shown to be useful in detecting non-decomposability.
Mutual information is a measure of total correlations [43] and is defined as IX:Y =
SX + SY − SXY , where, e.g., SX is the von Neumann entropy of subsystem X. It can
also be seen as a distance-based measure with the relative entropy as the distance
and a set of product states σX ⊗ σY as S [38]. The supremum in Eq. (4.17) is
attained by the state (recall that dA > dC ),

dC
1 X
|Ψi = √ |aj i|cj i, (4.20)
dC j=1

where |aj i and |cj i form orthonormal bases. One finds sup|ψi IA:C = 2 log2 (dC ).
An interesting quantifier in the context of non-classicality detection is the classi-
cal correlation in a quantum state. It is defined as mutual information of the classical
state obtained by performing the best local von Neumann measurements on the orig-
inal state ρ [66], i.e., CX:Y = supΠX ⊗ΠY IX:Y (ΠX ⊗ ΠY (ρ)), where ΠX ⊗ ΠY (ρ) =
P
xy |xyi hxy| ρ |xyi hxy|, and |xi, |yi form orthonormal bases. The supremum of
mutual information over classical states of AC is log2 (dC ).
Quantum discord is a form of purely quantum correlations that contain quantum
entanglement. It can be phrased as a distance-based measure. In particular, we
consider the relative entropy of discord [38], also known as the one-way deficit [39].
It is an asymmetric quantity defined as DX|Y = inf ΠY S(ΠY (ρ)) − S(ρ), where ΠY is
a von Neumann measurement conducted on subsystem Y . The relative entropy of
discord is maximized by the state (4.20), for which we have sup|ψi DA|C = log2 (dC ).
Our last example is negativity, a computable entanglement monotone [31–34].
For a bipartite system negativity is defined as NX:Y = (||ρTX ||1 − 1)/2, where ||.||1
2
In this context one might ask whether there exists a channel such that (11AC ⊗ ΛB )(ρABC ) =
ρAC ⊗ ρB , where ρAC and ρB are reduced density matrices. Such a channel does not exist, e.g., it
would be nonlinear. See also Ref. [65].

60
denotes the trace norm and ρTX is a matrix obtained by partial transposition of ρ
with respect to X. Negativity is maximized by the state (4.20), and the supremum
reads sup|ψi NA:C = (dC − 1)/2.
Clearly, many other correlation quantifiers are suitable for our detection method
because the assumptions behind Eqs. (4.1) and (4.17) are not demanding. For
those presented here, see Table 4.1 for a summary. In fact, one may wonder which
correlations do not qualify for our method. A concrete example is the geometric
quantum discord based on p-Schatten norms with p > 1, as it may increase under
local operations on BC [67, 68].

Table 4.1: Summary of particular correlation quantifiers and the corresponding

capacities. Note that we have assumed dC < dA , dB .
Quantifier Correlation capacity
Mutual information 2 log2 (dC )
Classical correlation log2 (dC )
Quantum discord log2 (dC )
Negativity (dC − 1)/2

4.4 Local environment & non-commutativity of

Hamiltonians
Finally, we wish to discuss a scenario where all the objects are open to their own
local environments, as realised, e.g., in [61]. We take the evolution following the
master equation in Lindblad form,

i X
ρ̇ = − [HAC + HBC , ρ] + LX ρ, (4.21)
~ X=A,B,C
X 1
LX ρ = QkX ρQk† k† k
X − {QX QX , ρ},
k
2

where the last term in (4.21) is the incoherent part of the evolution and LX de-
scribes interactions of system X with its local environment, i.e., the operators
QkX act on system X only. We denote LAC = −(i/~)[HAC , ·] + LA + LC and
LBC = −(i/~)[HBC , ·] + LB . One readily verifies that if [HAC , HBC ] = 0 and
[LC , HBC /~] = 0, we have commuting Lindblad operators [LAC , LBC ] = 0. Note
that, if one includes LC in LBC instead, the second condition for commuting Lind-
blad operators now reads [LC , HAC /~] = 0. For commuting Lindbladians, the cor-
responding evolution decomposes as ΛBC ΛAC , or in reverse order. Therefore, our
bounds apply accordingly. Their violation implies that either the Hamiltonians do

61
not commute or the operators describing dissipative channels on C do not commute
with HAC and HBC . In particular, if C is kept isolated so that its noise can be
ignored, the violation of our bounds is solely the result of the non-commutativity of
the interaction Hamiltonians.

4.5 Intuition behind excessive gain of correlations

In Chapter 8, I present high correlation gain (violating the capacity bounds) between
two orthogonally-polarised cavity fields that are individually interacting with a two-
level atom. Here, the origin of the violation will be explained.
Since the total Hamiltonian is of the form HAC + HBC , the Suzuki-Trotter ex-
pansion of the resulting evolution is particularly illuminating,
 n
i ~t (HAC +HBC ) i ∆t HBC i ∆t HAC
e = lim e ~ e ~ , (4.22)
n→∞

where ∆t = t/n. If Hamiltonians do not commute, it is necessary to think about

Eq. (4.22) as n sequences of pairwise interactions of C with A followed by C with B,
each for a time ∆t. Each pair of interactions can only increase correlations up to the
correlation capacity of the mediator, but their multiple use allows the accumulation
of correlations beyond what is possible with commuting Hamiltonians. Recall that,
in the latter case, we deal with only one exchange of system C, independently of the
duration of dynamics. We stress that Trotterization is just a mathematical tool and
in the laboratory system C is continuously coupled to A and B. It is rather as if
a virtual particle C were transmitted multiple times between A and B, interacting
with each of them for a time ∆t.

4.6 Summary
One of the most elementary non-classical traits of quantum observables is their
non-commutativity. In this chapter, we have linked non-commutativity of interac-
tion Hamiltonians (more generally, non-decomposability of time evolution) to the
amount of correlations that can be created in the associated dynamics. This led us
to a method for detection of non-decomposability of evolution in a scenario where a
mediator C mediates interactions between two probing objects A and B (all these
objects can interact with their local environments). The Hamiltonians or Lindblad
operators can remain unknown throughout the assessment, we only require knowl-
edge of the dimension of the mediator. Furthermore, no operation on C is necessary
at any time, which makes this strategy experimentally friendly. Technically, under
the assumption of decomposable evolution, we derived upper bounds on correlations

62
between the probes. Non-decomposability is detected by observing violation of cer-
tain bounds on AB correlations. A plethora of correlation quantifiers are helpful
in our method, e.g., quantum entanglement, discord, mutual information, and even
classical correlation. We also provided an intuitive explanation in terms of multiple
exchanges of a virtual particle which lead to the possible excessive accumulation of
correlations.

63
Chapter 5

Speed of distribution of quantum

entanglement

This chapter investigates the rate at which quantum entanglement is generated. I will
begin with a motivation comprising the well-known quantum speed limit and general
scenarios that will be considered here. Entangling speed limit for the direct inter-
action setting will be examined. The following section then formulates the limit for
its counterpart, the indirect interaction setting, with exemplary dynamics saturating
the speed. Next, I will revisit the entanglement localisation introduced in Chapter 3
and provide an upper bound for entanglement gain in terms of initial correlations
present in the initial state. I will also present the implications of the results and a
simple application to charging quantum batteries.

65
5.1 Motivation and objectives
A quantum state requires finite time to evolve into a state distinguishable from its
initial form. This limitation on the shortest time is widely known as the quantum
speed limit (QSL). The lower bound on time required for the evolution was first
obtained by Mandelstam and Tamm [69]. Since then, advances for the time bound
include unitary evolution of pure states [70–72] and for mixed states [73–76]. These
results have many applications, e.g., to the investigation of the rate of entropy
increase [77], the limit in quantum metrology [78], the limit on computation [79,
80], and the bound on charging power of quantum batteries [81–83]. Furthermore,
previous studies have also shown the application of QSL to classical domain [84, 85].
We will use the time bound that is given by the unified quantum speed limit
[86, 87], which, for an evolution of a state ρ to another state σ, reads

Θ(ρ, σ)
τ (ρ, σ) = ~ , (5.1)
min{hHi, ∆H}
p√ √

where Θ(ρ, σ) = arccos(F(ρ, σ)) is the Bures angle, F(ρ, σ) = tr ρσ ρ is the
Uhlmann root fidelity [26, 27], hHi = tr(Hρ)−Eg is the average energy with respect
p
to the ground state energy of H, and ∆H = tr[H 2 ρ] − tr[Hρ]2 is the standard
deviation of energy (SDE). Note that for pure states, the Bures angle is given by
the Fubini-Study distance, i.e., Θ(|ψi hψ| , |φi hφ|) = arccos |hψ|φi| [88–90].
In this chapter, we study the quantum speed of system AB with direct and
indirect (through a mediator) continuous interactions, see Fig. 1.1 and Fig. 1.2b.
The direct interaction case considers the objects A and B with Hamiltonian HAB
(local Hamiltonians HA and HB included). On the other hand, the interactions
between A and B are mediated by an object C, such that the total Hamiltonian is
of the form HAC + HBC (again, local terms HA , HB , and HC included). During the
evolution, we study properties of the principle objects A and B, such as quantum
entanglement between them and charging power. The latter has been defined as
the speed to evolve the quantum state of AB from |00i to |11i (fully charged).
We compare the entanglement creation and the charging power between the direct
and indirect interaction scenarios in terms of speed. In particular, we investigate
whether one setup has an advantage over the other and the factors that are relevant
for optimal speed.

5.2 Preliminaries
It can be seen from the QSL of Eq. (5.1) that the first relevant quantity is the
fidelity between the initial state ρAB (0) and the target state ρAB (t). The second

66
quantity is min{hHi, ∆H}, which involves the average energy or SDE. It is clear
that the change in Hamiltonian H → kH (e.g., having more energy) results in
hHi → khHi and ∆H → k∆H for a constant k. This means one can always speed
up the evolution of the quantum state by, for example, providing a higher amount
of energy. Therefore, to put all processes on equal footing, we fix min{hHi, ∆H} to
be ~Ω. For ease of computation, let us rewrite Eq. (5.1) as

Θ(ρ, σ)
Γ(ρ, σ) = , (5.2)
min{hM i, ∆M }

where Γ = τ Ω is a dimensionless time bound, hM i and ∆M are dimensionless

quantities representing the average energy and SDE, respectively, per unit ~Ω. In
this way, we fix the frequency Ω and set the resource on equal footing by having the
condition min{hM i, ∆M } = 1.1
We will use some correlation quantifiers (will be recalled briefly below) through-
out this chapter. For extensive definitions, see Chapter 2. As a quantifier for
entanglement, we mostly use negativity, which is a computable entanglement mono-
tone [31–34]. For two objects X and Y , the negativity is denoted by NX:Y whose
maximum is given by (d − 1)/2, where d = min{dX , dY } with dX (dY ) being the
dimension of the object X(Y ). This maximum entanglement is given by pure states
of the form
d
1 X
|Ψmax i = √ |xj i|yj i, (5.3)
d j=1

where |xj i and |yj i form orthonormal bases. We will also use the relative entropy
of entanglement (REE) [91], denoted as EX:Y . The maximum of REE is given by
log2 (d). We consider the relative entropy of discord [38], which is also referred to as
the one-way quantum deficit [39]. It is an asymmetric quantity denoted as ∆X|Y .
Similar to REE, maximum discord is given by log2 (d). To quantify the amount
of total correlation, we use the mutual information IX:Y [43]. Finally, on some
occasions, we use conditional entropy SX|Y .

5.3 Entangling speed limit: Direct interactions

Let us now consider the quickest way to entangle our principal objects A and B
with direct interaction. For an exemplary dynamics, assume for the moment, that
all the objects are qubits. Starting with an initially disentangled states, e.g., |00i,
the optimal entangling speed is obtained with the following Hamiltonian

HAB = ~Ω σAx ⊗ σBx , (5.4)

1
Throughout this chapter, this will be referred to as resource equality.

67
where σ x is the x-Pauli matrix and Ω denotes the frequency unit. It is easy to see
that the state at time t takes the form |ψ(t)i = cos(Ωt) |00i − i sin(Ωt) |11i, and
therefore the state of AB is oscillating between the disentangled states |00i or |11i
√
and a maximally entangled state (|00i − i |11i)/ 2.
We note that in small time, the component |11i grows ∝ ∆t, which means that
entanglement, as quantified by negativity, NA:B increases immediately. In other
words, the rate of entanglement follows ṄA:B (0) > 0. It is also easy to confirm that
min{hM i, ∆M } = 1 in this example, meaning that the time bound is simply given by
the Bures angle, i.e., Γ(ρAB (0), ρAB (T )) = Θ(ρAB (0), ρAB (T )).2 The corresponding
dynamics of entanglement and Bures angle is illustrated in Fig. 5.1. One can see
that this dynamics saturates the time bound, i.e., the time equals the time bound
Θ (dashed line in Fig. 5.1).

Figure 5.1: Exemplary dynamics showing maximum entangling speed between two
qubits. The Hamiltonian is given by HAB = ~Ω σAx ⊗ σBx with initial state |00i.
Maximum negativity (solid curve) is achieved in t = π/4Ω.

From the simple example above, it is apparent to see the generalisation of the
bound for the case where the objects under consideration are d-dimensional. In
particular, by having the resource equality, the direct bound reads
 
1
Γdi = arccos |h00|Ψmax i| = arccos √ . (5.5)
d

For the actual dynamics, one has to figure out the Hamiltonian that both gives the
resource equality and evolves the initial state to a state with given fidelity in the
shortest possible time.
2
We use T = Ωt as a dimensionless time variable.

68
5.4 Entangling speed limit: Indirect interactions
It is now reasonable to ask whether the dynamics in Fig. 5.1 can be sped up or at
least replicated through indirect interactions. Although it is definitely possible to
have entanglement growth from zero to maximum, we will show that it is impossible
to beat the speed set by the direct interaction scenario with initial state |00i for
system AB.
Let us begin with the following theorem, which investigates the initial rate of
entanglement gain in the partition A : B.
Theorem 5.4.1. Consider the indirect interaction scenario, where an ancilla C is
mediating interactions between objects A and B. We allow all the objects to be open
to their own local environments. If the initial state of the tripartite system is of the
form ρ0 = ρAB ⊗ρC , then the rate of any entanglement monotone reads ĖA:B (0) ≤ 0.
Proof. Let us start with the evolution of the tripartite state following the Lindblad
form

ρ∆t − ρ0 i X
= − [H, ρ0 ] + LX ρ0 , (5.6)
∆t ~ X=A,B,C
X 1 k† k
LX ρ 0 = QkX ρ0 Qk†
X − {QX QX , ρ0 },
k
2

where the first term in (5.6) is the coherent part of the evolution, whereas the second
term is the incoherent part and LX describes interactions of system X with its local
environment, i.e., the operators QkX act on system X only. Without loss of generality
we assume the Hamiltonians of the form H = HA ⊗ HC1 + HB ⊗ HC2 , where HC2 can
be different from HC1 . Note that the generalisation of the proof to Hamiltonians
H = µ HAµ ⊗ HCµ1 + ν HBν ⊗ HCν 2 is straightforward.
P P

The state of AB at time ∆t is calculated as follows

ρAB (∆t) = trC (ρ∆t )

∆t X
= trC (ρ0 − i [H, ρ0 ] + ∆t LX ρ0 )
~ X=A,B,C
∆t
= ρAB − i [HA EC1 + HB EC2 , ρAB ]
~
+∆t(LA + LB )ρAB , (5.7)

where we have used ρ0 = ρAB ⊗ ρC , and EC1 = tr(HC1 ρC ) and EC2 = tr(HC2 ρC )
are the initial mean energies.3 We have also utilised the cylic property of trace such
that trC (QkC ρC Qk† 1 k† k
C − 2 {QC QC , ρC }) = 0.
3
We will assume that these constants are dimensionless and the energy unit is transferred
entirely to the terms HA and HB , respectively.

69
 One can see from Eq. (5.7) that the state of AB at time ∆t is simply as a
result of effective local Hamiltonians HA EC1 + HB EC2 and local interactions with
environments. For any entanglement monotone, i.e., non-increasing under local
operations and classical communication (LOCC), we have EA:B (∆t) ≤ EA:B (0) and
therefore ĖA:B (0) ≤ 0.

For example, if we consider unitary dynamics and the initial pure state |00i
of AB, it requires C to be in a decoupled form, i.e., the whole system assumes
|00i h00| ⊗ ρC . From Eq. (5.7) of Theorem 5.4.1, it is easy to see that EA:B (∆t) =
EA:B (0), i.e., ĖA:B (0) = 0. This is due to the fact that unitary operations resulting
from local Hamiltonians leave the entanglement invariant. This implies that, in
order to have positive (or negative) initial entanglement rate, there has to be initial
correlation between the mediator C and the principal objects A and B.
Note also that if C is uncorrelated with AB at all times, i.e., IAB:C (t) = 0, one
can apply the method in Theorem 5.4.1 successively and conclude that entanglement
distribution is impossible as EA:B (t) ≤ EA:B (0). This has been proven similarly in
Chapter 3. This way, entanglement gain between the principal objects can be used
as a witness of correlation with the mediator C, i.e., IAB:C > 0 at some time during
the evolution.
Theorem 5.4.1 is not enough to conclude the impossibility of beating the direct
bound as it does not explain what is in play during the dynamics. It might happen
that although the initial rate of production is non-positive, the rate during the dy-
namics is high such that after some finite time T < Γdi the maximum entanglement
is reached. We will show that this scheme is impossible by utilising the following
ultimate bound.

Theorem 5.4.2. Consider any coherent dynamics of three objects A, B, and C with
the total Hamiltonian given by H. This may include the direct interaction scenario
between A and B as well as the indirect interaction where C is the mediator. Starting
with initial density matrix of the form ρ(0) = ρABC , where the reduced state ρAB is
separable, we show that the time it takes to maximally entangle A and B is lower
bounded as  
1
T ≥ arccos √ , (5.8)
d
where T = Ωt and min{hHi, ∆H} = ~Ω.

Proof. First let us note that maximum entanglement between A and B implies
that the state of AB is pure, and that the state of C is decoupled, i.e., ρ(T ) =

70
|Ψmax i hΨmax | ⊗ ρ0C . Taking the fidelity of the initial and final states gives

F(ρ(0), ρ(T )) = F(ρABC , |Ψmax i hΨmax | ⊗ ρ0C )

≤ F(ρAB , |Ψmax i hΨmax |)
p
= hΨmax |ρAB |Ψmax i
sX
≤ max pj |haj bj |Ψmax i|2
pj ,|aj bj i
j

≤ max |haj bj |Ψmax i|

|aj bj i
1
= √ , (5.9)
d

where we have used the non-decreasing property of fidelity under trace-preserving

completely positive maps (in our case, tracing-out object C) in the first line [92],
P
expressed the separable state as ρAB = j pj |aj bj i haj bj |, and used convexity in
the last inequality.
Finally, by utilising the QSL of Eq. (5.2) and noting that min{hHi, ∆H} = ~Ω,
one obtains  
1
T ≥ arccos (F(ρ(0), ρ(T ))) ≥ arccos √ = Γdi . (5.10)
d

Let us now discuss a special case of Theorem 5.4.2 above. In particular, consider
the initially decoupled mediator, i.e., ρ(0) = ρAB ⊗ ρC . Consequently, we have
q 
√ √
F(ρ(0), ρ(T )) = tr ρAB ρC |Ψmax i hΨmax | ρ0C ρAB ρC
q 
√ √ √ 0 √
= tr ρAB |Ψmax i hΨmax | ρAB ρC ρC ρC
q  q 
√ √ √ 0 √
= tr ρAB |Ψmax i hΨmax | ρAB tr ρC ρC ρC

= F(ρAB , |Ψmax i hΨmax |) F(ρC , ρ0C ). (5.11)

Note that as the fidelity 0 ≤ F(ρC , ρ0C ) ≤ 1, we recover the second line of Eq. (5.9).
Now let us discuss the consequence of the initial state ρ(0) = ρAB ⊗ ρC . From
Eq. (5.11), we notice that if ρAB is strictly mixed, then for a unitary dynamics
(purity-preserving), ρ0C will be more mixed than ρC , i.e., ρC 6= ρ0C . This means
F(ρC , ρ0C ) < 1, making the corresponding first inequality in (5.9) strict, and resulting
in a strict time bound T > Γdi . Therefore, in order to saturate the direct bound,
the state ρAB has to be pure and F(ρC , ρ0C ) = 1. A trivial dynamics (via direct
interactions in a tripartite setting) is given by the example presented in Section 5.3
with the addition of a decoupled mediator ρC and a local Hamiltonian HC .

71
5.4.1 Saturating the limit
Now we present examples of indirect interactions that saturate the direct time
bound. As initial states of the form ρAB ⊗ ρC cannot be used for maximum entan-
glement speed, one would normally think of utilising entanglement in the partition
AB : C. In particular, consider the following initial state and Hamiltonian:

1
|ψ(0)i = √ (|000i + |111i),
2
~Ω z
H = √ (σA ⊗ HC1 + σBz ⊗ HC2 ), (5.12)
2 2

where HC1 = −(1 + σCx + σCy + σCz ) and HC2 = 1 − σCx − σCy + σCz . One finds that this
example has min{hM i, ∆M } = 1 and therefore the bound is also given by the Bures
angle Θ(ρAB (0), ρAB (T )), where now ρAB (T ) = trC (ρABC (T )). Furthermore, the
resulting dynamics is the same as that in Fig. 5.1. From this example we note that
the initial state has some purely quantum properties: entanglement NAB:C (0) = 0.5
and mutual information IAB:C (0) = 2.
However, is it necessary to have entanglement with the mediator C to saturate
the entangling speed of direct interactions? We now show that this is not the case.
In particular, it is enough to have classical correlation with C initially. For this
purpose, we take the initial state and Hamiltonian

1 1
ρ(0) = |ψ+ i hψ+ | ⊗ |+i h+| + |φ+ i hφ+ | ⊗ |−i h−| ,
2 2
~Ω x
H = (σ ⊗ σCx + σBx ⊗ σCx ), (5.13)
2 A
√ √
where σ x |±i = ± |±i, and |ψ+ i = (1/ 2)(|01i+|10i) and |φ+ i = (1/ 2)(|00i+|11i)
are two Bell states of party AB.4 One can see that the initial state in Eq. (5.13)
is disentangled in the partition AB : C and is correlated with mutual information
IAB:C (0) = 1. This example also has the resource equality satisfied and the dynamics
of that in Fig. 5.1. Furthermore, the dynamics from Eq. (5.13) has been shown to
have zero quantum discord DAB|C at all times (see Chapter 3), showing that the
correlation between C and AB is purely classical. One can extend further and see
that initial states of the form p |ψ+ i hψ+ | ⊗ |+i h+| + (1 − p) |φ+ i hφ+ | ⊗ |−i h−|,
where p stands for probability, can be made maximally entangled by Hamiltonian
in Eq. (5.13). This maximum entanglement is achieved in T = π/4 for 0 < p < 1,
see Fig. 5.2. Independent of p, the Bures angle dynamics is given by the dashed line
in Fig. 5.1, which indicates saturation.
Let us compare the three examples shown above. First of all, it is clear that
4
Note that this example is similar to that already presented in Chapter 3 to demonstrate
entanglement localisation.

72
Figure 5.2: Maximum entangling speed for various initial states. The Hamiltonian
is taken to be H = ~Ω2
(σAx ⊗ σCx + σBx ⊗ σCx ) with initial state p |ψ+ i hψ+ | ⊗ |+i h+| +
(1−p) |φ+ i hφ+ |⊗|−i h−|. Entanglement is plotted for p = 0.1 (red curve), 0.3 (blue
curve), and 0.5 (black curve).

starting with pure states, e.g., |00i, it is possible to get maximum entanglement via
direct interactions. Also note that during the evolution, the purity of the state is
preserved. This is because unitary operations do not change purity. Therefore, if
one were to start with a mixed initial state ρAB , the maximum entanglement can
never be achieved. This is true as maximum entanglement is given by pure states (as
measured by any entanglement monotone) [62]. In the indirect interaction scenario,
the evolution of the state of AB is not unitary. Therefore, the dynamics can change
the purity of the state ρAB . For example, in the last two examples, ρAB changes from
having zero entanglement with purity 0.5 to maximum entanglment with purity 1.
Hence, for mixed initial states, one would want to utilise mediators to distribute
maximum entanglement between the principal objects A and B.
Finally, one might wonder, for an initial mixed state, is it possible to achieve
maximum entanglement with direct interaction setup, this time allowing the prin-
cipal objects to be open to their local environments? This question is sensible as
interactions with environments can change the purity of the principal objects, and
therefore allowing them to reach maximum entanglement. To illustrate this, let us
assume a scenario where one of the principal objects, say A, is interacting with a
single-qubit environment C. Furthermore, take the initial state of Eq. (5.13) and
Hamiltonian of the form H = ~Ω σAx ⊗ σCx . One can confirm that the resulting A : B
entanglement dynamics is identical as that in Fig. 5.1.

73
5.5 Initial correlation with mediators bounds lo-
calised entanglement
Note that the example in (5.13), the entanglement localisation, realises maximum
entanglement with dynamics saturating the direct bound by having initial state with
only classical correlation in the partition AB : C. It is then natural to ask what
bounds the entanglement gain. In particular, whether the gain is related to the
initial mutual information IAB:C (0). In order to answer this question, let us begin
by presenting the following Lemma.

Lemma 5.5.1. For a three-party system ABC, where the subsystem C is only clas-
P
sically correlated, i.e., ρ = c pc ρAB|c ⊗ |ci hc| with {|ci} forming an orthonormal
basis, the relative entropy of entanglement follows the bound

EA:BC − EA:B ≤ IAB:C . (5.14)

Proof. By definition of the REE we have EA:BC (ρ) = −tr(ρ log2 σ) − S(ρ), where σ
is the closest separable state to ρ [91]. According to the flags condition EA:BC =
P
c pc EA:B (ρAB|c ), where EA:B (ρAB|c ) = −tr(ρAB|c log2 σAB|c ) − S(ρAB|c ) [40]. Also,
P
it has been shown that σ = c pc σAB|c ⊗ |ci hc|, where σAB|c is a separable state
closest to ρAB|c [40]. Next, we note for the first term in the definition of EA:BC :
!
X X
−tr pc ρAB|c ⊗ |ci hc| log2 ( pj σAB|j ⊗ |ji hj|)
c j
!
X X
= −tr pc ρAB|c ⊗ |ci hc| log2 (pj σAB|j ) ⊗ |ji hj|
c j
!
X
= −tr pc ρAB|c log2 (pc σAB|c )
c
X
= − pc (log2 (pc ) + tr( ρAB|c log2 (σAB|c )))
c
X
= SC + pc (−tr( ρAB|c log2 (σAB|c ))), (5.15)
c

where each term in the sum is minimised as σAB|c is closest to ρAB|c . On the other
P
hand, we have EA:B = −tr( c pc ρAB|c log2 σAB ) − SAB , where σAB is the closest
P
separable state to c pc ρAB|c . Putting things together we have EA:BC − EA:B =

74
IAB:C + η, where IAB:C = SAB + SC − SABC and
X
η = pc [−tr( ρAB|c log2 (σAB|c ))]
c
X
− pc [−tr( ρAB|c log2 (σAB ))]. (5.16)
c

One can see that η is negative, since −tr( ρAB|c log2 (σAB|c )) is minimised, confirming
the Lemma.

Now, for the dynamics, let us consider the indirect interaction scenario where all
our objects are allowed to be open to their own local environment. The entanglement
bound is proven in the following Theorem.

Theorem 5.5.2. Consider the indirect interaction scenario where all the objects A,
B, and the mediator C are open to their local environment. If C is classical at all
times, i.e., DAB|C (t) = 0 for t ∈ [0, τ ], we have

EA:B (τ ) − EA:B (0) ≤ IAB:C (0). (5.17)

Proof. The proof is done with the following chain of equations:

EA:B (τ ) − EA:B (0) ≤ EA:BC (τ ) − EA:B (0) (5.18)

≤ EA:BC (0) − EA:B (0) (5.19)
≤ IAB:C (0), (5.20)

where the steps are justified as follows. The inequality (5.18) is due to the mono-
tonicity of entanglement under local operations (tracing out the mediator C), i.e.,
EA:B (τ ) ≤ EA:BC (τ ). Line (5.19) follows from Theorem 3.3.1, stating that entan-
glement in the partition A : BC cannot grow via classical C, that is EA:BC (τ ) ≤
EA:BC (0). Finally, Lemma 5.5.1 confirms line (5.20).

The bound in Eq. (5.17) can be made simpler at the expense of tightness. Let us
note that for states separable in the partition AB : C (including states where C is
P
classical, i.e., ρ = c pc ρAB|c ⊗ |ci hc|), the conditional entropy {SAB|C , SC|AB } ≥ 0.
This means that the mutual information can be bounded as IAB:C = SAB − SAB|C ≤
SAB or IAB:C = SC − SC|AB ≤ SC . Therefore, entanglement gain via classical
mediators cannot be larger than log2 (dC ) as it is the maximum entropy for system
C. We would also like to mention that if one observes EA:B (t) − EA:B (0) > SAB (0)
then there had to be non-zero quantum discord DAB|C during the evolution. This
is a witness of non-classicality of C by observing only the objects A and B, similar
to the detection of quantum discord in Chapter 3.

75
 As mentioned in Preliminaries, the evolution of a quantum state can be sped up
simply by providing more energy to the system. From the results presented in this
chapter, one can also speed up the creation of entanglement between two objects
via a mediating system that is initially correlated with them. Indeed, for initially
uncorrelated systems, it would take longer time to reach maximum entanglement.
In a way, this puts the two quantities, energy and correlations, on the same footing.
Both can be seen as a resource for fast distribution of quantum entanglement.

5.6 Charging power

In this section we study the quickest way to flip a quantum bit, i.e., to evolve
it from the ground state |0i to the excited state |1i. This is known as charging.
Also for this task, we normalise the resource by having min{hM i, ∆M } = 1. For
simplicity, let us first consider a system of two qubits. One would say that even free
Hamiltonian of the form H = ~Ω(σAx + σBx ) will suffice. While this will charge the
system, it does not provide the quickest process as we will show below. Starting
with the state |00i, the resource equality requirement sets the free Hamiltonian to
√
be Hfree = ~Ω(σAx + σBx )/ 2. As can be seen from the blue dashed line in Fig. 5.3,
this dynamics does not saturate the time bound.
We consider another Hamiltonian, Hint = ~Ω(σAx ⊗ σBx ), which involves interac-
tions between A and B. Indeed, with this Hamiltonian, charging is done fastest as
illustrated by the black dashed line in Fig. 5.3. The generalisation of the fastest
charging to N qubits simply reads

|ψ(0)i = |0i⊗N ,
H = ~Ω (σ1x ⊗ σ2x ⊗ · · · ⊗ σN
x
). (5.21)

Independent of N , fully charged state |1i⊗N is achieved in T = π/2.

5.6.1 Charge quantifier

Let us consider two quantum objects A and B. The goal is to evolve the state from
|00i to the fully charged state |11i. As a way of measuring how close the state
ρAB (t) is to the target state, one may define the charge using the root fidelity as
ΞAB (t) = F(|11i h11| , ρAB (t)). The value would go from 0 to 1 (fully charged).
For direct interactions, where the state is pure, the charge simply reduces to
ΞAB (t) = |h11|ψAB (t)i|. This would mean that for immediate or fast charging,
the state |ψAB (∆t)i has to have a component |11i in the time of order ∆t such that
ΞAB (∆t) > 0, i.e., Ξ̇AB (0) > 0. This scenario is nicely illustrated in Fig. 5.3 by using

76
the Hint . On the other hand, free Hamiltonians of the form HA + HB would evolve
the subsystems separately, such that the state at time ∆t reads |ψAB (∆t)i = (1 −
i∆tHA /~) |0i⊗(1−i∆tHB /~) |0i. One can easily see that, in this case, ΞAB (∆t) = 0.
From these observations, we note that positive charging rate is linked with the
presence of entanglement between objects A and B. This is because entanglement
is nonzero for the state |ψAB (∆t)i with |11i component.

Figure 5.3: Dynamics of charge Ξ (solid curves) and Bures angle (dashed lines) for
interacting (black) and non-interacting (blue) qubits. In both cases, the resource
has been fixed to the same amount, i.e., min{hM i, ∆M } = 1.

For indirect interactions, with system C as the mediator, the initial state is of the
form |00i h00| ⊗ ρC . We show with the following Lemma that an initially correlated
mediator is also a factor for having positive charging rate. This is similar to the
entanglement distribution task.
Lemma 5.6.1. For indirect interactions, where all objects can be open to their local
environment, the charging rate follows Ξ̇AB (0) = 0.
Proof. First, let us express the charge as
p
ΞAB (t) = F(|11i h11| , ρAB (t)) = h11|ρAB (t)|11i. (5.22)

Next, from Eq. (5.7), the state of AB at time ∆t is given by

∆t
ρAB (∆t) = |00i h00| − i [HA EC + HB Eγ , |00i h00|]
~
+∆t(LA + LB ) |00i h00| , (5.23)

where we have used ρ0 = |00i h00| ⊗ ρC . One can see that ΞAB (∆t) = 0 from the
observation that h11|ρAB (∆t)|11i = 0, which proves the Lemma.

77
5.7 Summary
Quantum speed limit sets the minimum time required for evolving a quantum state
and therefore also its properties. In this chapter we have presented the time bound
for the aim of distributing quantum entanglement between two objects both via
direct as well as indirect interactions. We have shown that the indirect interac-
tion setting cannot beat the direct interaction setting in terms of entangling speed.
Furthermore, the correlation between the mediator and the principal objects is re-
quired for optimal distribution. From the perspective of speeding up the creation
of correlations, we discussed that both energy and correlated mediators play similar
roles. We also presented briefly a simple application of the quantum speed limit
to charging of quantum batteries and compared the direct and indirect interaction
settings.

78
Chapter 6

Observable quantum entanglement

due to gravity

This chapter presents experimental setups where quantum entanglement is gener-

ated through gravitational coupling.1 Under certain natural assumptions, this in
turn shows the quantum nature of gravity, which is one of the most important quests
in modern physics. Firstly, a motivation is presented with a focus on the need for
experimental evidence of quantum gravity. Some of past experiments where gravity
affected quantum matter are briefly reviewed. I will then proceed to introduce our
proposal that involves two masses coupled gravitationally. Two cases will be con-
sidered in which the masses are either trapped in harmonic potentials or released
from the traps. I will present details of the dynamics including the calculations of
analytic figures of merit that characterise the amount of generated entanglement and
the accumulation time. An analysis of other effects including environmental noises,
decoherence mechanisms, and Casimir interactions will also be presented. Finally,
I discuss the conclusion one can draw from our proposal.

1
Parts of this chapter are reproduced from our published article of Ref. [93],
which is licensed under the Creative Commons Attribution 4.0 International License
(http://creativecommons.org/licenses/by/4.0/). Where applicable, changes made will be indi-
cated.

79
6.1 Motivation and objectives
The successful unification of electromagnetic, weak, and strong interactions within
the quantum framework strongly suggests that gravity should also be quantised.
Up to date, however, there is no experimental evidence of quantum features of grav-
ity. In numerous experiments gravity is key to the interpretation of the observed
data, but it is often sufficient to use Newtonian theory (quantum particle moving
in a background classical field) or general relativity (quantum particle moving in
a fixed spacetime) to gather a meaningful understanding of such data. Milestone
experiments described within Newtonian framework include gravity-induced quan-
tum phase shift in a vertical neutron interferometer [94], precise measurement of
gravitational acceleration by dropping atoms [95], or quantum bound states of neu-
trons in a confining potential created by the gravitational field and a horizontal
mirror [96]. Quantum experiments that require general relativity include gravita-
tional redshift of electromagnetic radiation [97] or time dilation of atomic clocks at
different heights [98].
A number of theoretical proposals discussed scenarios capable of revealing quan-
tumness of gravity. For example Refs. [99–107] proposed the observation of a probe
mass interacting with the gravitational field generated by another mass. More recent
proposals put gravity in a role of mediator of quantum correlations and are based
on the fact that quantum entanglement between otherwise non-interacting objects
can only increase via a quantum mediator [45, 108, 109], see Chapter 3. Motivated
by these proposals and by advances in optomechanics [110], in particular the cooling
of massive mechanical (macroscopic) oscillators close to their quantum ground state
[111–113] and the measurement of quantum entanglement of a two-mode system
[114–116], we study two nearby cooled masses interacting gravitationally.
We propose two scenarios capable of increasing gravitational entanglement be-
tween the masses. In the first scenario, we consider the masses trapped at all times
in 1D harmonic potentials (optomechanics). In the second one, the masses are re-
leased from the optical traps. For both settings, we derive an analytic figure of merit
characterising the amount of gravitationally induced entanglement and the time it
takes to observe it. The derivation includes various initial states and shows that the
objects have to be cooled down very close to their ground states and that squeezing
of their initial state significantly enhances the amount of generated entanglement.
We then formulate a numerical approach, which accounts for all the relevant sources
of noise affecting the settings that we propose, to identify a set of parameters re-
quired for the observation of such entanglement. Finally, we discuss the conclusions
that can be drawn from this experiment with emphasis on the need for indepen-
dent verification that the gravitational interaction between nearby objects is indeed

80
mediated.

6.2 Proposed experimental setups

Consider two particles, separated by a distance L, as depicted in Fig. 6.1. In what
follows, we study the setting where the massive particles are either held or released
from unidimensional harmonic traps. In the former case one can treat the particles
as identical harmonic oscillators, with the same shape, mass m, and vibrational
frequency ω. The two oscillators and the gravitational interaction between them
give rise to the total Hamiltonian H = H0 + Hg , where

p2A 1 2 2 p2B 1
H0 = + mω xA + + mω 2 x2B (6.1)
2m 2 2m 2

and Hg describes the gravitational term. If the harmonic traps are removed the
corresponding Hamiltonian simplifies to H0 = (p2A + p2B )/2m. Before we proceed
with detailed calculations, we shall discuss generic features of the gravitational term
and the conditions required for the creation of entanglement.

Figure 6.1: Thought experiment for the generation of entanglement via gravitational
interactions. Two particles are a distance L apart and trapped with unidimensional
harmonic potentials. It is assumed that both particles are cooled down near their
ground state. We consider two scenarios in which the particles are either trapped
at all times or released from the traps. In the text, we investigate entanglement
dynamics for both scenarios. Note that the generated entanglement can be probed
with weak lights. Apart from the dominant gravitational coupling, our analysis
takes noise, damping, decoherence, and Casimir forces into account.

In general, the gravitational term Hg depends on the geometry of the objects.

Various configurations have been analysed (will be presented later in Section 6.5).
The results of such analysis suggest that spherical masses give rise to the high-
est amount of generated entanglement. The Newtonian gravitational energy of
this setting is the same as if the two objects were point-like masses, that is Hg =
−Gm2 /(L + xB − xA ), where L is the distance between the objects at equilibrium
and xA (xB ) is the displacement of mass A (B) from equilibrium. By expanding
the energy in the limit xA − xB  L, which is well justified for oscillators that are

81
cooled down close to their ground state, one gets

Gm2 (xA − xB ) (xA − xB )2

 
Hg = − 1+ + + ··· . (6.2)
L L L2

The first term is a rigid energy offset, while the second is a bi-local term and cannot
thus give rise to quantum entanglement. The third term, which is proportional to
(xA − xB )2 , is the first that couples the masses. When written in second quan-
tisation, it becomes apparent that this term includes contributions responsible for
the correlated creation of excitations in both oscillators. In the quantum optics
language, this is commonly referred to as a “two-mode squeezing” operation, which
can in principle entangle the masses provided a sufficient strength of their mutual
coupling.

6.3 Dynamics of gravitationally induced entan-

glement: Oscillators
We first provide an intuitive argument setting the scales of experimentally relevant
parameters, which will then be proven rigorously. In order to achieve considerable
entanglement, we should ensure that the coupling (third term) in Eq. (6.2) is com-
parable to the energy ~ω of each oscillator, that is Gm2 (xA − xB )2 /L3 ∼ ~ω. As we
assume that the oscillators are near their ground state, we estimate their displace-
ments by the ground state extension, (xA − xB )2 ∼ 2~/mω. We thus introduce the
(dimensionless) figure of merit
2Gm
ηg ≡ 2 3 . (6.3)
ω L
We should have ηg ∼ 1 in order for the oscillators to be significantly entangled. This
sets the requested values of the experimentally relevant parameters m, ω, and L.
Below we will demonstrate the following results, which embody the key find-
ings of our investigation: (i) Starting from the ground state of each oscillator and
assuming (for the sake of argument) only negligible environmental noise, the maxi-
mum entanglement (as quantified by the logarithmic negativity [34, 35]) generated
max
during the dynamics is given by Eth ≈ ηg / ln 2.2 Moreover, the time taken for
entanglement to reach such maximum value is tmax th = π/2(1 − ηg )ω; (ii) Single-
mode squeezing of the initial ground state of each oscillator substantially enhances
the gravity-induced entanglement. The corresponding maximum entanglement be-
max
comes Esq ≈ |sA + sB |/ ln 2, where sj (j = A, B) is the degree of squeezing of the
jth oscillator, and we assume ηg  sA , sB . In this case, the maximum entanglement
2
Note that in this chapter, we shall use E to denote the logarithmic negativity for simplicity.

82
is reached in a time tmax
sq = π/2ηg ω; (iii) Weaker entanglement is generated with
increasing temperature of the masses or coupling to the environment.
We will begin by constructing Langevin equations of the masses in Heisenberg
picture. As we will be dealing with Gaussian states, we provide the dynamics of the
covariance matrix, which completely describes the state of the whole system. From
this dynamics, we will calculate important quantities such as quantum entanglement
between the masses, both analytically for special cases and numerically for general
cases.

6.3.1 Langevin equations

As the third term in Eq. (6.2) is already very small under usual experimental con-
ditions,3 we neglect all terms of order higher than the second in the displacement
from equilibrium. We note Ref. [117] for similar treatment of linearised central-
potential interactions. By taking the total Hamiltonian with a suitably truncated
gravitational term Hg , one gets a set of Langevin equations in Heisenberg picture

Ẋj = ω Pj (j = A, B),
ṖA = −ω (1 − ηg ) XA − ωηg XB − γ PA + ξA + ν, (6.4)
ṖB = −ω (1 − ηg ) XB − ωηg XA − γ PB + ξB − ν,
√
where we have introduced the constant frequency ν = Gm2 / ~mωL4 and the di-
p √
mensionless quadratures Xj = mω/~ xj and Pj = pj / ~mω. These equations
incorporate Brownian-like noise – described by the noise operators ξj – and damp-
ing (at rate γ) affecting the dynamics of the mechanical oscillators, due to their
interactions with their respective environment. We assume the (high mechanical
quality) conditions Q = ω/γ  1, as it is the case experimentally, so that the
Brownian noise operators can de facto be treated as uncolored noise and we can
write hξj (t)ξj (t0 ) + ξj (t0 )ξj (t)i/2 ' γ(2n̄ + 1)δ(t − t0 ) for j = A, B [118, 119]. Here,
n̄ = (eβ − 1)−1 is the thermal phonon number with β = ~ω/kB T and T the temper-
ature of the environment with which the oscillators are in contact.
The linearity of Eqs. (6.4) and the Gaussian nature of the noise make the theory
of continuous variable Gaussian systems very well suited to the description of the
dynamics and properties of the oscillators under scrutiny. In this respect, the key
tool to use is embodied by the covariance matrix V (t) associated with the state
of the system, whose elements Vij (t) = hui (t)uj (t) + uj (t)ui (t)i/2 − hui (t)ihuj (t)i
encompass the variances and correlations of the elements of the quadrature vector
3
p Note that the ratio between any two consecutive terms in Eq. (6.2) is given by (xA − xB )/L ∼
~/mωL2 . For instance, taking m = 100 µg, ω = 100 kHz, and L = 0.1 mm gives this ratio
∼ 10−12 , and for macroscopic values m = 1 kg, ω = 0.1 Hz, and L = 1 cm the ratio is ∼ 10−15 .

83
u(t) = (XA (t), PA (t), XB (t), PB (t))T . The temporal behaviour of physically rele-
vant quantities for our system of mechanical oscillators can be drawn from V (t)
by making use of the approach for the solution of the dynamics that is illustrated
below.

6.3.2 Dynamics of covariance matrix

In this section we provide the solution of the Langevin equations, and consequently
the covariance matrix. One can rewrite the equations in (6.4) as a single matrix
equation u̇(t) = Ku(t) + l(t), with the vector u(t) = (XA (t), PA (t), XB (t), PB (t))T
and a drift matrix
 
0 ω 0 0
 
 ηg − ω(1 − ηg ) −γ −ωηg 0 
K=  . (6.5)
 0 0 0 ω 

−ωηg 0 −ω(1 − ηg ) −γ

We split the last term in the matrix equation into two parts, representing the noise
and constant term respectively, i.e., l(t) = υ(t)+κν , where υ(t) = (0, ξA (t), 0, ξB (t))T
√
and the constant κν = ν(0, 1, 0, −1)T with ν = Gm2 / ~mωL4 .
The solution to the Langevin equations is given by
Z t
u(t) = W+ (t)u(0) + W+ (t) dt0 W− (t0 )l(t0 ), (6.6)
0

where W± (t) = exp (±Kt). This allows one to calculate the expectation value of
the ith quadrature hui (t)i numerically, which is given by the ith element of
Z t
W+ (t)hu(0)i + W+ (t) dt0 W− (t0 )κν , (6.7)
0

where we have used the fact that the noises have zero mean, i.e., hυi (t)i = 0 and
that hκν i = tr(κν ρ) = κν . From Eq. (6.6), one can also calculate other important
quantities via the covariance matrix as shown below.
Covariance matrix of our system is defined as Vij (t) ≡ h{∆ui (t), ∆uj (t)}i/2 =
hui (t)uj (t)+uj (t)ui (t)i/2−hui (t)ihuj (t)i where we have used ∆ui (t) = ui (t)−hui (t)i.
This means that κν does not contribute to ∆ui (t) (and hence the covariance matrix)
since hκν i = κν . We can then construct the covariance matrix at time t from Eq.

84
(6.6) without considering κν as follows

Vij (t) = hui (t)uj (t) + uj (t)ui (t)i/2 − hui (t)ihuj (t)i
V (t) = W+ (t)V (0)W+T (t)
Z t
+W+ (t) dt0 W− (t0 )DW−T (t0 ) W+T (t), (6.8)
0

where D = Diag[0, γ(2n̄ + 1), 0, γ(2n̄ + 1)] and we have assumed that the initial
quadratures are not correlated with the noise quadratures such that the mean values
of the cross terms are zero. A more explicit solution of the covariance matrix, after
integration in Eq. (6.8), is given by

KV (t) + V (t)K T = −D + KW+ (t)V (0)W+T (t)

+W+ (t)V (0)W+T (t)K T
+W+ (t)DW+T (t), (6.9)

which is linear and can be solved numerically.

Consider a special case, in which the damping term γ is negligible, giving D = 0.
In this case, Eq. (6.9) simplifies to

V (t) = W+ (t)V (0)W+T (t). (6.10)

In this regime we will obtain analytical results of Section 6.3.3.

Entanglement from covariance matrix

We use logarithmic negativity to quantify the amount of entanglement between the

coupled masses. One can calculate this from the covariance matrix by using the
method in Chapter 2, which will be briefly repeated here for convenience. The
covariance matrix V (t) describing our two-mode system can be written in a block
form !
IA L
V (t) = , (6.11)
LT IB
where the component IA (IB ) is a 2 × 2 matrix describing local mode correlation
for A (B) while L is a 2 × 2 matrix characterising the intermodal correlation. A
two-mode covariance matrix has two symplectic eigenvalues {ν1 , ν2 }. A physical
system has ν1 , ν2 ≥ 1/2 [25].
For entangled modes, the covariance matrix will not be physical after partial
transposition with respect to mode B (this is equivalent to flipping the sign of the
oscillator’s momentum operator PB in V (t)). This unphysical V (t)TB is shown by

85
the minimum symplectic eigenvalue ν̃min < 1/2. The explicit expression is given by
√ √
ν̃min = (Σ − Σ2 − 4 detV )1/2 / 2, where Σ = detIA + detIB − 2 detL. Entangle-
ment between mode A and mode B is then quantified by logarithmic negativity as

E = max 0, − log2 (2ν̃min ) [34, 35]. Note that the separability condition, when
V (t)TB has ν̃min ≥ 1/2, is sufficient and necessary for two-mode systems [120].

6.3.3 Noiseless dynamics: Analytical solution

Thermal initial state

Due to weakness of the gravitational coupling, we have ηg  1 in practically any

realistic experimental situation, and we thus assume such condition here.4 In the
case of no damping (i.e., γ = 0) and assuming an initial (uncorrelated) thermal
state of the oscillators, a tedious but otherwise straightforward analytical derivation
shows that the entanglement between the mechanical systems, as quantified by the
logarithmic negativity, oscillates in time with an amplitude of ηg / ln 2 − log2 (2n̄ + 1).
At low operating temperature, a condition achieved through a combination of passive
and radiation-pressure cooling [110], n̄ ≈ 0 and the maximum entanglement between
max
the oscillators is Eth ≈ ηg / ln 2, a value reached at a time tmax
th such that ωtmax
th =
π/2(1 − ηg ). We present the dynamics of entanglement for varying values of ηg in
Fig. 6.2.

Figure 6.2: Entanglement dynamics with initial ground state for each mass. The
figure of merit is varied as ηg = 0.01 (red curve), 0.05 (blue curve), and 0.1 (black
curve). The maximum entanglement approximately follows ηg / ln 2.

4
We use ηg  1 for all analytical derivations in this chapter.

86
Squeezed thermal initial state

An analytic solution is also possible for the case of mechanical systems initially pre-
pared in squeezed thermal states, a situation that can be arranged by suitable optical
driving [121, 122]. Each mass is prepared in a state Sρth S † , where ρth is a thermal
state and S = exp (−i s(X 2 − P 2 )/2) is the squeezing operator with strength s
(assumed to be real). This operator corresponds to anti-squeezing (squeezing) the
position quadrature for s > 0 (s < 0). By writing individual-oscillator squeezing as
sj and assuming sj  ηg , the entanglement is again observed to oscillate, but with
amplitude |sA + sB |/ ln 2 − log2 (2n̄ + 1). We present the entanglement dynamics
in Fig. 6.3 for varying values of the squeezing strength. Note that it is irrelevant
whether quadratures of both masses are squeezed or anti-squeezed. We provide ex-
planation in the Details of entanglement dynamics (Section 6.3.5). Therefore, only
the degree of pre-available single-oscillator squeezing and the environmental tem-
perature set a limit to the amount of entanglement that can be generated between
the mechanical systems through the gravitational interaction. In the low temper-
max
ature limit, where Esq ≈ |sA + sB |/ ln 2, which is in principle arbitrarily larger
than the case without squeezing, a time tmaxsq = π/(2ηg ω)  tth
max
would be required
for such entanglement to accumulate. Needless to say, long accumulation times are
far from the possibilities offered by state-of-the-art optomechamical experiments,
which prompts an assessment that includes ab initio the effects of environmental
interactions.

Figure 6.3: Entanglement dynamics with squeezed initial ground state for each
mass. The figure of merit is taken as ηg = 10−4 and the squeezing strength is varied
as sA,B = 0.1 (red curve), 0.5 (blue curve), and 1 (black curve). The maximum
entanglement approximately follows 2sA,B / ln 2.

87
6.3.4 Noisy dynamics: Numerical simulation
In the case of noisy dynamics, however, an analytical solution is no longer available
and we have to resort to a numerical analysis. Let us therefore consider the figure
of merit ηg in order to set the parameters for numerical investigation. We consider
two oscillators of spherical shape with uniform density ρ and radius R, which are
separated by a distance L = 2.1R. This might be a situation matching current
experiments in levitated optomechanics [123, 124], which are rapidly evolving to-
wards the possibility of trapping multiple dielectric nano-spheres in common optical
traps and controlling their relative positions [125]. However, low-frequency oscilla-
tors, which are favourable for the figure of merit and typically associated with large
masses, are unsuited to such platforms and would require a different arrangement,
such as LIGO-like ones [111].
In terms of the density ρ, we have ηg = 8πGρ/3(2.1)3 ω 2 , which does not depend
on the dimensions of the oscillators nor their mass. As the density of materials
currently available for such experiments varies within a range of only two orders of
magnitude, the linear dependence on ρ sets a considerable restriction on the values
that ηg can take. The densest naturally available material is Osmium, which has ρ =
22.59 g/cm3 and, in order to provide an upper bound to the generated entanglement
that would be attainable using other materials, we shall use this density in our
numerical simulations. Accordingly, ηg = 1.36 × 10−6 /ω 2 , where ω is in Hz.

Figure 6.4: Entanglement dynamics between two spherical oscillators. Dynamics

for shorter times (a), showing entanglement resolution, and for longer times (b),
showing high accumulation. The varying parameters are the temperature (mean
phonon number n̄) and the damping of the oscillators (mechanical quality factor
Q). We have fixed both the frequency of the mirrors ω = 0.1 Hz and the amount of
squeezing sA,B = 1.73. Note that entanglement is quantified using the logarithmic
negativity.

Fig. 6.4 shows exemplary entanglement dynamics for different values of the ther-
mal phonon number n̄ and mechanical quality factor Q.5 The frequency has been
5
The oscillations of entanglement for unsqueezed initial state are still present in this dynamics,

88
fixed to ω = 0.1 Hz (cf. Section 6.8). As expected, higher damping (lower Q) results
in the decay of entanglement, and the higher the temperature of the mirror (higher
n̄) the higher the mechanical quality factor needed to maintain entanglement. The
setup allows for high entanglement, even with low coupling strength ηg ∼ 10−4 .
However, this comes at the expense of the time for which the dynamics of the oscil-
lators should be kept coherent. In the same figure one can see the time required for
accumulation of entanglement. It shows that cooling down the masses close to their
ground state, n̄ ≈ 0, is crucial for the reduction of the required coherence time.

6.3.5 Details of entanglement dynamics

In this section we show that entanglement gain is linked to the evolution of the
position variance of each mass. This is intuitive because bigger variance means
stronger gravitational coupling for parts of the wave functions which are closer. In
order to illustrate this, we take, as an example, the oscillators setup with squeezed
initial ground state for each mass.

Figure 6.5: Exemplary dynamics of entanglement and width of two coupled oscil-
lators. Each oscillator is made of Osmium and has a spherical shape with mass
m = 1 kg. The equilibrium distance is L = 2.1R and the frequency of the trapping
potential is ω = 0.1 Hz. (a) Short dynamics within 100 s for squeezing parameters
sA,B = ±1.73. (b) Higher gain of entanglement requires longer time. Note that
the dynamics in (b) is approximately the same for positive and negative squeezing
parameters.

Fig. 6.5a shows the entanglement dynamics with both sA,B = 1.73 and −1.73,
corresponding to initial states of the masses that are both anti-squeezed or squeezed,
respectively, in position quadrature. During the evolution, the width of each mass
oscillates as illustrated in the inset of Fig. 6.5a. It is clear that the oscillation of the
width matches the oscillation of entanglement, for both squeezing parameters. For
longer time, i.e., t ∼ 104 s, the oscillation in width leads to an accumulation of high
entanglement as can be seen in Fig. 6.5b. Note that the oscillation of the position
showing repeating pattern with a period of π/[(1 − ηg )ω] ≈ 31 s.

89
variance is very rapid on this timescale and the envelope of these oscillations is the
same for both positive and negative s, see the inset of Fig. 6.5b. As a result, the
entanglement dynamics is approximately equal for squeezed and anti-squeezed cases.
This confirms our analytical result regarding the maximum entanglement gain being
|sA + sB |/ ln 2.

6.4 Dynamics of gravitationally induced entan-

glement: Released masses
As seen previously, the experimental parameters required for detectable gravitational
entanglement of masses in harmonic traps are demanding. We therefore study one
more feasible system, where the traps are switched off after cooling the masses. All
solutions are analytical under the assumed weak gravitational coupling.

6.4.1 Langevin equations and covariance matrix

Similar to the treatment of two oscillators, one starts with the total Hamiltonian for
free masses and truncated gravitational term, and obtains the following equations
of motion
Ẋj = ω Pj (j = A, B),
ṖA = ωηg XA − ωηg XB + ν, (6.12)
ṖB = ωηg XB − ωηg XA − ν.
Note that ω here just sets the conversion between xj , pj and their dimensionless
parameters Xj , Pj . In what follows, we will consider starting the dynamics with
thermal state for each mass. For example, the ground state is a Gaussian state with
p
width ∆x(0) = ~/2mω. This way, one can think of ω as a parameter characterising
the initial spread of the wave function.
For the solution of the dynamics, one can follow similar treatments as in Sec-
tion 6.3.2, keeping in mind γ = 0 and υ(t) = (0, 0, 0, 0)T such that the quadrature
dynamics and covariance matrix is given by Eq. (6.7) and Eq. (6.10) respectively
with a new drift matrix
 
0 ω 0 0
 
 ωηg 0 −ωηg 0 
K=  0
. (6.13)
 0 0 ω  
−ωηg 0 ωηg 0

90
6.4.2 Noiseless dynamics: Analytical solution
One can obtain the covariance matrix V (t) from Eqs. (6.12) and consequently derive
the entanglement dynamics using the approach described above. After imposing the
√
limits ηg  1 and ηg ωt  1, which apply in typical experimental situations, one
obtains the analytical expression for the entanglement dynamics as follows

Eth (t) = max 0, Egnd (t) − log2 (2n̄ + 1) , (6.14)
q p 
Egnd (t) = − log2 1 + 2σ(t) − 2 σ(t)2 + σ(t) ,

where Egnd (t) is the entanglement with initial ground state for each mass and σ(t) =
4G2 m2 ω 2 t6 /9L6 . Since entanglement is an increasing function of σ(t), the latter is
a figure of merit for entanglement gain relevant in the case of released masses.
We present exemplary entanglement dynamics in Fig. 6.6 for which entanglement
∼ 10−2 is achieved within seconds. The parameters used here are m = 100 µg,
ω = 100 kHz, and L = 3R. We will show later that with these values gravity is the
dominant interaction and coherence times are much longer than 1 s. Note that this
setup does not require any squeezing.

Figure 6.6: Entanglement dynamics between two released spherical particles. Each
particle with mass 100 µg is initially trapped in 1D harmonic potential (ω =
100 kHz) and having phonon number n̄. This corresponds to the initial state of
each mass being Gaussian thermal state. We take the equilibrium distance to be
L = 3R ≈ 0.1 mm, making gravitational coupling more dominant than the Casimir
interactions. We note that detectable entanglement, E ∼ 0.01, is generated in sec-
onds: 0.8, 4.5, and 7.5 s for n̄ = 0, 1, and 5 respectively. It is shown in the text that
this accumulation is within the coherence times of the particles.

These improvements over the scheme with trapped masses are the result of un-
limited expansion of the wave functions. For example, for initial ground state, the
p √
evolution of the width of each sphere closely follows ∆x(t) ≈ ~/2mω 1 + ω 2 t2 ,
which is an exact solution to a free non-interacting mass. The effect of gravity is

91
stronger attraction of parts of the spatial superposition that are closer, and hence
generation of position and momentum correlations, leading to the growth of quan-
tum entanglement, see inset in Fig. 6.6.
In order to understand the effect of squeezing in this setup, let us suppose, for
simplicity, the squeezing strengths sA,B = s. It is as if one initially prepared each
mass in a Gaussian state with a new initial spread ∆x0 (0) = ∆x(0) exp(s). One can
then calculate the entanglement dynamics using Eq. (6.14) with a new frequency
ω 0 = ω exp(−2s). This means that anti-squeezing the initial position quadrature
(s > 0) would decrease entanglement gain, a situation opposite to the oscillators
setup. This is because a Gaussian state with smaller ∆x(0) spreads faster, such that
during the majority of the evolution, the width is larger than that if one started with
larger ∆x(0). In principle, one obtains higher entanglement gain by squeezing the
position quadrature (s < 0). However, this will result in higher final width, making
it more susceptible to decoherence by environmental particles (see calculations in
Section 6.7).
From numerical simulations, one confirms that, within t = [0, 10]s, the displace-
ments of the two masses follow xA − xB  L. Furthermore, the trajectories coin-
cide for both quantum treatment with truncated gravitational energy and classical
treatment with full Hg (see Section 6.4.3). This justifies the approximations used. A
summary of comparison between our two proposed setups can be seen in Table 6.1.

Table 6.1: Summary of parameters for the proposed experimental setups: Oscillators
and released masses. Note that for the dynamics (shaded), we assume initial ground
state for each mass.
Parameters Oscillators Released masses
Mass mA,B 1 kg 100 µg
Radius R 2.2 cm 0.1 mm
Equilibrium distance L 4.6 cm 0.3 mm
Trap frequency ω 0.1 Hz 100 kHz
Squeezing strength sA,B 1.73 0
−17
Average width ∆x 8.3 × 10 m 2.8 × 10−12 m
Entangling time τe 31.7 s 0.8 s

6.4.3 Quantum and classical trajectories

From the quadrature dynamics of Eq. (6.7) above, one can calculate the expecta-
tion value of position for both masses. They are presented in Fig. 6.7, where we
have assumed initial conditions hu(0)i = (0, 0, 0, 0)T . On the other hand, without
truncating the gravitational interaction, one can easily solve the classical dynamics

92
and obtain the following equation for xt :
r
2Gm p L π −1

t = xt (L − 2xt ) + √ − tan (θ(xt )) , (6.15)
L 2 2 2
p
where θ(xt ) = (L − 4xt )/ 8xt (L − 2xt ) and t is the time taken for the left mass
to move a distance xt . The trajectory of the right mass is simply −xt . One can
confirm that the classical trajectories indeed coincide with the quantum ones. This
justifies the truncation of Hg in our calculations. Note also that, within 10 s, the
displacement (xA − xB ) ∼ 10−9 m, which is much smaller than the initial distance
between the masses L ≈ 0.3 mm. This validates the use of the limit xA − xB  L.

Figure 6.7: Expectation value of the displacement of the masses. Two Osmium
spheres, each with mass 100 µg, are separated by L = 3R. The dynamics shown is
independent of the temperature and the frequency of the initial trapping potential
ω.

6.5 Other shapes of masses

First, let us note, from Eqs. (6.4) and (6.12), that the magnitude of the interaction
rate between different modes (couples the momentum of one mass to the position
of the other) in the case of identical spheres is given by r1 = 2Gm/ωL3 . One can
alternatively see this from an element of the drift matrix, i.e., r1 = |K23 |. Assuming,
e.g., the material is Osmium and L = 2.1R, one gets r1 = 1.36 × 10−6 /ω. Below, we
will compare the interaction rate for different shapes of the objects to this reference.
Consider a particle of spherical shape, mass mA , and frequency ωA that inter-
acts with a second sphere with mass mB and frequency ωB , see Fig. 6.8a (the
same treatment applies to released masses with initial trap frequencies ωA and ωB ).
Moreover, we assume both spheres are made of Osmium and take RB = αRA , where
α = [0, ∞). After taking similar steps as those in Section 6.3.1 one obtains a new
intermodal interaction rate r2 = 1.36 × 10−6 f (α)/ωA , where f (α) = α9/4 and we

93
have used L = 2.1RA . We have also assumed spring-like scaling for the frequency,
p
i.e., ωB = ωA mA /mB . This result is intuitive as one expects stronger gravitational
interaction by making the second sphere larger, i.e., larger α.
More intriguing is the setting in Fig. 6.8b, in which a rod with length d, mass
λB d, and radius RB is interacting with a sphere of mass mA and radius RA . For
simplicity, we assume both objects have a single mechanical frequency ωA and ωB
respectively, and that the rod is thin such that its radius is much smaller than L.
The gravitational interaction reads
p !
d/2 + (d/2)2 + (L0 )2
Hg = −2GmA λB ln , (6.16)
L0

where L0 = L − (xA − xB ) with xA (xB ) being the displacement of mass A (B).

Note that this expression is the same as if we had a one-dimensional rod and a point
mass. By expanding Eq. (6.16) in the limit xA − xB  L and keeping only up to
the quadratic term in displacements, one can show that the intermodal interaction
rate is given by r3 = 2.18 × 10−7 f (ς)/ωA , where
√
ς 2 ((ς 2 − 1) 1 + ς 2 − 1)
 
1/4
f (ς) = (ς) 1− √ , (6.17)
(1 + 1 + ς 2 )2 (1 + ς 2 )3/2

and we define ς = 2L/d. We have also taken L = 1.1RA , RB = 0.1RA , and the
same spring-like scaling as in the two-sphere configuration. Although the strength
of the gravitational energy (6.16) increases monotonically with d, it is not the case
for f (ς), and hence r3 , which peaks at ς ≈ 1.14, i.e., d ≈ 1.75L ≈ 1.93RA . The
maximum fmax = 1.07 gives maximum interaction rate 2.33 × 10−7 /ωA . For higher
d the rate r3 decreases, implying that r3 is always weaker than r1 .
We also note that both gravitational field and field gradient are necessary (but
not sufficient) for producing entanglement. This is clear from consideration of yet
another configuration – an infinite plane and a point mass separated by a distance
L. One immediately observes that there is no field gradient here and that the
gravitational energy of this configuration is proportional to L − (xA − xB ), which
does not couple the masses and therefore does not contribute to entanglement.

6.6 Casimir interactions

In light of their proximity, apart from gravitational interaction, the two masses can
also interact via Casimir force. It has been shown that the Casimir energy between
two nearby spheres is given by a fraction of the “proximity force approximation”
Ec = −f0 (π 3 /1440)~cR/(L−2R+xB −xA )2 , with the factor 0 ≤ f0 ≤ 1 [126, 127]. As

94
Figure 6.8: Other configurations and notation for the two interacting masses. A
mass with spherical shape is situated next to another spherical mass with a different
radius (a) or a thin rod (b).

typically xA −xB  L−2R, we can expand such expression to find a quadratic term
in xA − xB that can produce entanglement between the masses. The strength of this
term, however, is much weaker than the strength of the corresponding entangling
term of gravitational origin: for Osmium oscillators with mass ∼ 1 kg separated
by L = 2.1R, the ratio between the Casimir and gravitational term is rcg ∼ 10−12 .
Similar calculations made for released masses of the same material with m = 100 µg
and L = 3R, give rcg ∼ 10−2 . It is thus legitimate to ignore Casimir interaction in
both schemes.

6.7 Standard decoherence

Let us also discuss common decoherence mechanisms, i.e., due to interactions with
thermal photons and air molecules [128]. All the situations we consider follow the
limit ∆x  λ, i.e., the “size” of superposition is much smaller than the wavelength
of the particles causing the decoherence. In this regime, the coherence time due to
interactions with thermal photons is given by τph = 1/Λph (∆x)2 with

Λph = 1036 R6 T 9 [1/m2 s], (6.18)

where R is the radius of the sphere and T is the temperature of the environment.
Note that all variables are in SI units.
The decoherence due to interactions with other scattering particles, e.g., air
molecules, gives
8 N√
Λam = 2 2πmair R2 (kB T )3/2 , (6.19)
3~ V
where N/V is the density of air molecules with mass mair . We take mair ≈ 0.5 ×
10−25 kg. For ultrahigh vacuum, pressure ∼ 10−10 Pa, the density is ∼ 1012
particles/m3 . One could also consider performing these experiments in space. In
this case, by taking the pressure ∼ 10−15 Pa, the density can be as low as ∼
107 particles/m3 .
We take the average width of the wave function as an estimate for the super-

95
position that is subjected to decoherence. For oscillators made of Osmium, we use
m ∼ 1 kg and frequency ω ∼ 0.1 Hz. Taking L = 2.1R and starting with ground
state give ∆x ≈ 8 × 10−17 m. From interactions with thermal photons at environ-
mental temperature of 4 K (liquid Helium), the coherence time for the oscillators is
τph ∼ 5s. The coherence time due to collisions with air molecules can be improved
by evacuating the chamber with the oscillators – for about 1012 molecules/m3 (ul-
trahigh vacuum), the coherence time is also τam ∼ 5s. For the space experiment, by
taking the temperature as 2.7 K (cosmic microwave background) and assuming 107
molecules/m3 , we obtain τph ∼ 170s and τam ∼ 106 s.
By making similar calculations for released masses, with parameters considered
in Fig. 6.6, one obtains τph ∼ 105 s and τam ∼ 10−4 s for the experiment on Earth
with liquid Helium temperature and ultrahigh vacuum. For the space experiment
the coherence times improve to τph ∼ 107 s and τam ∼ 41s respectively. We present
in Table 6.2 a summary of coherence times.

Table 6.2: Coherence times for oscillators (m = 1 kg, ω = 0.1 Hz) and released
masses (m = 100 µg, ω = 100 kHz) both for experiments on Earth and in space. We
use the pressure P ∼ 10−10 Pa and the temperature T = 4 K for Earth experiment,
while P ∼ 10−15 Pa and T = 2.7 K for space experiment.
Coherence Oscillators Released masses
time Earth Space Earth Space
τph 5s 170 s 4 × 10 s 2 × 107 s
5

τam 5s 106 s 2 × 10−4 s 41 s

Other schemes have been proposed for gravitationally induced entanglement [108,
109]. They are based on Newtonian interaction between spatially superposed mi-
crospheres with embedded spins. In those proposals, entanglement is reached faster
and the small size of the experiment is the main advantage. However, in order
to separate gravitational and Casimir contributions in that setup, each diamond
sphere with mass m = 10−14 kg has to be superposed across 250 µm. Decoherence
due to scattering of molecules then becomes the main limiting factor. The schemes
we discussed here are complementary in a sense that vibrations of each oscillator
are minute (no macroscopic superposition) but larger mass, 100 µg, is required for
observable entanglement.

96
6.8 Comparison with recent experimental achieve-
ments
We have shown that two nearby masses – both trapped and released – can be-
come entangled via gravitational interaction. Let us now discuss the experimental
conditions required to observe this entanglement in light of recent experimental
achievements.
Logarithmic negativity in the order 10−2 has already been observed experimen-
tally between mechanical motion and microwave cavity field [114]. Extrapolating
the same entanglement resolution to the case of two massive oscillators sets the
required frequency to ω ∼ 10−2 Hz, see Eq. (6.3) and its expression in terms of
ω. Interestingly, kilogram-scale mirrors of similar frequency (ω ∼ 10−1 Hz) were
recently cooled down near their quantum ground state [111]. Furthermore, recent
experiments on squeezed light have reported high squeezing strength [129, 130] (see
also a review in this context [131]), up to 15 dB, which corresponds to s ≈ 1.73.
Advances in the state transfer between light and optomechanical mirrors [110] make
this high squeezing promising also for mechanical systems.
For released masses, the experimental requirements are more relaxed. Their
mass can be considerably smaller while the frequency for initial trapping consider-
ably higher, which is close to common experimental parameters used for optome-
chanical system [110, 112, 113]. Note that higher frequency (lower ∆x(0)) actually
improves entanglement gain, unlike in the oscillators case where small ω is prefer-
able. However, one has to be cautious of decoherence mechanisms as a result of faster
spreading rate of the wave functions. For future experiments, an improvement in
the sensitivity of entanglement detection will also be beneficial.

6.9 Gravity: Direct or mediated interaction?

We conclude with analysis of implications on the nature of the gravitational coupling
that one can draw from such experiments and future research directions. In labo-
ratory, we deal with two nearby masses which are experimentally shown to become
entangled. These setups can be theoretically treated in different ways depending
on the assumptions one makes about gravity. In a “conservative approach” the two
masses are coupled via Newtonian potential. As seen from our calculations and
those in Refs. [108, 109] this indeed leads to gravitationally-induced entanglement.
In this picture gravity is a direct interaction and hence it is difficult to draw conclu-
sions about the form of quantumness of the gravitational field. We note that even
in this conservative approach such an experiment has considerable value as it would
show the necessity of at least the quadratic term in the expansion of the Newtonian

97
potential for generating entanglement.
The objection to the conservative approach is instantaneity of gravitational in-
teraction: Newtonian potential directly couples masses independently of their sepa-
ration. On the other hand, it has been shown that gravitational waves travel with
finite speed [132]. For nearby masses this retardation is hardly measurable and
Newtonian potential is dominant and expected to correctly describe the amount
of generated entanglement. A more consistent option in our opinion, motivated
by quantum formalism and comparison with other fundamental interactions, is to
treat gravitational field as a separate physical object. In this picture the masses are
not directly coupled, but each of them individually interacts with the field. It has
been argued within this mass-field-mass setting that entanglement gain between the
masses implies non-classical features of the field [45, 108, 109], see Chapter 3.
This discussion shows that it would be useful to provide methods for independent
verification of the presence or absence of a physical object mediating the interaction.
We finish with a toy example of a condition capable of revealing that there was no
mediator. To this end we consider two scenarios: (i) evolving a bipartite system
described at time t by a density matrix ρ12 (t); (ii) two objects interacting via a
mediator M , i.e., with Hamiltonian H1M + HM 2 , described by a tripartite state
ρ1M 2 (t). We ask whether there exists bipartite quantum dynamics ρ12 (t) that cannot
be obtained by tracing out the mediator in scenario (ii). Indeed, if ρ12 (t) is a pure
state at all times and entanglement increases, the dynamics could not have been
mediated. The purity assumption requires the mediator to be uncorrelated from
ρ12 (t), and uncorrelated mediator is not capable of entangling the principal system,
composed of particles 1 and 2 [45]. It would be valuable to generalise this argument
to mixed states measured at finite number of time instances.
For example, in an experimental situation, the state of particles 1 and 2 might
only be close to a pure state (with purity 1−, where  is a small positive parameter)
and therefore they could be weakly entangled to the mediator M (with entanglement
∼ O()). One would naturally expect that entanglement gain between particles 1
and 2 is bounded, e.g., as a function of . An observation of higher entanglement
gain therefore excludes the possibility of mediated dynamics.

6.10 Summary
No experiment to date verified that gravity possesses quantum nature. A possible
way towards providing such evidence requires the generation of quantum entangle-
ment between massive objects. In this chapter, we have provided systematic study
of entanglement dynamics between two masses that are coupled gravitationally. We
put forward two proposals where the masses are either trapped in 1D harmonic po-

98
tentials or released from such traps. We derived figures of merit that are useful for
estimating experimental parameters needed to produce entanglement. Our schemes
have the advantage that no macroscopic superpositions develop during the dynam-
ics, resulting in observable entanglement generated within the coherence times of the
masses. Finally, we concluded with a discussion on the nature of the gravitational
interactions that can be infered from our proposals.

99
Chapter 7

Probing quantum features of

photosynthetic organisms

This chapter studies the presence of quantum features in photosynthetic organisms,

particularly the green sulphur bacteria.1 Our aim is to reveal that the organisms can
possess quantumness without directly measuring them. For this purpose, I will start
by presenting our experimental proposal that involves placing the bacteria inside a
Fabry-Perot cavity. In order to show that our proposal is viable, I proceed with
the modelling of the bacterial interactions with cavity fields and environment. The
dynamics of interesting features, such as quantum entanglement, will be presented.
It is confirmed with our simulations that quantum entanglement between the cavity
fields indeed shows that parts of the bacteria are non-classical.

1
Parts of this chapter are reproduced from our published article of Ref. [133],
which is licensed under the Creative Commons Attribution 4.0 International License
(http://creativecommons.org/licenses/by/4.0/). Where applicable, changes made will be indi-
cated.

101
7.1 Motivation and objectives
There is no a priori limit on the complexity, size or mass of objects to which quantum
theory is applicable. Yet, whether or not the physical configuration of macroscopic
systems could showcase quantum coherences has been the subject of a long-standing
debate. The pioneers of quantum theory, such as Schrödinger [134] and Bohr [135],
wondered whether there might be limitations to living systems obeying the laws of
quantum theory. Wigner even claimed that their behaviour violates unitarity [136].
A striking way to counter such claims on the implausibility of macroscopic quan-
tum coherence would be the successful preparation of quantum superposition states
of living objects. A direct route towards such goal is provided by matter-wave inter-
ferometers, which have already been instrumental in observing quantum interference
from complex molecules [137], and are believed to hold the potential to successfully
show similar results for objects as large as viruses in the near future.
However, other possibilities exist that do not make use of interferometric ap-
proaches. An instance of such alternatives is to interact a living object with a
quantum system in order to generate quantum correlations. Should such correla-
tions be as strong as entanglement, measuring the quantum system in a suitable
basis could project the living object into a quantum superposition. Furthermore,
requesting the establishment of entanglement is, in general, not necessary as the
presence of quantum discord, that is a weaker form of quantum correlations, would
already provide evidence that the Hilbert space spanned by the living object must
contain quantum superposition states [24, 36, 37, 46, 47]. For example, by operating
on the quantum system alone one could remotely prepare quantum coherence in the
living object [42].
A promising step in this direction, demonstrating strong coupling between living
bacteria and optical fields and suggesting the existence of entanglement between
them [138], has recently been realised [139]. See also Refs. [140–150] for a broader
picture of quantum effects in photosynthetic organisms. However, the experimental
results reported in Ref. [139] can as well be explained by a fully classical model [138,
139, 151, 152], which calls loud for the design of a protocol with more conclusive
interpretation.
In this chapter we make a proposal in such a direction by designing a thought
experiment in which the bacteria are mediating interactions between otherwise un-
coupled light modes. This scheme fits into the general framework of Ref. [45], which
shows in the present context that quantum entanglement between the light modes
can only be created if the bacteria are non-classically correlated with them during
the process. It is important to realise that in this way we bypass the need of exact
modelling of the living organisms and their interactions with external world. In-

102
deed, experimenters are never asked to directly operate on the bacteria, it is solely
sufficient to observe the light modes. A positive result of this experiment, i.e. obser-
vation of quantum entanglement between the light modes, provides an unambiguous
witness of quantum correlations, in the form of quantum discord, between the light
and bacteria.
In order to demonstrate that there should be observable entanglement in the
experiment we then propose a plausible model of light-bacteria interactions and
noises in the experiment. We focus on the optical response of the bacteria and
model their light-sensitive part by a collection of two-level atoms with transition
frequencies matching observed bacterial spectrum [139]. All processes responsible
for keeping the organisms alive are thus effectively put into the environment of these
atoms. We argue that standard Langevin approach gives a sensible treatment of this
environment due to its quasi-thermal character, low energies compared to optical
transitions and no evidence for finite-size effects. Within this model we find scenarios
with non-zero steady-state entanglement between the light modes which is always
accompanied by light-bacteria entanglement (in addition to quantum discord), which
is in turn empowered by the ultra-strong coupling2 between such systems.

7.2 Proposed setup: Bacteria in a Fabry-Perot

cavity
Our idea is to design a setup which, on one hand, is close to what has already been
realised with bacteria and light, in order to utilise their strong coupling, and whose
description, on the other hand, can be phrased within the framework of Chapter 3.
It was shown there that two physical systems, A and B, coupled via a mediator C,
i.e., described by a total Hamiltonian of the form HAC +HBC , can become entangled
only if quantum discord DAB|C is generated during the evolution. This also holds
if each system is allowed to interact with its own local environment. Therefore,
observation of quantum entanglement between A and B is a witness of quantum
discord DAB|C during the evolution if one can ensure the following conditions:
(i) A and B do not interact directly, i.e., there is no term HAB in the total
Hamiltonian.

(ii) All environments are local, i.e., they do not interact with each other.

(iii) The initial state is completely unentangled (otherwise entanglement between

A and B can grow via classical C [45]).
2
Note that we refer to the ultra-strong coupling as the regime where the so-called counter-
rotating terms in the Hamiltonian have considerable strength. These terms are responsible for
simultaneous creation (and annihilation) of excitations both in the light and bacterial modes.

103
 We now propose a concrete scheme for revealing non-classicality of the bacteria
and argue how it meets the conditions above. Consider the arrangement in Fig. 7.1.
The bacteria are inside a driven single-sided multimode Fabry-Perot cavity where
they interact independently with a few cavity modes. The cavity modes are divided
into two sets which play the role of systems A and B in the general framework. The
bacteria are mediating the interaction between the modes and hence they represent
system C. Condition (i) above can be realised in practice in at least two ways.
An experimenter could utilise the polarisation of electromagnetic waves and group
optical modes polarised along one direction to system A and those polarised orthog-
onally to system B. Another option, which we will study in detail via a concrete
model below, is to choose different frequency modes and arbitrarily group them into
systems A and B. Condition (ii) holds under typical experimental circumstances
where the environment of the cavity modes is outside the cavity whereas that of
the bacteria is inside the cavity or even part of bacteria themselves. The electro-
magnetic environment outside the cavity is a large system giving rise to the decay
of cavity modes but having no back-action on them. Therefore each cavity mode
decays independently and cannot get entangled via interactions with the electro-
magnetic environment. Finally, condition (iii) is satisfied right before placing the
bacteria into the cavity, because at this time all three systems A, B, and C are in
a completely uncorrelated state ρA ⊗ ρB ⊗ ρC .

Figure 7.1: Proposed experimental setup for probing quantum features of photosyn-
thetic organisms. In particular, green sulphur bacteria are mediating interactions
between a few cavity modes in a driven multimode Fabry-Perot cavity, which are
otherwise non-interacting. The cavity has an input mirror with reflectivity R1 and
an end mirror with perfect reflectivity, i.e., R2 ≈ 1. As a result of interactions with
the environment, the mth cavity mode has a decay rate κm , while the bacterial sys-
tem dissipates at a rate γn . In the text, we treat cavity fields and bacterial modes
as open systems.

We note again that this discussion is generic with almost no modelling of the
involved systems. In particular, nothing has been assumed regarding the physics of
the bacteria and their interactions with light and the external world. This makes

104
our proposal experimentally attractive.
In order to make concrete predictions about the amount of intermodal entangle-
ment EA:B we now study a specific model for the energy of the discussed system.
This additional assumption about the overall Hamiltonian will allow us to demon-
strate that the entanglement EA:B is accompanied by light-bacteria entanglement
EAB:C . This independently confirms the presence of light-bacteria discord as entan-
glement is a stronger form of quantum correlations than discord [24, 46, 47]. In the
remainder of this chapter we will therefore only calculate entanglement.

7.3 Model of optical coupling

We consider a photosynthetic bacterium, Chlorobaculum tepidum, that is able to
survive in extreme environments with almost no light [153]. Each bacterium, which
is approximately 2µm × 500nm in size, contains 200 − 250 chlorosomes, each having
200, 000 bacteriochlorophyll c (BChl c) molecules. Such pigment molecules serve as
excitons that can be coupled to light [139, 154]. The extinction spectrum of the
bacteria (BChl c molecules) in water shows two pronounced peaks, at wavelengths
λI = 750nm and λII = 460nm (see Fig. 1b of Ref. [139]). We therefore model the
light-sensitive part of the bacteria by two collections of N two-level atoms with
transition frequencies (ΩI , ΩII ) = (2.5, 4.1) × 1015 Hz. Simplification of this model to
atoms with a single transition frequency was already shown to be able to explain the
results of recent experiments [139, 154]. This simplification was adequate because
only one cavity mode was relevant in the previous experiments. In contrast, several
cavity modes are required for the observation of intermodal entanglement and it
is correspondingly more accurate to include also all relevant transitions of BChl c
molecules. We assume that the molecules (two-level atoms in our model) are coupled
through a dipole-like mechanism to each light mode. For N  1, such collections
of two-level systems can be approximated to spin N/2 angular momenta. In the
low-excitation approximation (which we will justify later), such angular momentum
can be mapped into an effective harmonic oscillator through the use of the Holstein-
Primakoff transformation [155]. This allows us to cast the energy of the overall
system as
X X X
H = ~ωm a†m am + ~Ωn b†n bn + ~Gmn (am + a†m )(bn + b†n )
m n m,n
X
+ i~Em (a†m e−iΛm t − am e iΛm t
). (7.1)
m

Here, m = 1, . . . , M is the label for the mth cavity mode, whose annihilation (cre-
ation) operator is denoted by am (a†m ) and having frequency ωm . Moreover, both

105
harmonic oscillators describing the bacteria are labelled by n = I, II with bn (b†n ) de-
noting the corresponding bosonic annihilation (creation) operator. Each oscillator
is coupled to the mth cavity field at a rate Gmn . The collective form of the coupling
√ p
allows us to write Gmn = gmn N with gmn = µn ωm /2~r 0 Vm , where µn is the
dipole moment of the nth two-level transition, r relative permitivity of medium,
and Vm the mth mode volume [152], see also [156, 157] for similar treatments. The
cavity is driven by a multimode laser, each mode having frequency Λm , amplitude
p
Em = 2Pm κm /~Λm , power Pm , and amplitude decay rate of the corresponding
cavity mode κm . It is important to notice that in Eq. (7.1) we have not invoked the
rotating-wave approximation but actually retained the counter-rotating terms am bn
and a†m b†n . These cannot be ignored in the regime of ultra-strong coupling and we
will show that they actually play a crucial role in our proposal.

7.4 Dynamics of light-bacteria system

7.4.1 Langevin equations
We assume the local environment of the light-sensitive part of the bacteria to give
rise to Markovian open-system dynamics, which is modelled as decay of the two-
level systems. For justification we note that in actual experiments the bacteria are
surrounded by water which can be treated as a standard heat bath and although
the environment of interest cannot be in a thermal state (because the bacteria are
alive) its state is expected to be quasi-thermal. Given that the bacterial environ-
ment of the BChl c molecules is of finite size we should also justify the Markovianity
assumption. To the best of our knowledge there is no experimental evidence against
this assumption. Likely this is due to the fact that all excitations arriving at this en-
vironment are further rapidly dissipated to the large thermal environment of water,
whose energy is small compared to the optical transitions.
We treat the environment of the cavity modes as the usual electromagnetic en-
vironment outside the cavity [158, 159]. This results in independent decay rates
of each mode. Taken all together, the dynamics of the optical modes and bacte-
ria can be written using the standard Langevin formulation in Heisenberg picture.
This gives the following equations of motion, taking into account noise and damping
terms coming from interactions with the local environments
X √
ȧm = −(κm + iωm )am − i Gmn (bn + b†n ) + Em e−iΛm t + 2κm Fm ,
n
X p (7.2)
ḃn = −(γn + iΩn )bn − i Gmn (am + a†m ) + 2γn Qn ,
m

106
where γn is the amplitude decay rate of the bacterial system. Fm and Qn are op-
erators describing independent zero-mean Gaussian noise affecting the mth cavity
field and the nth bacterial mode respectively. The only nonzero correlation func-
tions between these noises are hFm (t)Fm† 0 (t0 )i = δmm0 δ(t − t0 ) and hQn (t)Q†n0 (t0 )i =
δnn0 δ(t − t0 ) [158, 159]. We note that in this model the light-sensitive part of the
bacteria is treated collectively, i.e. all its two-level atoms are indistinguishable. This
assumption is standardly made in present-day literature, see e.g. [139, 154] where
modelling of the bacteria / chlorosomes as a harmonic oscillator fits observed exper-
imental results. But it should be stressed that this assumption deserves an in-depth
experimental assessment.
We express the Langevin equations in terms of mode quadratures. In particular,
√ √
we use Xm ≡ (am + a†m )/ 2 and Ym ≡ (am − a†m )/i 2. This allows us to write
the Langevin equations for the quadratures in a simple matrix equation u̇(t) =
Ku(t) + l(t), with the vector u = (X1 , Y1 , · · · , XM , YM , XI , YI , XII , YII )T and
 
I1 0 ··· 0 L1I L1II
0 I2 ··· 0 L2I L2II
 
 
 .. .. ... .. .. .. 
. . . . .
 
K= , (7.3)
 
 0
 0 · · · IM LM I LM II 

 L1I L2I · · · LM I II 0
 

L1II L2II · · · LM II 0 III

where the components are 2 × 2 matrices given by

! !
−κm ωm 0 0
Im = , Lmn = ,
−ωm −κm −2Gmn 0
!
−γn Ωn
In = , (7.4)
−Ωn −γn
and 0 is a 2 × 2 zero matrix. Note that we have used the index m = 1, 2, · · · , M for
the cavity modes and n = I, II for the bacterial modes. We split the last term in the
matrix equation into two parts, representing the noise and pumping respectively,

107
i.e. l(t) = η(t) + p(t) where
 √   
κ1 X1 (t) E1 cos Λ1 t
 √   
 κ1 Y1 (t)   −E1 sin Λ1 t 
.. ..
   
   

 √ . 


 . 

 κ X (t)
 p(t)  EM cos ΛM t
  
M M
η(t) 
√ =  √κM YM (t)

 , √ =  −EM sin ΛM t . (7.5)
   
2  √  2  
 γI XI (t)   0 
 √
   
  
 √ γI YI (t) 0
   
  
 γ X (t)   0 
II II
√
   
γII YII (t) 0
√
We have also used quadratures for the noise terms, i.e. through Fm = (Xm +iYm )/ 2
√
and Qn = (Xn + iYn )/ 2.
The solution to the Langevin equations is given by
Z t
u(t) = W+ (t)u(0) + W+ (t) dt0 W− (t0 )l(t0 ), (7.6)
0

where W± (t) = exp (±Kt). This allows numerical calculation of expectation value
of the quadratures as a function of time, i.e. hui (t)i is given by the ith element of
Z t
W+ (t)hu(0)i + W+ (t) dt0 W− (t0 )p(t0 ), (7.7)
0

which is obtained as follows. Since every component of p(t) is not an operator, we

have hpk (t)i = tr(pk (t)ρ) = pk (t). Also, we have used the fact that the noises have
zero mean, i.e. hηk (t)i = 0.

7.4.2 Covariance matrix

Covariance matrix of our system is defined as Vij (t) ≡ h{∆ui (t), ∆uj (t)}i/2 =
hui (t)uj (t)+uj (t)ui (t)i/2−hui (t)ihuj (t)i where we have used ∆ui (t) = ui (t)−hui (t)i.
This means that p(t) does not contribute to ∆ui (t) (and hence the covariance ma-
trix) since hpk (t)i = pk (t). We can then construct the covariance matrix at time t
from Eq. (7.6) without considering p(t) as follows

Vij (t) = hui (t)uj (t) + uj (t)ui (t)i/2 − hui (t)ihuj (t)i
V (t) = W+ (t)V (0)W+T (t)
Z t
+W+ (t) dt0 W− (t0 )DW−T (t0 ) W+T (t), (7.8)
0

108
where D = Diag[κ1 , κ1 , · · · , κM , κM , γI , γI , γII , γII ] and we have assumed that the
initial quadratures are not correlated with the noise quadratures such that the mean
of the cross terms are zero. A more explicit solution of the covariance matrix, after
integration in Eq. (7.8), is given by

KV (t) + V (t)K T = −D + KW+ (t)V (0)W+T (t)

+W+ (t)V (0)W+T (t)K T
+W+ (t)DW+T (t), (7.9)

which is linear and can be solved numerically.

Time evolution of important quantities can then be calculated from the covari-
ance matrix, e.g., entanglement and excitation number. If one is only interested in
the steady state, it is guaranteed when all real parts of the eigenvalues of K are
negative. In this case the covariance matrix satisfies Lyapunov-like equation

K V (∞) + V (∞) K T + D = 0, (7.10)

where D = Diag[κ1 , κ1 , · · · , κM , κM , γI , γI , γII , γII ]. Note that the steady-state co-

variance matrix does not depend on the initial conditions, i.e. V (0). Moreover, as
the Langevin equations are linear and due to the gaussian nature of the quantum
noises, the dynamics of the system is preserving gaussianity. Therefore the steady
state is a continuous variable gaussian state completely characterised by V (∞).

7.4.3 Quantum entanglement

Entanglement from covariance matrix

Logarithmic negativity is chosen as entanglement quantifier and below we provide

the details on how this quantity is calculated.
The covariance matrix V describing our system can be written in block form
 
B11 B12 · · · B1Z
T
B12 B22 · · · B2Z
 
 
V = .. .. .. .. , (7.11)

 . . . .


T T
B1Z B2Z · · · BZZ

where Z is the total number of modes, which is M + 2 in our case. The block
component, here denoted as Bjk , is a 2 × 2 matrix describing local mode correlation
when j = k and intermodal correlation when j 6= k. A Z-mode covariance matrix
has symplectic eigenvalues {νk }Zk=1 that can be computed from the spectrum of

109
matrix |iΩZ V | [25] where !
Z
M 0 1
ΩZ = . (7.12)
k=1
−1 0
For a physical covariance matrix 2νk ≥ 1.
Entanglement is calculated as follows. For example, the calculation in the par-
tition 12 : 34 only requires the covariance matrix of modes 1, 2, 3, and 4:
 
B11 B12 B13 B14
T
 
 B12 B22 B23 B24 
V = T T
, (7.13)

 B13 B23 B33 B34 

T T T
B14 B24 B34 B44

that can be obtained from Eq. (7.11). If the covariance matrix Ṽ , after partial trans-
position with respect to mode 3 and 4 (this is equivalent to flipping the sign of the
operator Y3 and Y4 in V ) is not physical, then our system is entangled. This unphys-
ical Ṽ is shown by its minimum symplectic eigenvalue ν̃min < 1/2. Entanglement is
then quantified by logarithmic negativity as follows E12:34 = max[0, − log2 (2ν̃min )]
[34, 35].3 Note that the separability condition, when ν̃min ≥ 1/2, is sufficient and
necessary when one considers bipartitions with one mode on one side [120], e.g.,
partition between bacterial modes I : II.

Simulation parameters

The parameters used in our simulations are taken, wherever possible, from the ex-
periments of Ref. [139]. We place the bacteria in a single-sided Fabry-Perot cavity
of length L = 518 nm (cf. Fig. 7.1). The refractive index due to aqueous bacterial
√
solution embedded in the cavity is nr = r ≈ 1.33, which gives the frequency of
the mth cavity mode ωm = mπc/nr L ≈ 1.37m × 1015 Hz. The reflectivities of the
mirrors are engineered such that R2 = 100% and R1 = 50%. We assume the reflec-
tivities are the same for all the optical modes, giving κm ≈ 7.5 × 1013 Hz through
the finesse Fi = −2π/ ln (R1 R2 ) = πc/2κm nr L. The decay rate of the excitons can
be calculated as γn = 1/2τn where τn = 2h/Γn is the coherence time with Γn be-
ing the full-width at half-maximum (FWHM) of the bacterial spectrum [160]. We
approximate the spectrum in Fig. 1b of Ref. [139] as a sum of two Lorentzian
functions centred at ΩI and ΩII having FWHM of (ΓI , ΓII ) = (130, 600) meV, giving
(γI , γII ) ≈ (0.78, 3.63)×1013 Hz respectively. Note that the decay rate solely depends
on the coherence time, i.e., we assume only homogenous broadening of the spectral
lines.
3
Note that we have used logarithm with base 2, instead of natural logarithm as in Ref. [133],
to maintain consistencies in this thesis.

110
 All the spectral components of the driving laser are assumed to have the same
power Pm = 50mW and frequency Λm = ωm . By using the mode volume Vm =
2πL3 /m(1−R1 ) [161], we can express the interaction strength as Gmn = mG̃n , where
p
we define G̃n ≡ µn c(1 − R1 )N/4~n3r 0 L4 . This quantity is a rate that characterises
the base collective interaction strength of the cavity mode and the nth bacterial
mode. Instead of fixing the value of G̃n , we vary this quantity G̃n = [0, 0.2] 1015 Hz,
which is within experimentally achievable regime (cf. Refs. [138, 139]).

Entanglement dynamics

Let us consider as initial the time right before the bacteria are inserted into the
cavity. Then all the cavity modes and the bacteria are completely uncorrelated and
do not interact. The dynamics is then started by placing the bacteria in the cavity.
In what follows, as an example of the dynamics we start with vacuum state for the
cavity modes and ground state for the bacteria. The initial state of the bacteria is
justified by the fact that ~Ωn  kB T , even at room temperature.
Fig. 7.2 shows the resulting entanglement dynamics. Panel (a) displays exis-
tence of steady-state entanglement between cavity modes 1, 2 and 3, 4, which is not
altered heavily if the calculations take into account five and six cavity modes in
total. Therefore, we consider 4 cavity modes in all other calculations. In recent
experiments, the rate G̃I was shown to be 3.9 × 1013 Hz [139] and the corresponding
G̃II = 6 × 1013 Hz. In our calculations we vary the latter rate as in panels (b) and
(c) (also see Fig. 7.3). As expected the higher the rate the more entanglement gets
generated. It is also apparent that entanglement between the cavity modes and
bacteria E1234:I II grows faster than entanglement between the cavity modes. More
precisely, nonzero E12:34 implies nonzero E1234:I II .

Steady state entanglement

We consider four cavity modes as the addition of higher modes shows negligible
effects to the steady state entanglement. In the steady state regime, we calculate
entanglement between the cavity modes E12:34 , between the cavity modes and bac-
teria E1234:I II , and between the bacterial modes EI:II , cf. Fig. 7.3. This steady
state regime is reached in ∼ 100 fs (see Fig. 7.2), which is faster than relaxation
processes (∼ ps) occuring within green sulphur bacteria [154]. Our results show
that the steady state entanglement E12:34 is always accompanied by E1234:I II , i.e.,
the bacteria are non-classically correlated with the cavity modes. This is in agree-
ment with the general detection method of Ref. [45] as entanglement is a stronger
type of quantum correlation than discord, i.e., nonzero E1234:I II implies nonzero cav-
ity modes-bacteria discord D1234|I II . Our results also show that the entanglement

111
Figure 7.2: Entanglement dynamics in the field-bacteria-field system. (a) Entangle-
ment E12:34 between the cavity modes {1, 2} and {3, 4}. The entanglement is shown
up to 4, 5, and 6 cavity modes in the cavity. Higher modes do not contribute to the
steady state entanglement due to higher detuning from the bacterial system. We
therefore only consider 4 cavity modes for the rest of the dynamics. (b) Entangle-
ment between the cavity modes for varied interaction strength with the bacterial
system. (c) Cavity modes–bacteria entanglement. The correlation from this parti-
tion is higher and showing faster initial growth compared to the others. The coupling
parameters used are G̃I = 3.9 × 1013 Hz for all graphs and G̃II = 6 × 1013 Hz for (a).
Note that the steady state of all dynamics is reached within ∼ 100 fs and we have
used the logarithmic negativity as entanglement quantifier.

dynamics of E12:34 is dominated by modes 2 and 3 since other modes are further
off resonance with the bacterial modes. Moreover, there is entanglement generated
within the bacteria. This requires both G̃I and G̃II to be nonzero and relatively
high. We see that the bacteria can be strongly entangled with the cavity modes,
much stronger than entanglement between the cavity modes. While the latter is in
the order of 10−2 − 10−3 , we note that entanglement in the range 10−2 has already
been observed experimentally between mechanical motion and microwave cavity
field [114]. We have also indicated, as black dots in Fig. 7.3, the coupling strengths

112
G̃I = 3.9 × 1013 Hz from Ref. [139] and the corresponding G̃II = 6 × 1013 Hz, which
R
is estimated as follows. From the relation µ2n ∝ f (ω)dω/ωn [162], where f is the
extinction coefficient, one can obtain the ratio G̃II /G̃I = µII /µI ≈ 1.53.

Figure 7.3: Steady state entanglement in field-bacteria-field system for varied inter-
action strengths. In all panels, the horizontal axis is G̃I = [0, 0.2] 1015 Hz, while G̃II
is varied in 1015 Hz as: 0 (red curves), 0.05 (green curves), 0.1 (blue curves), 0.15
(magenta curves), and 0.2 (black curves). Black dots in the panels show the values
of entanglement for recently realised coupling strengths G̃I = 3.9 × 1013 Hz by Coles
et al. [139] and the corresponding G̃II = 6 × 1013 Hz. We note from our simulations
that the entanglement E1234:I II is nonzero whenever E12:34 is present. Furthermore,
entanglement in the bacterial system, panel (c), also persists for stronger couplings.

7.4.4 Excitation number

We point out that the covariance matrix V (t), and hence the entanglement, does not
depend on the power of the lasers. This is a consequence of the dipole-dipole coupling
and classical treatment of the driving field. Therefore, the system gets entangled also
in the absence of the lasers. There is no fundamental reason why this entanglement
with vacuum could not be measured, but practically it is preferable to pump the
cavity in order to improve the signal-to-noise ratio. Of course quantities other than
entanglement may depend on driving power, for example the light intensity inside
the cavity.
This finding is quite different from results in optomechanical system where the
covariance matrix depends on laser power [45, 163]. The origin of this difference is
the nature of the coupling. For example, in an optomechanical system consisting of

113
a single cavity mode A and a mechanical mirror B the coupling is proportional to
a† aXB , which is a third-order operator [164]. This results in the effective coupling
strength being proportional to the classical cavity field intensity α after linearisation
of the Langevin equations. This classical signal enters the covariance matrix via the
effective coupling strength and introduces the dependence on the driving power.
In order to justify the low atomic excitation limit we first note that the number
of steady-state photons for the mth cavity mode without the presence of the bacteria
2
is given by Em /κ2m ∝ Pm . When one considers the bacteria in the cavity having the
base interaction strength G̃n and a decay rate γn in the same order as the cavity
decay rate, the number of excitations of the bacterial modes would also be in the
2
order of Em /κ2m , which in our case is 103 . With ∼ 108 actively coupled dipoles in
the cavity [139], this gives ∼ 10−3 % excitation, which justifies the low-excitation
approximation. We also plotted the evolution of excitation numbers of the bacterial
modes (together with the number of photons in different cavity modes) within our
model in Fig. 7.4. It shows that excitation numbers are oscillating in the “steady
state”. The oscillations are caused by the combination of interactions between the
light and bacteria (Rabi-like oscillations) and the time-dependent driving laser. Set-
ting the interactions Gmn = 0 or the driving off (Pm = 0) indeed produces constant
steady-state value. We observe that the excitation number of the bacterial system
is always below 2000, which is in agreement with the statement above.
The excitation number of the cavity modes and bacteria as a function of time
can be calculated from hui (t)i and Vii (t). For example, the mean excitation number
for the first cavity mode is given by

1
N̄1 (t) = ha†1 (t)a1 (t)i = (V11 (t) + V22 (t) + hu1 (t)i2 + hu2 (t)i2 − 1). (7.14)
2

We present the evolution of photon number of the cavity modes and excitation
of the bacterial modes in Fig. 7.4. Note that photon number of the third cavity
mode (solid magenta line) is showing oscillations well below its “off-interaction”
value (dashed magenta line). This is because ω3 is almost in resonance with the
frequency of the atomic transition ΩII .

7.5 Entanglement as a witness of ultra-strong cou-

pling
We also performed similar calculations in which we neglected the counterrotating
terms in Eq. (7.1), the model known as Tavis-Cummings. This resulted in no
entanglement generated in the steady state and can be intuitively understood as
follows. Since the steady-state covariance matrix does not depend on the initial

114
Figure 7.4: Dynamics of the bacterial excitation numbers and photon number of
the cavity modes. Solid curves represent the interacting system: N̄1 , N̄2 , N̄3 , N̄4
for the cavity modes and N̄I , N̄II for the bacterial system. The couplings are taken
as G̃I , G̃II = (3.9, 6) × 1013 Hz. The corresponding dynamics for non-interacting
system, i.e., Gmn = 0, are given by the dashed curves.

state and on the power of the driving lasers, we might start with all atoms in the
ground state, vacuum for the light modes, and no driving. Under such circumstances
there is no interaction between bacterial modes and light modes as every term in the
interaction Hamiltonian contains an annihilation operator. In physical terms, since
we begin with the lowest energy state and the interaction Hamiltonian preserves
energy, the ground state will be the state of affairs at any time. Therefore, nonzero
entanglement observed in experiments will provide evidence of the existence of the
counterrotating terms, showing a signature of the ultra-strong coupling regime.

7.6 Summary
Recent experimental studies have shown that the regime of strong coupling between
photosynthetic organisms such as green sulphur bacteria and light is accessible. In
this chapter, we put forward a proposal for probing a quantum property of the bac-
teria (as characterised by quantum correlations) without directly measuring them.
Our general proposal bypasses the need of modelling both the bacteria and the in-
teractions they have with external world. The proposed scheme utilises the bacteria
as mediators between two groups of non-interacting cavity light modes. In order to
show feasibility, we modelled the light-sensitive part of the bacteria and their interac-
tions with the cavity modes and environments using recent experimental parameters.
Within this model, our simulations show that, in most cases, quantum entanglement
endures the environmental noises and remains present in the steady state. Our sim-
ulations also confirmed that entanglement between the cavity light modes reveals a
quantum property of the bacteria, i.e., quantum entanglement between the bacterial
modes and cavity modes, and for stronger interactions, even quantum entanglement

115
within the bacterial modes. We note that the presence of entanglement also provides
independent evidence of the ultra-strong coupling regime.

116
Chapter 8

Other applications

In this chapter, I will present other applications that result from the principles of
distribution of correlations between two objects whose interactions are mediated by
an ancillary system – investigated in Chapters 3 to 5.1 First, I will apply our
non-classicality detection method to an experimentally relevant platform of optome-
chanics, in particular, the membrane-in-the-middle setting. Our protocol indirectly
detects the quantum property of the mechanical membrane by observing the dynamics
of entanglement between probing light modes. Next, I present a protocol to reveal
the correlation between a system and its environment in a scenario of open system
dynamics. Then, high gain of correlations will be demonstrated in a dynamics of
two cavity fields interacting with a two-level atom. This gain will be shown to imply
a quantum trait – non-commutativity of interaction Hamiltonians. Finally, a brief
discussion will be presented that shows an application of our method to estimate
dimensionality of a quantum object.

1
Parts of this chapter are reproduced from our published articles of Refs. [45, 55], c [2019]
American Physical Society. Where applicable, changes made will be indicated.

117
8.1 Membrane-in-the-middle optomechanics
In this section, we address the practical implication of our criteria from Chapter 3 to
experiments of cavity optomechanics [110]. This is a paradigmatic open mesoscopic
quantum system for which the non-classicality detection method in Chapter 3 holds
the potential to be practically significant. In fact, one of the goals of optomechanics
is to infer the non-classicality of the state of a massive mechanical system, in similar
spirit as “certification” in Refs. [165, 166], without affecting its (in general fragile)
state. A possible setting for such a task is given by the so-called membrane-in-
the-middle configuration, where a mechanical oscillator (a membrane) is suspended
at the centre of a two-sided optical cavity [163]. By driving the cavity with laser
fields from both its input mirrors, respectively, we realise a situation completely
analogous to that in Fig. 3.1 (cf. Fig. 8.1). We now show that our scheme detects
non-classicality of the membrane without measuring it.

8.1.1 Experimental setup

The hamiltonian of our setup (Fig. 8.1 below) in a rotating frame with frequency of
the lasers can be written as H = Hloc + Hint where [163]:

~ωC 2
Hloc = ~∆0A a† a + ~∆0B b† b + (PC + XC2 )
2
+i~EA (a† − a) + i~EB (b† − b) (8.1)

and
Hint = −~G0A a† a XC + ~G0B b† b XC , (8.2)

where the annihilation (creation) operator of field J = A, B is denoted by the

corresponding lowercase letter j (j † ) with [j, j † ] = 1, PC and XC are dimensionless
quadratures of the membrane with [XC , PC ] = i, EJ is the driving strength of laser
J with |EJ | = (2PJ κJ /~ωlJ )1/2 , where PJ is the laser power and ωlJ denotes its
frequency. κJ = πc/2Fi,J lJ is decay rate of cavity J with finesse Fi,J . Cavity-laser
detuning is defined as ∆0J ≡ ωJ − ωlJ , where ωJ is the frequency of the cavity and
G0J = (ωJ /lJ )(~/mC ωC )1/2 represents field-membrane coupling strength, where lJ
is the length of the cavity, mC is the mass of the membrane and ωC is its natural
frequency. Note that Hloc is local in A : B : C partition and the two terms in Hint
represent coupling in the partition A : C and B : C respectively. All the other
interactions are local, i.e., A (B) is coupled to its own environment A0 (B 0 ) and C is
coupled to its thermal phonon reservoir C 0 , responsible for the Brownian motion of
the membrane. Thus, our Theorem 3.3.1 directly applies here and we can implement
the detection method of Fig. 3.4a.

118
Figure 8.1: Membrane-in-the-middle optomechanical setup. Cavity field A is inter-
acting indirectly with cavity field B via a perfectly reflecting mechanical membrane
C. The interactions of the movable mechanical mirror with its thermal environment
result in the Brownian noise. The cavities are driven by lasers and experience energy
dissipation through the fixed input mirrors.

8.1.2 Dynamics of the system

Langevin equations

The dynamics of the operators, adding into account noise and damping terms (also
local), can be well written by a set of Langevin equations in Heisenberg picture
√
ȧ = −(κA + i∆0A )a + iG0A aXC + EA + 2κA ain
√
ḃ = −(κB + i∆0B )b − iG0B bXC + EB + 2κB bin
X˙ C = ωC PC
P˙C = −ωC XC + G0A a† a − G0B b† b − γC PC + ξ (8.3)

where γC is damping rate of the membrane. Also jin is input noise of field J associ-
ated with cavity-input mirror interface and has only correlation function
†
hjin (t)kin (t0 )i = δjk δ(t − t0 ) [159], whereas ξ is Brownian noise of the membrane and
has correlation function hξ(t)ξ(t0 ) + ξ(t0 )ξ(t)i/2 ≈ γC (2n̄ + 1)δ(t − t0 ) in the limit of
interest that is large mechanical quality of the membrane, i.e., ωC /γC  1 [118, 119].
The mean phonon number of the membrane reads n̄ = 1/(exp (~ωC /kB T ) − 1).
The linearised Langevin equations can be obtained by splitting the operators into
steady state values and fluctuating terms. In particular we write XC = XCs + δXC ,
PC = PCs + δXC , and j = αJs + δj. By inserting these into Eq. (8.3) and ignoring
nonlinear terms δj † δj and δj δXC one gets a set of linear Langevin equations for

119
the fluctuations of the quadratures
√
δ X˙ A = −κA δXA + ∆A δYA + 2κA xin,A
√
δ Y˙A = −κA δYA − ∆A δXA + GA δXC + 2κA yin,A
√
δ X˙ B = −κB δXB + ∆B δYB + 2κB xin,B
√
δ Y˙B = −κB δYB − ∆B δXB − GB δXC + 2κB yin,B
δ X˙ C = ωC δPC
δ P˙C = −ωC δXC − γC δPC + GA δXA − GB δXB + ξ (8.4)

where effective detuning ∆A ≡ ∆0A − G0A XCs , ∆B ≡ ∆0B + G0B XCs , and effective
√
coupling GJ ≡ 2G0J αJs . The steady state values are given by PCs = 0, XCs =
p
(G0A |αAs |2 − G0B |αBs |2 )/ωC , and αJs = |EJ |/ κ2J + ∆2J . The quadratures of the
√
field XJ and YJ are related to the field operator j through j = (XJ + iYJ )/ 2.
√
This relation also applies for the input noise, i.e., jin = (xin,J + iyin,J )/ 2.
For simplicity one can re-write Eqs. (8.4) as a single matrix equation u̇(t) =
Ku(t) + n(t) where the vector u(t) = (δXA , δYA , δXB , δYB , δXC , δPC )T ,
√ √ √ √
n(t) = ( 2κA xin,A , 2κA yin,A , 2κB xin,B , 2κB yin,B , 0, ξ)T , (8.5)

and  
−κA ∆A 0 0 0 0
 

 −∆A −κA 0 0 GA 0 

 0 0 −κB ∆B 0 0 
K= . (8.6)
 

 0 0 −∆B −κB −GB 0 


 0 0 0 0 0 ωC 

GA 0 −GB 0 −ωC −γC
The solution to this linearised Langevin equation is then given by u(t) = M (t)u(0)+
Rt
0
dsM (s)n(t − s) where M (t) = exp (Kt).

Dynamics of covariance matrix

The quantum state of the fluctuations is fully characterised by covariance ma-

trix Vij (t) ≡ hui (t)uj (t) + uj (t)ui (t)i/2 − hui (t)ihuj (t)i. Note that the Gaussian
nature of the initial state is maintained since we have linear dynamics and the
noises involved are zero mean Gaussian noises. One can show that the covari-
Rt
ance matrix at time t is V (t) = M (t)V (0)M T (t) + 0 ds M (s)DM T (s) where
D = Diag[κA , κA , κB , κB , 0, γC (2n̄ + 1)]. A more explicit solution of the covariance

120
matrix, after integration, is given by

KV (t) + V (t)K T = −D + KM (t)V (0)M T (t)

+M (t)V (0)M T (t)K T
+M (t)DM T (t), (8.7)

which is linear and can easily be solved numerically. For our simulations of the
dynamics below, we take the initial state to be thermal state for c and coherent
state for field j, this gives V (0) = Diag[1, 1, 1, 1, 2n̄ + 1, 2n̄ + 1]/2. If one is only
interested in steady state solution, it is guaranteed when all real parts of eigenvalues
of K are negative, giving M (∞) = 0 such that the steady state covariance matrix
can be calculated from a simpler equation KV (ts ) + V (ts )K T = −D.

Quantum entanglement: dynamics and steady state

In order to calculate entanglement, we utilise here the method presented in Chap-

ter 2. In particular, the covariance matrix V describing our three-mode optome-
chanical system can be written in block form
 
LAA LAB LAC
VABC =  LTAB LBB LBC  (8.8)
 

LTAC LTBC LCC

where for j, k = A, B, C the block component Ljk is a 2 × 2 matrix describing local

mode correlation when j = k and intermodal correlation when j 6= k. An N -mode
covariance matrix has symplectic eigenvalues {νk }N
k=1 that can be computed from
the spectrum of matrix |iΩN V | [25] where

N
!
M 0 1
ΩN = . (8.9)
k=1
−1 0

For a physical covariance matrix 2νk ≥ 1. For an entangled system, e.g., in the par-
tition AB : C, the covariance matrix will not be physical after partial transposition
with respect to mode C (this is equivalent to flipping the sign of the membrane’s
momentum fluctuation operator δPC in V ). For our system, this unphysical V TC
is shown by one of its three symplectic eigenvalues ν̃min < 1/2. Entanglement is
then quantified by logarithmic negativity as follows EAB:C = max[0, − log2 (2ν̃min )]
[34, 35].2 Note that the separability condition, when V TC has ν̃min ≥ 1/2, is suffi-
cient and necessary for 1 : N mode partition [120]. Entanglement EA:B is calculated
2
Note that we have used logarithm with base 2, instead of natural logarithm as in Ref. [45], to
maintain consistencies in this thesis.

121
in similar manner by only considering system AB where the covariance matrix is
now !
LAA LAB
VAB = . (8.10)
LTAB LBB
In order to independently confirm the non-classicality of the membrane and
demonstrate that there is considerable entanglement to be detected we calculate the
ensuing entanglement dynamics. As mentioned above, we start with the experimen-
tally natural state where C is in a thermal state and A and B are coherent states,
and calculate the dynamics of entanglement EA:B and EAB:C . Since initially there
is no entanglement, the first step in Fig. 3.4a can be omitted.
The parameters used in our simulations all adhere to present-day technology
[167]. This includes mC = 145 ng, T = 300 mK, lJ = 25 mm, and (ωC , ωlJ , γC ) =
2π(947 × 103 , 2.8 × 1014 , 140) Hz. Finesse of each cavity is 1.4 × 104 . The results of
our analysis are presented in Fig. 8.2 for varying power of the right laser. We see
that non-zero EA:B (τ ) is always accompanied by non-zero EAB:C at some time (0, τ ).
Note that entanglement is a stronger type of quantum correlations than discord. We
have also performed similar calculations by varying the power of the left laser as well
as the frequencies of the lasers within experimentally accessible ranges and observed
consistent results.

Figure 8.2: Exemplary dynamics of entanglement in the membrane-in-the-middle

optomechanical setup. We take the feasible parameters from recent experiments by
Gröblacher et al. [167]. In particular, the mirror has mass mC = 145 ng, damping
rate γC = 2π × 140 Hz, natural frequency ωC = 2π × 947 kHz, and environmental
temperature T = 0.3 K. Each cavity has length 25 mm, finesse 1.4 × 104 , and is
driven by 1064 nm laser. We fixed the power of the left laser to be PA = 100 mW,
and the detunings ∆A = ωC and ∆B = −ωC , while varying the power of the right
laser as PB = 20 mW (green curves), 40 mW (blue curves), 60 mW (red curves),
and 80 mW (black curves). Our simulations show that positive entanglement EA:B
implies that the mirror is entangled with the cavity fields, i.e., nonzero EAB:C during
the evolution.

If one is interested only in the steady state regime, Fig. 8.3 shows the corre-

122
sponding entanglement EA:B while EAB:C is zero in this range (not shown). Note
that red colour has been used in the plots for parameters that do not correspond to
steady state solution.

Figure 8.3: Exemplary steady state entanglement between the cavity fields in the
membrane-in-the-middle optomechanical setup. The power of the right laser is var-
ied as PB = [0, 100] mW and the detuning ∆B = [−1.05, −0.85] ωC . Other param-
eters are the same as those stated in Fig. 8.2. Red colour region corresponds to
non-steady state regime.

8.2 Detecting system-environment correlations

In this section we present a brief review of schemes for detecting system-environment
correlations. We show that our protocol is capable of such detection without the
need for state tomography.

8.2.1 Previous detection schemes

A vast body of literature exists on the study of the influence of initial system-
environment correlations (SECs) on the evolution of the open system [168]. Propos-
als for the detection of SECs based on monitoring the dynamics of distinguishability
[169–173] or purity [174, 175] of the open system have been put forward. Such
proposals have been implemented experimentally by means of quantum tomogra-
phy [176, 177]. Moreover, the possible non-classical nature of SECs was linked to
the impossibility of describing the evolution of an open system through completely
positive maps [178]. Hence detection schemes of quantum discord in the initial
system-environment state have been proposed [179, 180] and recently assessed ex-
perimentally [181–183].

123
8.2.2 Our protocol
In order to apply our method in this scenario, let us consider again a closed-system
dynamics and, in line with the assumed inaccessibility of the mediator, focus the at-
tention to the probes only (see Fig. 3.1). We could thus think of C as an environment
in contact with the open system AB, as seen in Fig. 8.4.

Figure 8.4: A scenario to reveal system-environment correlations. Two mutually

non-interacting objects A and B are in contact with a common environment C.
We show that correlations between the open system AB and the environment can
be inferred from the gain of entanglement between A and B. Note that the setup
considered here is different from the one in Fig. 3.1 of Chapter 3, where now C
includes local or global environments or mediating objects.

Now we use our scheme of Fig. 3.4a to reveal SECs, with the advantage that
the applications of an entanglement breaking channel and state tomography are
not necessary. This is a consequence of Lemma 3.3.2 that, if rephrased in terms
of the aim in this section, states that quantum entanglement in the open system
between two objects A and B cannot increase if the environment is not correlated
with them at all times. Note that the amount of entanglement can be quantified by
any entanglement monotone.
Recall also that our detection protocol of Fig. 3.4a is used to infer non-classicality
of correlations, i.e., quantum discord, between the environment and the open system.
We note that previous schemes detect the non-classicality of the system [179, 180],
i.e., presence of DC|AB , whereas our schemes ascertain the non-classicality of the
environment, DAB|C , which is perhaps a prime example of an inaccessible object.

8.3 Two cavity fields coupled via a two-level atom

In this section, we demonstrate, with concrete dynamics generated by non-commuting
Hamiltonians, that the correlation capacity bounds derived in Chapter 4 can be vi-
olated. We also provide similar calculations for commuting Hamiltonians.

124
8.3.1 Witnessing non-commutativity and discord
Consider a two-level atom C, i.e., dC = 2, mediating interactions between two cavity
fields A and B as illustrated in Fig. 8.5. A similar scenario has been considered and
implemented, for example, in Refs. [60, 61, 184, 185]. The interaction between the
atom and each cavity field is taken to follow the Jaynes-Cummings model,

H = ~g(aσ + + a† σ − ) + ~g(bσ + + b† σ − ), (8.11)

where a (b) is the annihilation operator of field A (B), while σ + (σ − ) is the raising
(lowering) operator of the two-level atom. For simplicity, we have assumed that the
interaction strengths between the two-level atom and the fields are the same. Note
that H is of the form HAC + HBC with non-commuting components.

Figure 8.5: Two orthogonally polarised cavity fields interact via a two-level atom,
but not with each other. In the text, we show that correlation gain between A and
B is linked with non-commutativity of interaction Hamiltonians, or more generally,
non-decomposability of time evolution.

The resulting correlation dynamics are plotted in Fig. 8.6. Mutual informa-
tion and negativity were calculated directly, whereas for the classical correlation
and the relative entropy of discord, we provide the lower bounds C̃A:B and −SA|B ,
respectively. C̃A:B is calculated as the mutual information of the state resulting
from projective local measurements in the Fock basis (no optimization over mea-
surements performed). The negative conditional entropy −SA|B is a lower bound
on the distillable entanglement [186], which in turn is a lower bound on the rela-
tive entropy of entanglement EA:B [187]. Therefore, we note the chain of inequal-
ities −SA|B ≤ EA:B ≤ DA|B ≤ IA:B , where the last two inequalities follow from
[38]. Already these lower bounds can beat the limit set by decomposable evolu-
tion, and therefore, all mentioned correlations can detect non-decomposability of
the evolution. Since we consider closed systems, this infers non-commutativity of
the Jaynes-Cummings couplings. We also note another non-classical feature of the
studied dynamics: since Fig. 8.6 shows entanglement gain, according to Chapter 3
there must be quantum discord DAB|C during the evolution.

125
 It is apparent that the detection is easier (faster and with more pronounced vio-
lation) with a higher number of photons in the initial states of the cavity fields.
We offer an intuitive explanation. Consider, for example, |mn0i as the initial
√
state of ABC. By defining ξ = (a + b)/ 2, the Hamiltonian of Eq. (8.11) be-
√
comes 2~g(ξσ + + ξ † σ − ) and it is straightforward to obtain the unitary evolu-
tion [188]. One finds that the quantum state of the fields oscillates incoherently
between m+n
P Pm+n−1
j=0 cj (t)|jiA |m + n − jiB and j=0 dj (t)|jiA |m + n − 1 − jiB . Both of
these states are superpositions of essentially m + n bi-orthogonal terms giving rise
to high entanglement and, therefore, also other forms of correlations.
Figure 8.6 illustrates that different correlation quantifiers have different detection
capabilities and it is not clear at this stage whether there is a universal measure with
which non-commutativity is detected, e.g., the fastest. For most initial states we
studied mutual information detected non-commutativity the most rapidly, but there
are exceptions, as shown by the black curve corresponding to the initial state |101i.
With this initial state the mutual information never violates its bound, but the
negativity does.

Figure 8.6: Correlation dynamics in the field-atom-field system. The solid curves
represent the Jaynes-Cummings coupling while the dashed lines show the corre-
sponding correlation bound for decomposable dynamics. In the panels, we show
correlations between field A and field B: (a) the mutual information, (b) lower
bound on the classical correlation, (c) lower bound on the relative entropy of dis-
cord, and (d) the entanglement as quantified by negativity. The varied parameter
is the initial state of the tripartite system ABC: |110i (red curves), |101i (black
curves), |210i (green curves), and |220i (blue curves). Note that the horizontal axis
is given by the dimensionless time gt.

126
8.3.2 Strong interactions and bounded entanglement
Our results imply that the non-commutativity (non-decomposability in general) is a
desired feature of interactions in the task of correlation distribution, which is impor-
tant for quantum information applications. As a contrasting physical illustration,
we consider the strong dipole-dipole interactions in our field-atom-field example.
The Hamiltonian reads

H 0 = ~g(a + a† )(σ + + σ − ) + ~g(b + b† )(σ + + σ − ), (8.12)

with commuting components, i.e., [HAC , HBC ] = 0. One can verify that the results
of this model are in agreement with all the bounds we derived in Chapter 4. Further-
more, we prove below that, with this coupling, the state of AB at time t is effectively
given by a two-qubit separable state. This makes NA:B (t) = 0 and IA:B (t) ≤ 1. Note
the counter-intuitive result that strong interactions produce bounded correlations
between the probes, while weak interactions (Jaynes-Cummings coupling) can in-
crease the correlations above the bounds.
√
Let us define ξ = (a + b)/ 2. The dipole-dipole Hamiltonian, Eq. (8.12), is
√
reformulated as H 0 = 2~g(ξ + ξ † )σ x , where σ x = σ + + σ − and [ξ, ξ † ] = 1. The
unitary evolution operator is given by

iH 0 t
Ut = e− ~ (8.13)
1 √ †
√ †
= [(1 − σ x )ei 2gt(ξ+ξ ) + (1 + σ x )e−i 2gt(ξ+ξ ) ]
2
1
= [(1 − σ x )Da (α)Db (α) + (1 + σ x )Da (−α)Db (−α)],
2

where α = igt and, e.g., Da (α) = exp(αa† − α∗ a). Given an initial state |mn0i, the
state at time t reads

1 (mn) (m) (n) (mn) (m) (n)

|ψt i = [(d |D+ , D+ i + d−− |D− , D− i)|0i
4 ++
(mn) (m) (n) (mn) (m) (n)
− (d+− |D+ , D− i + d−+ |D− , D+ i)|1i],

where
q
(mn)
d±± = 2 [1 ± e−2|α|2 Lm (4|α|2 )][1 ± e−2|α|2 Ln (4|α|2 )],
(n) 1
|D± i = q [D(α) ± D(−α)]|ni.
(nn)
d±±

(n) (n) (n) (n)

Note that hD+ |D− i = 0 and hD± |D± i = 1. Ln (|α|2 ) is the Laguerre polynomial,
2
which comes from the relation hn|D(α)|ni = e−|α| /2 Ln (|α|2 ). After tracing-out of

127
the atomic mode C, the state of the fields is effectively given by a two-qubit state,
 (mn) (mn) (mn)

(d++ )2 0 0 d++ d−−
 (mn) 2 (mn) (mn) 
1 
 0 (d+− ) d+− d−+ 0 
,
(mn) (mn) (mn)
16 
 0 d+− d−+ (d−+ )2 0 

(mn) (mn) (mn) 2
d++ d−− 0 0 (d−− )
(8.14)

which is positive under partial transposition and, hence, separable [30, 51]. The
same result follows for initial state |mn1i.

8.3.3 Estimating dimension of mediators

Last but not least, we note an application of our correlation capacity bounds to es-
timate the dimension of the mediator; see, e.g., Refs. [189–191] for other dimension
witnesses. For decomposable evolution (including discrete sequential operators con-
sidered in Refs. [7, 15–20]), the amount of correlation between the probes is bounded
by the correlation capacity sup|ψi QA:C , which is a function of dC . If one observes
QA:B (t) value that is larger than the correlation capacity of a certain dC , then the
dimension of the mediator must be larger than dC .

128
Chapter 9

Conclusion and future work

In this chapter, I present the conclusion of the work that has been reported in this
thesis. Additionally, some immediate questions arise, which have not yet been set-
tled. I will discuss these key questions with preliminary results, wherever possible.
In particular, this includes a natural question on the general entanglement bound
for the indirect interaction scenario, a quantitative bound on the amount of non-
commutativity of Hamiltonians (in general, non-decomposability of time evolution
operator), a request on the protocol capable of witnessing the presence of quantum
objects (i.e., the ones that can be given a description within the quantum framework
such as living in a Hilbert space), and a strict bound on entangling time for initial
states where the mediator is decoupled.

129
9.1 Conclusion
We have studied the creation of correlations, mostly quantum entanglement, be-
tween two principal quantum objects that are continuously interacting via a medi-
ator – the indirect interaction setting. The research in this thesis has been devoted
to understanding the factors that are crucial in this scenario. It includes the re-
quired property of the mediator, the amount of correlation gain, and the speed at
which entanglement is created. Our work also resulted in key applications ranging
from the current technologically available platform detecting non-classicality of an
optomechanical mirror to a proposal on revealing a quantum nature of gravity.
In particular, Chapter 3 has shown that, for initially uncorrelated objects, quan-
tum discord between the mediator and the principal objects is a necessary condi-
tion for distribution of quantum entanglement. The contrapositive of our theorem
is therefore the revelation of quantum discord (a form of non-classical feature of
the mediator) from the observation of entanglement gain between the principal ob-
jects. Next, in Chapter 4, we have made a connection between the amount of
distributed correlations and non-commutativity of Hamiltonians or, more generally,
non-decomposability of the dynamical operator (another form of non-classicality).
The speed of entanglement creation has been investigated in Chapter 5. We have
presented a lower bound on entangling time for both direct and indirect interaction
settings. We proved that entanglement cannot be indirectly distributed faster than
the direct time bound. However, some examples, all of which require an initially
correlated mediator, have been shown to saturate the bound. We note that in Chap-
ters 3–5, wherever possible, we have used minimalistic assumptions in order to make
our theorems as general as possible. This includes the relaxation of knowledge of
the dimensionality of all the objects under scrutiny, the initial state of the system,
and details of interactions. All the objects can also be open systems.
Applications resulted from this work have been reported in three chapters. First,
Chapter 6 focussed on our proposal on the observation of quantum entanglement
between two masses. This has been argued as a potential route towards the detec-
tion of a quantum nature of gravitational interactions. Next, Chapter 7 has shown
another proposal on probing of non-classicality of photosynthetic organisms, in par-
ticular, the green sulphur bacteria, without measuring them directly. From our
model of the bacterial modes and their interactions, we have confirmed our probing
scheme. Finally, other applications were presented in Chapter 8. This includes,
among others, the revelation of a non-classical property of a mechanical mirror that
is mediating interactions between two otherwise non-interacting cavity light modes.

130
9.2 Future work
9.2.1 Entanglement bound for indirect continuous interac-
tions
In Chapter 3, we considered the indirect interaction setting where the interactions
are continuous. An illustration can be seen in Fig. 3.1 where an ancillary object C
is mediating interactions between two principal objects A and B. In this case, we
have proven that classical mediator (zero discord DAB|C at all times) cannot increase
quantum entanglement, i.e., EA:BC (τ ) ≤ EA:BC (0). See Theorem 3.3.1 for details.
However, we note that having positive discord is not sufficient for entanglement gain.
Moreover, we often deal with imperfections in real situations. It has been shown
that states with zero discord are impossible to prepare in experiments, i.e., small
perturbations will drive a zero-discord state into a state having positive, albeit small,
discord [192]. As an example, let us consider a state of the form

ρ = p ρAB ⊗ |αi hα| + (1 − p) ρ0AB ⊗ |α0 i hα0 | , (9.1)

where p stands for probability, {ρAB , ρ0AB } are states of system AB, and {|αi , |α0 i}
are coherent states of C. Here we note that the states |αi and |α0 i are non-
orthogonal. In the limit where the expectation values of position, hα|x|αi and
hα0 |x|α0 i are far apart, the states will be effectively orthogonal. Theoretically, how-
ever, quantum discord DAB|C of the state in (9.1) will never be exactly zero. Yet, one
expects that not much entanglement can be gained with the effectively orthogonal
states. This calls for a bound on entanglement gain in terms of quantum discord
DAB|C , or other related quantities.1 Indeed, an observation of finite entanglement
gain would certify that useful quantum discord (i.e., finite) formed during the dy-
namics.
We have performed preliminary investigation towards this direction. First, we
note from the example in Fig. 3.2 that entanglement increment in the partition
A : BC within ωt = π/4 (EA:BC = 1 at this time), is not bounded by the maximum
discord (≈ 0.81) or the average discord (≈ 0.38). For a special class of initial states
and Hamiltonian, we have obtained a simple bound. In particular, the gain of
entanglement (as quantified by REE) is always bounded by entanglement with the
mediator, i.e.,
|EA:BC (τ ) − EA:BC (0)| ≤ EAB:C (τ ). (9.2)

This applies for a coherent dynamics with commuting interaction Hamiltonians,

1
For example, one might consider quantum coherence of object C (see Ref. [41] for a recent
review on quantum coherence).

131
[HAC , HBC ] = 0, and pure decoupled states as the initial condition [44]. The present
aim is therefore to generalise this result, taking into account more general states and
Hamiltonians with non-commuting components. One could also make it experimen-
tally friendly by considering incoherent interactions with environments.

9.2.2 The strength of non-decomposability and correlation

gain
Here we discuss a future direction that is an extension of Chapter 4. Previously
we have established, in the indirect interaction setting, that the gain of correla-
tions between the objects A and B exceeding certain bounds, would imply that
the the Hamiltonians are non-commuting (for closed systems), or more generally,
non-decomposability of the dynamical operator (for open systems).
Now let us consider the following bound on the correlation between A and B (as
quantified by a quantifier Q)

QA:B (τ ) − QA:B (0) ≤ η, (9.3)

where η is the to-be-proven bound that is independent of initial state. One can then
think of the Trotterized process, as depicted in Fig. 9.1. For a general evolution,
suppose one writes the dynamical operator as

(n) (n) (2) (2) (1) (1)

Λ = ΛBC ΛAC · · · ΛBC ΛAC ΛBC ΛAC , (9.4)

where the superscript denotes the number of sequences where C first interacts with
A and then with B. Next, one can apply the bound of Eq. (9.3) to the sequences
above and obtain

QA:B (∆t) − QA:B (0) ≤ η

QA:B (2∆t) − QA:B (∆t) ≤ η
..
.
QA:B (n∆t) − QA:B ((n − 1)∆t) ≤ η. (9.5)

The sum would give

QA:B (τ ) − QA:B (0) ≤ nη, (9.6)

where τ = n∆t. In real situation, the observation of correlation gain that is ex-
ceeding certain n, would imply that the corresponding evolution operator cannot
be modelled by n sequences of decomposable operator of the form ΛBC ΛAC . We
propose this as characterisation of the strength of non-decomposability of the actual

132
dynamical operator Λ. In particular, the higher the observed correlation QA:B (τ )
the stronger the non-decomposability in the dynamics.

Figure 9.1: Illustration of Trotterized dynamics. The evolution of the whole system
consists of sequences of interactions between the mediator object C with A, followed
by C with B, each for a time ∆t.

We now provide a simple example of the bound in Eq. (9.3). Consider the mutual
information as the correlation quantifier. The corresponding bound is given by the
following Lemma.

Lemma 9.2.1. For a tripartite system with a dynamical operator having the form
ΛBC ΛAC , the mutual information follows

IA:B (t) − IA:B (0) ≤ 4 log2 (dC ), (9.7)

where dX is the dimension of object X and we have assumed dA , dB > dC for sim-
plicity.

Proof. Let us begin with the monotonicity of mutual information under local oper-
ations to arrive at

IA:B (t) ≤ IA:BC (t) = IA:BC (ΛBC ΛAC [ρ(0)]) ≤ IA:BC (ΛAC [ρ(0)]). (9.8)

Next, we have the following arguments

0
IA:B (t) − IA:B (0) ≤ IA:BC − IA:B (0)
0 0 0
= IAC:B − IB:C + IA:C − IA:B (0)
0 0
≤ IAC:B (0) − IA:B (0) − IB:C + IA:C
0
≤ IB:C|A (0) + IA:C
≤ 4 log2 (dC ), (9.9)

133
where the prime symbol denotes the state ΛAC [ρ(0)]. We justify the steps as follows.
Eq. (9.8) has been used in the first line. The second line is apparent by utilising the
definition of mutual information between two parties, i.e., IX:Y = SX + SY − SXY ,
where for example SX is the von Neumann entropy of object X. The third line uses
the monotonicity property under the operation ΛAC that is local in the partition
AC : B. We have used the definition of conditional entropy and the positivity of
mutual information in the fourth line. The mutual information between B and C
(conditioned on A) is bounded by 2 log2 (dC ). Note that we assume dB > dC . This is
also true for mutual information between A and C, which completes the proof.

Therefore, the corresponding bound for n sequences of decomposable dynamics

reads
IA:B (t) − IA:B (0) ≤ 4n log2 (dC ). (9.10)

To demonstrate violations of the bound for certain n, one can make use of the setup
described in Section 8.3, where two cavity fields are interacting via a two-level atom.
Our simulations show that high correlations can be obtained by having pure states
with more photons as the initial condition. Future work is aimed at reducing the
bound in Eq. (9.9) as our simulations show the gain of mutual information being
bounded by only 2 log2 (dC ). Also, one might consider other correlation quantifiers.

9.2.3 Protocols witnessing the presence or absence of quan-

tum objects
Chapter 6 has brought to our attention the need of a protocol for witnessing quantum
mediators. To some extend, this has been discussed in Section 6.9. In particular,
if quantum entanglement between the principal objects A and B increases and the
state of system AB is pure at all times, then the dynamics is only possible through
direct interactions. This shows the absence of a mediating object. An experimentally
friendly scheme requires one to generalise this method to mixed states, open system
dynamics, and finite number of measurements.
Moreover, we are also interested in a protocol showing the presence of a mediating
object. Based on the study of the speed of distributing quantum entanglement in
Chapter 5, a simple way can be proposed as follows. If one starts with mixed initial
state for system AB, by considering closed dynamics, one cannot get maximum
entanglement in the partition A : B. This is because unitary operator preserves
the purity of the state and that maximum entanglement is given by a pure state.
Therefore, the observation of maximum entanglement between the principal objects,
would reveal the presence of a mediating object. This mediating object results in
the change of purity of the state of AB, making it possible to reach maximum

134
entanglement. This protocol is in general hard to implement due to the requirement
of maximum entanglement. We aim at making this scheme also experimentally
friendly. A potential direction would be to consider an entanglement bound for a
given purity. If the entanglement EA:B reaches a higher value, this would imply the
presence of a quantum mediator.

9.2.4 Complete link between initial correlations with medi-

ators and entangling time limit
In Chapter 5, we have proven that the speed limit of entanglement creation in the in-
direct interaction setting cannot be faster than the direct setting. See Theorem 5.4.2
for details. However, the saturation of the speed limit is possible and we have shown
this with exemplary dynamics. Our examples, e.g., the ones in Section 5.4.1, show
that initial correlation in the partition AB : C is important for the task. Therefore,
for completeness, we aim to show that for initial state of the form ρAB ⊗ ρC , the
entangling time for the indirect setting follows a strict bound T > Γdi .
We also performed preliminary simulations with coherent dynamics in the in-
direct interaction setting. For initial assessment, we consider three qubits A, B,
and C. We have taken random initial states (separable in A : B partition) and
interaction Hamiltonians.2 Our simulations suggest a time bound

π
T ≥ . (9.11)
2

Therefore, in this scenario, it follows that T ≥ 2Γdi for the case of three qubits. A
simple example saturating this limit is given as follows

|ψ(0)i = |110i ,
~Ω
H = √ σA+ ⊗ σC− + σA− ⊗ σC+ + σB+ ⊗ σC− + σB− ⊗ σC+ ,


(9.12)
2
+(−)
where σX is the raising (lowering) operator of object X. Indeed, maximum en-
tanglement (negativity NA:B = 0.5) is reached in T = Ωt = π/2.

9.2.5 Gravity-mediated entanglement: Full treatment

The results we presented in Chapter 6 have shown the possibility of having entan-
glement gain between two spherical masses that are interacting via gravitational
2
Note that the resource equality has been implemented, i.e., min{hM i, ∆M } = 1.

135
potential. The gravitational energy is taken as

Gm2
Hg = − (9.13)
(L + xB − xA )
Gm2 (xA − xB ) (xA − xB )2
 
' − 1+ + , (9.14)
L L L2

where we have applied the approximation (xA − xB )  L. As gravitational coupling

is weak in practically any experimental situation, the resulting entanglement is either
very small or requires a long time to accumulate. One might wonder if there is more
entanglement generated by taking the full Hamiltonian as in Eq. (9.13). This is
intuitive since higher order terms in the expansion provide coupling between the
two masses, although at the expense of strength.
In this direction, we aim to perform simulations of the dynamics considering full
gravitational energy. We note that logarithmic negativity is a valid entanglement
quantifier only for Hamiltonians that have up to quadratic terms in operators, e.g.,
Eq. (9.14). This is because these Hamiltonians would preserve Gaussianity of the
state, allowing logarithmic negativity to be a monotone quantifier of entanglement.
However, we note that the simulation with full term has to converge to the same
dynamics in the correct limit, i.e., (xA − xB )  L. For general regime, one can use
other quantifiers. As an example, we propose the use of pure state for the initial
condition. For a closed dynamics, one can then calculate the entropy of one of the
masses as a quantifier of entanglement.

136
Appendix A

Trotter expansion

A simple derivation of the Trotter formula [193] is provided in Theorem A.0.1 below.

Theorem A.0.1. For arbitrary square matrices A and B, the following holds
 A B
n
A+B
e = lim e e
n n . (A.1)
n→∞

Proof. We recall the Baker-Campbell-Hausdorff equation

1 1 1
ln eA eB = A + B + [A, B] + [A, [A, B]] + [B, [B, A]] + · · ·


(A.2)
2 12 12

and the identity limx ef (x) = elimx f (x) .

Taking the natural logarithm of (eA/n eB/n )n gives
 
 A B
 A B 1 1 1
n ln e en n=n + + 2 [A, B] + [A, [A, B]] + [B, [B, A]] + · · · ,
n n 2n 12n3 12n3
(A.3)
where we have used Eq. (A.2).
By applying the limit n → ∞ to the exponent of Eq. (A.3) one has
 n 1 1 1
= elimn→∞ (A+B+ 2n [A,B]+ 12n2 [A,[A,B]]+ 12n2 [B,[B,A]]+··· )
A B
lim en en
n→∞
= eA+B , (A.4)

where we have utilised the limit identity in the first line.

137
Appendix B

Entanglement localisation:
Prescription

As shown in Chapter 3, entanglement distribution between remote objects A and

B is possible through a classical mediator C. Here I provide an insight into this
example and prescribe a method for applying it to a more general setting. We also
note that our example has been realised recently [49].
For simplicity, let us assume that the Hamiltonian is of the form

H = HA ⊗ HC1 + HB ⊗ HC2 , (B.1)

where [HC1 , HC2 ] = 0, i.e., these components share the same eigenbasis. Now we
take a quantum–classical initial state in the partition AB : C,
X
ρ(0) = pc ρAB|c ⊗ |ci hc| , (B.2)
c

where {|ci} form the common eigenbasis of HC1 and HC2 . Let us also use, as the
eigenvalues (assumed dimensionless), E1c and E2c respectively.
The state after a short time ∆t reads
   
i∆t i∆t
ρ(∆t) = 1− H ρ(0) 1 + H
~ ~
X  i∆t c i∆t c
= pc ρAB|c − E1 HA ρAB|c − E HB ρAB|c
c
~ ~ 2
i∆t c i∆t c

+ ρAB|c E1 HA + ρAB|c E2 HB ⊗ |ci hc|
~ ~
X i∆t
[(E1c HA + E2c HB ), ρAB|c ] ⊗ |ci hc|


= pc ρAB|c −
c
~
X
= pc ρ∆t
AB|c ⊗ |ci hc| , (B.3)
c

139
where we have used ρ∆t AB|c to denote the short coherent evolution of ρAB|c with
weighted local Hamiltonians E1c HA + E2c HB . Note that we have ignored terms pro-
portional to (∆t)2 . The evolution for a time t simply requires successive applications
of the above process, i.e., ρ(t) = c pc ρtAB|c ⊗ |ci hc|.
P

We show that entanglement in the partition A : BC stays constant as follows

X X
EA:BC (t) = pc EA:B (ρtAB|c ) = pc EA:B (ρAB|c ) = EA:BC (0), (B.4)
c c

where we have used the flags condition [40] on the quantum–classical state in the
first and last equality, and the monotonicity of entanglement under local unitary
operations in the second equality.
Now let us consider entanglement between A and B. In particular we have
X
X X
EA:B (t) = EA:B pc ρtAB|c ≤ pc EA:B (ρtAB|c ) = pc EA:B (ρAB|c ), (B.5)
c c c

where we have used the convexity of entanglement for the inequality and the mono-
tonicity for the last equality. One immediately realises that, in order to have entan-
glement gain in the partition A : B with this method, there has to be entanglement
already present initially, i.e., in the state ρAB|c .
Note that the necessary requirements above are satisfied in the example given in
Chapter 3, where H = (σAx ⊗ 1 ⊗ σCx + 1 ⊗ σBx ⊗ σCx )~ω, and the initial state reads
1
2
|ψ+ i hψ+ | ⊗ |+i h+| + 12 |φ+ i hφ+ | ⊗ |−i h−|. In particular, the classical basis of sys-
tem C is the eigenbasis of the Pauli matrix σx and there is maximum entanglement
in the initial states {|ψ+ i , |φ+ i}. Furthermore, the evolution of the states for a time
τ = π/8ω follows

|ψ+ i hψ+ | → ρτAB|1 = ρmax

|φ+ i hφ+ | → ρτAB|2 = ρmax , (B.6)

where ρmax is, up to a universal phase factor, a density matrix of a pure maximally
entangled state
1
√ (|0i |y+ i + |1i |y− i) , (B.7)
2
with {|y+ i , |y− i} being the eigenbasis of the σy Pauli matrix. Therefore, one can see
that EA:B (0) = 0 and EA:B (τ ) = 1, where we have used REE as the entanglement
quantifier.

140
Bibliography

[1] A. Einstein, B. Podolsky, and N. Rosen. Can Quantum-Mechanical Descrip-

tion of Physical Reality be Considered Complete? Phys. Rev., 47:777, 1935.

[2] E. Schrödinger. Discussion of probability relations between separated systems.

Math. Proc. Camb. Philos. Soc., 31:555, 1935.

[3] A. K. Ekert. Quantum cryptography based on Bell’s theorem. Phys. Rev.

Lett., 67:661, 1991.

[4] C. H. Bennett and S. J. Wiesner. Communication via one- and two-particle

operators on Einstein-Podolsky-Rosen states. Phys. Rev. Lett., 69:2881, 1992.

[5] C. H. Bennett, G. Brassard, C. Crépeau, R. Jozsa, A. Peres, and W. K. Woot-

ters. Teleporting an unknown quantum state via dual classical and Einstein-
Podolsky-Rosen channels. Phys. Rev. Lett., 70:1895, 1993.

[6] R. Horodecki, P. Horodecki, M. Horodecki, and K. Horodecki. Quantum en-

tanglement. Rev. Mod. Phys., 81:865, 2009.

[7] T. S. Cubitt, F. Verstraete, W. Dür, and J. I. Cirac. Separable States Can Be

Used To Distribute Entanglement. Phys. Rev. Lett., 91:037902, 2003.

[8] M. Żukowski, A. Zeilinger, M. A. Horne, and A. K. Ekert. “Event-ready-

detectors” Bell experiment via entanglement swapping. Phys. Rev. Lett., 71:
4287, 1993.

[9] J.-W. Pan, D. Bouwmeester, H. Weinfurter, and A. Zeilinger. Experimental

Entanglement Swapping: Entangling Photons That Never Interacted. Phys.
Rev. Lett., 80:3891, 1998.

[10] X. Ma, S. Zotter, J. Kofler, R. Ursin, T. Jennewein, Č. Brukner, and

A. Zeilinger. Experimental delayed-choice entanglement swapping. Nat. Phys.,
8:479, 2012.

[11] L. Mišta Jr. and N. Korolkova. Distribution of continuous-variable entangle-

ment by separable Gaussian states. Phys. Rev. A, 77:050302, 2008.

141
[12] L. Mišta Jr. and N. Korolkova. Improving continuous-variable entanglement
distribution by separable states. Phys. Rev. A, 80:032310, 2009.

[13] L. Mišta Jr. Entanglement sharing with separable states. Phys. Rev. A, 87:
062326, 2013.

[14] A. Kay. Using Separable Bell-Diagonal States to Distribute Entanglement.

Phys. Rev. Lett., 109:080503, 2012.

[15] X.-D. Yang, A.-M. Wang, X.-S. Ma, F. Xu, H. You, and W.-Q. Niu. Experi-
mental Creation of Entanglement Using Separable States. Chin. Phys. Lett.,
22:279, 2005.

[16] A. Fedrizzi, M. Zuppardo, G. G. Gillett, M. A. Broome, M. P. Almeida, M. Pa-

ternostro, A. G. White, and T. Paterek. Experimental Distribution of Entan-
glement with Separable Carriers. Phys. Rev. Lett., 111:230504, 2013.

[17] C. E. Vollmer, D. Schulze, T. Eberle, V. Händchen, J. Fiurášek, and R. Schn-

abel. Experimental Entanglement Distribution by Separable States. Phys.
Rev. Lett., 111:230505, 2013.

[18] C. Peuntinger, V. Chille, L. Mišta, N. Korolkova Jr., M. Förtsch, J. Korger,

C. Marquardt, and G. Leuchs. Distributing Entanglement with Separable
States. Phys. Rev. Lett., 111:230506, 2013.

[19] A. Streltsov, H. Kampermann, and D. Bruß. Quantum Cost for Sending

Entanglement. Phys. Rev. Lett., 108:250501, 2012.

[20] T. K. Chuan, J. Maillard, K. Modi, T. Paterek, M. Paternostro, and M. Piani.

Quantum Discord Bounds the Amount of Distributed Entanglement. Phys.
Rev. Lett., 109:070501, 2012.

[21] A. Streltsov, R. Augusiak, M. Demianowicz, and M. Lewenstein. Progress

towards a unified approach to entanglement distribution. Phys. Rev. A, 92:
012335, 2015.

[22] M. Zuppardo, T. Krisnanda, T. Paterek, S. Bandyopadhyay, A. Banerjee,

P. Deb, S. Halder, K. Modi, and M. Paternostro. Excessive distribution of
quantum entanglement. Phys. Rev. A, 93:012305, 2016.

[23] M. A. Nielsen and I. L. Chuang. Quantum computation and quantum infor-

mation. Cambridge University Press, 2010.

142
[24] K. Modi, A. Brodutch, H. Cable, T. Paterek, and V. Vedral. The classical-
quantum boundary for correlations: Discord and related measures. Rev. Mod.
Phys., 84:1655, 2012.

[25] C. Weedbrook, S. Pirandola, R. Garcı́a-Patrón, N. J. Cerf, T. C. Ralph, J. H.

Shapiro, and S. Lloyd. Gaussian quantum information. Rev. Mod. Phys., 84:
621, 2012.

[26] W. K. Wootters. Statistical distance and Hilbert space. Phys. Rev. D, 23:357,
1981.

[27] A. Uhlmann. The Metric of Bures and the Geometric Phase, in Groups and
Related Topics (Springer, Netherlands, Dordrecht, 1992), pp. 267-274.

[28] C. Helstrom. Quantum detection and estimation theory. J. Stat. Phys., 1:231,
1969.

[29] V. Vedral, M. B. Plenio, M. A. Rippin, and P. L. Knight. Quantifying Entan-

glement. Phys. Rev. Lett., 78:2275, 1997.

[30] A. Peres. Separability Criterion for Density Matrices. Phys. Rev. Lett., 77:
1413, 1996.

[31] K. Życzkowski, P. Horodecki, A. Sanpera, and M. Lewenstein. Volume of the

set of separable states. Phys. Rev. A, 58:883, 1998.

[32] J. Lee and M. S. Kim. Entanglement Teleportation via Werner States. Phys.
Rev. Lett., 84:4236, 2000.

[33] J. Lee, M. S. Kim, Y. J. Park, and S. Lee. Partial teleportation of entanglement

in a noisy environment. J. Mod. Opt., 47:2151, 2000.

[34] G. Vidal and R. F. Werner. Computable measure of entanglement. Phys. Rev.

A, 65:032314, 2002.

[35] G. Adesso, A. Serafini, and F. Illuminati. Extremal entanglement and mixed-

ness in continuous variable systems. Phys. Rev. A, 70:022318, 2004.

[36] L. Henderson and V. Vedral. Classical, quantum and total correlations.

J. Phys. A, 34:6899, 2001.

[37] H. Ollivier and W. H. Zurek. Quantum Discord: A Measure of the Quantum-

ness of Correlations. Phys. Rev. Lett., 88:017901, 2001.

[38] K. Modi, T. Paterek, W. Son, V. Vedral, and M. Williamson. Unified View of

Quantum and Classical Correlations. Phys. Rev. Lett., 104:080501, 2010.

143
[39] M. Horodecki, P. Horodecki, R. Horodecki, J. Oppenheim, A. Sen, U. Sen, and
B. Synak-Radtke. Local versus nonlocal information in quantum-information
theory: Formalism and phenomena. Phys. Rev. A, 71:062307, 2005.

[40] M. Horodecki. Simplifying Monotonicity Conditions for Entanglement Mea-

sures. Open Sys. Inf. Dyn., 12:231, 2005.

[41] A. Streltsov, G. Adesso, and M. B. Plenio. Quantum coherence as a resource.

Rev. Mod. Phys., 89:041003, 2017.

[42] E. Chitambar, A. Streltsov, S. Rana, M. N. Bera, G. Adesso, and M. Lewen-

stein. Assisted Distillation of Quantum Coherence. Phys. Rev. Lett., 116:
070402, 2016.

[43] B. Groisman, S. Popescu, and A. Winter. Quantum, classical, and total

amount of correlations in a quantum state. Phys. Rev. A, 72:032317, 2005.

[44] E. K. S. Ho. Bounds on continuous entanglement gain. Thesis, Nanyang

Technological University, 2018.

[45] T. Krisnanda, M. Zuppardo, M. Paternostro, and T. Paterek. Revealing Non-

classicality of Inaccessible Objects. Phys. Rev. Lett., 119:120402, 2017.

[46] L. C. Céleri, J. Maziero, and R. M. Serra. Theoretical and experimental

aspects of quantum discord and related measures. Int. J. Quantum Inf., 9:
1837, 2011.

[47] G. Adesso, T. R. Bromley, and M. Cianciaruso. Measures and applications of

quantum correlations. J. Phys. A, 49:473001, 2016.

[48] C. H. Bennett, D. P. DiVincenzo, J. A. Smolin, and W. K. Wootters. Mixed-

state entanglement and quantum error correction. Phys. Rev. A, 54:3824,
1996.

[49] S. Pal, P. Batra, T. Krisnanda, T. Paterek, and T. S. Mahesh. Experimental

localisation of quantum entanglement through monitored classical mediator.
arXiv:1909.11030.

[50] L. Gyongyosi. The Correlation Conversion property of quantum channels.

Quantum Inf. Process., 13:467, 2014.

[51] M. Horodecki, P. Horodecki, and R. Horodecki. Separability of mixed states:

necessary and sufficient conditions. Phys. Lett. A, 223:1, 1996.

144
[52] T. A. Baart, T. Fujita, C. Reichl, W. Wegscheider, and L. M. K. Vandersypen.
Coherent spin-exchange via a quantum mediator. Nat. Nanotechnol., 12:26,
2017.

[53] S. Sahling, G. Remeny, C. Paulsen, P. Monceau, V. Saligrama, C. Marin,

A. Revcolevschi, L. P. Regnault, S. Raymond, and J. E. Lorenzo. Experimental
realization of long-distance entanglement between spins in antiferromagnetic
quantum spin chains. Nat. Phys., 11:255, 2015.

[54] W. Y. Kon, T. Krisnanda, P. Sengupta, and T. Paterek. Non-classicality of

spin structures in condensed matter: An analysis of Sr14 Cu24 O41 . Phys. Rev.
B, 100:235103, 2019.

[55] T. Krisnanda, R. Ganardi, S.-Y. Lee, J. Kim, and T. Paterek. Detecting

nondecomposability of time evolution via extreme gain of correlations. Phys.
Rev. A, 98:052321, 2018.

[56] H. P. Robertson. The Uncertainty Principle. Phys. Rev., 34:163, 1929.

[57] L. J. Landau. On the violation of Bell’s inequality in quantum theory. Phys.

Lett. A, 120:54, 1987.

[58] A. Peres. Quantum Theory: Concepts and Methods. Kluwer Academic, Dor-
drecht, Netherlands, 2002.

[59] J. Thompson, P. Kurzyński, S.-Y. Lee, A. Soeda, and D. Kaszlikowski. Recent

Advances in Contextuality Tests. Open Syst. Info. Dynam., 23:1650009, 2016.

[60] A. Rauschenbeutel, P. Bertet, S. Osnaghi, G. Nogues, M. Brune, J. M. Rai-

mond, and S. Haroche. Controlled entanglement of two field modes in a cavity
quantum electrodynamics experiment. Phys. Rev. A, 64:050301, 2001.

[61] C. Hamsen, K. N. Tolazzi, T. Wilk, and G. Rempe. Strong coupling between

photons of two light fields mediated by one atom. Nat. Phys., 14:885, 2018.

[62] A. Streltsov, G. Adesso, M. Piani, and D. Bruß. Are General Quantum Cor-
relations Monogamous? Phys. Rev. Lett., 109:050503, 2012.

[63] A. Uhlmann. Relative entropy and the Wigner-Yanase-Dyson-Lieb concavity

in an interpolation theory. Commun. Math. Phys., 54:21, 1977.

[64] M. Fannes. A continuity property of the entropy density for spin lattice sys-
tems. Commun. Math. Phys., 31:291, 1973.

145
[65] X. Yuan, S. M. Assad, J. Thompson, J. Y. Haw, V. Vedral, T. C. Ralph, P. K.
Lam, C. Weedbrook, and M. Gu. Replicating the benefits of Deutschian closed
timelike curves without breaking causality. npj Quantum Inf., 1:15007, 2015.

[66] B. M. Terhal, M. Horodecki, D. W. Leung, and D. P. DiVincenzo. The entan-

glement of purification. J. Math. Phys., 43:4286, 2002.

[67] M. Piani. Problem with geometric discord. Phys. Rev. A, 86:034101, 2012.

[68] F. M. Paula, T. R. Oliveira, and M. S. Sarandy. Geometric quantum discord

through the Schatten 1-norm. Phys. Rev. A, 87:064101, 2013.

[69] L. Mandelstam and I. Tamm. The Uncertainty Relation Between En-

ergy and Time in Non-relativistic Quantum Mechanics, in Selected papers
(Springer, Berlin, Heidelberg, 1945), pp. 115-123.

[70] G. N. Fleming. A unitarity bound on the evolution of nonstationary states.

Nuovo Cimento A, 16:232, 1973.

[71] J. Uffink. The rate of evolution of a quantum state. Am. J. Phys., 61:935,
1993.

[72] N. Margolus and L. B. Levitin. The maximum speed of dynamical evolution.

Physica D, 120:188, 1998.

[73] A. Uhlmann. An energy dispersion estimate. Phys. Lett. A, 161:329, 1992.

[74] S. Deffner and E. Lutz. Energy-time uncertainty relation for driven quantum
systems. J. Phys. A, 46:335302, 2013.

[75] Y.-J. Zhang, W. Han, Y.-J. Xia, J.-P. Cao, and H. Fan. Quantum speed limit
for arbitrary initial states. Sci. Rep., 4:4890, 2014.

[76] D. Mondal, C. Datta, and S. Sazim. Quantum coherence sets the quantum
speed limit for mixed states. Phys. Lett. A, 380:689, 2016.

[77] S. Deffner and E. Lutz. Generalized Clausius Inequality for Nonequilibrium

Quantum Processes. Phys. Rev. Lett., 105:170402, 2010.

[78] V. Giovannetti, S. Lloyd, and L. Maccone. Advances in quantum metrology.

Nat. Photonics, 5:222, 2011.

[79] S. Lloyd. Ultimate physical limits to computation. Nature, 406:1047, 2000.

[80] S. Lloyd. Computational Capacity of the Universe. Phys. Rev. Lett., 88:
237901, 2002.

146
[81] F. C. Binder, S. Vinjanampathy, K. Modi, and J. Goold. Quantacell: powerful
charging of quantum batteries. New J. Phys., 17:075015, 2015.

[82] F. C. Binder. Work, Heat, and Power of Quantum Processes. PhD The-
sis, University of Oxford, 2016.

[83] F. Campaioli et al. Enhancing the Charging Power of Quantum Batteries.

Phys. Rev. Lett., 118:150601, 2017.

[84] B. Shanahan, A. Chenu, N. Margolus, and A. Del Campo. Quantum Speed

Limits across the Quantum-to-Classical Transition. Phys. Rev. Lett., 120:
070401, 2018.

[85] M. Okuyama and M. Ohzeki. Quantum Speed Limit is Not Quantum. Phys.
Rev. Lett., 120:070402, 2018.

[86] L. B. Levitin and T. Toffoli. Fundamental Limit on the Rate of Quantum

Dynamics: The Unified Bound Is Tight. Phys. Rev. Lett., 103:160502, 2009.

[87] F. Campaioli, F. A. Pollock, F. C. Binder, and K. Modi. Tightening Quantum

Speed Limits for Almost All States. Phys. Rev. Lett., 120:060409, 2018.

[88] I. Bengtsson and K. Zyczkowski. Geometry of Quantum States: An Intro-

duction to Quantum Entanglement (Cambridge University Press, Cambridge,
England, 2008) p. 419.

[89] G. Fubini. Atti Istit. Veneto, 63:502, 1904.

[90] E. Study. Kürzeste Wege im komplexen Gebiet. Math. Ann., 60:321, 1905.

[91] L. Henderson and V. Vedral. Information, Relative Entropy of Entanglement,

and Irreversibility. Phys. Rev. Lett., 84:2263, 2000.

[92] M. A. Nielsen. The entanglement fidelity and quantum error correction.

arXiv:quant-ph/9606012, 1996.

[93] T. Krisnanda, G. Y. Tham, M. Paternostro, and T. Paterek. Observable

quantum entanglement due to gravity. npj Quantum Inf., 6:12, 2020.

[94] R. Colella, A. W. Overhauser, and S. A. Werner. Observation of Gravitation-

ally Induced Quantum Interference. Phys. Rev. Lett., 34:1472, 1975.

[95] A. Peters, K. Y. Chung, and S. Chu. Measurement of gravitational acceleration

by dropping atoms. Nature, 400:849, 1999.

147
 [96] V. V. Nesvizhevsky, H. G. Börner, A. K. Petukhov, H. Abele, S. Baeßler,
F. J. Rueß, T. Stöferle, A. Westphal, A. M. Gagarski, G. A. Petrov, and
A. V. Strelkov. Quantum states of neutrons in the Earth’s gravitational field.
Nature, 415:297, 2002.

[97] R. V. Pound and G. A. Rebka Jr. Apparent Weight of Photons. Phys. Rev.
Lett., 4:337, 1960.

[98] J. C. Hafele and R. E. Keating. Around-the-World Atomic Clocks: Predicted

Relativistic Time Gains. Science, 177:166, 1972.

[99] D. N. Page and C. D. Geilker. Indirect Evidence for Quantum Gravity. Phys.
Rev. Lett., 47:979, 1981.

[100] M. Bahrami, A. Bassi, S. McMillen, M. Paternostro, and H. Ulbricht. Is

Gravity Quantum? arXiv:1507.05733.

[101] M. Carlesso, A. Bassi, M. Paternostro, and H. Ulbricht. Testing the gravita-

tional field generated by a quantum superposition. New J. Phys., 21:093052,
2019.

[102] C. Anastopoulos and B.-L. Hu. Probing a gravitational cat state. Class.
Quantum Grav., 32:165022, 2015.

[103] M. Derakhshani, C. Anastopoulos, and B.-L. Hu. Probing a gravitational cat

state: Experimental Possibilities. J. Phys. Conf. Ser., 701:012015, 2016.

[104] M. Derakhshani. Probing Gravitational Cat States in Canonical Quantum

Theory vs Objective Collapse Theories. arXiv:1609.01711.

[105] C. Marletto and V. Vedral. Witness gravity’s quantum side in the lab. Nature,
547:156, 2017.

[106] A. Belenchia et al. Quantum superposition of massive objects and the quan-
tization of gravity. Phys. Rev. D, 98:126009, 2018.

[107] A. Balushi, W. Cong, and R. B. Mann. Optomechanical quantum Cavendish

experiment. Phys. Rev. A, 98:043811, 2018.

[108] C. Marletto and V. Vedral. Gravitationally Induced Entanglement between

Two Massive Particles is Sufficient Evidence of Quantum Effects in Gravity.
Phys. Rev. Lett., 119:240402, 2017.

[109] S. Bose, A. Mazumdar, G. W. Morley, H. Ulbricht, M. Toroš, M. Paternostro,

A. A. Geraci, P. F. Barker, M. S. Kim, and G. Milburn. Spin Entanglement
Witness for Quantum Gravity. Phys. Rev. Lett., 119:240401, 2017.

148
[110] M. Aspelmeyer, T. J. Kippenberg, and F. Marquardt. Cavity optomechanics.
Rev. Mod. Phys., 86:1391, 2014.

[111] B. Abbott et al. Observation of a kilogram-scale oscillator near its quantum

ground state. New J. Phys., 11:073032, 2009.

[112] J. D. Teufel et al. Sideband cooling of micromechanical motion to the quantum

ground state. Nature, 475:359, 2011.

[113] J. Chan, T. P. M. Alegre, A. H. Safavi-Naeini, J. T. Hill, A. Krause,

S. Gröblacher, M. Aspelmeyer, and O. Painter. Laser cooling of a nanome-
chanical oscillator into its quantum ground state. Nature, 478:89, 2011.

[114] T. A. Palomaki, J. D. Teufel, R. W. Simmonds, and K. W. Lehnert. Entangling

Mechanical Motion with Microwave Fields. Science, 342:710, 2013.

[115] R. Riedinger, A. Wallucks, I. Marinković, C. Löschnauer, M. Aspelmeyer,

S. Hong, and S. Gröblacher. Remote quantum entanglement between two
micromechanical oscillators. Nature, 556:473, 2018.

[116] I. Marinković, A. Wallucks, R. Riedinger, S. Hong, M. Aspelmeyer, and

S. Gröblacher. Optomechanical Bell Test. Phys. Rev. Lett., 121:220404, 2018.

[117] S. Qvarfort, S. Bose, and A. Serafini. Mesoscopic entanglement from central

potential interactions. arXiv:1812.09776.

[118] R. Benguria and M. Kac. Quantum Langevin Equation. Phys. Rev. Lett., 46:
1, 1981.

[119] V. Giovannetti and D. Vitali. Phase-noise measurement in a cavity with a

movable mirror undergoing quantum Brownian motion. Phys. Rev. A, 63:
023812, 2001.

[120] R. F. Werner and M. M. Wolf. Bound Entangled Gaussian States. Phys. Rev.
Lett., 86:3658, 2001.

[121] M. R. Vanner, J. Hofer, G. D. Cole, and M. Aspelmeyer. Cooling-by-

measurement and mechanical state tomography via pulsed optomechanics.
Nat. Commun., 4:2295, 2013.

[122] M. Rashid, T. Tufarelli, J. Bateman, J. Vovrosh, D. Hempston, M. S. Kim,

and H. Ulbricht. Experimental Realization of a Thermal Squeezed State of
Levitated Optomechanics. Phys. Rev. Lett., 117:273601, 2016.

149
[123] N. Kiesel, F. Blaser, U. Delić, D. Grass, R. Kaltenbaek, and M. Aspelmeyer.
Cavity cooling of an optically levitated submicron particle. Proc. Natl. Acad.
Sci. U.S.A., 110:14180, 2013.

[124] J. Vovrosh, M. Rashid, D. Hempston, J. Bateman, M. Paternostro, and H. Ul-

bricht. Parametric feedback cooling of levitated optomechanics in a parabolic
mirror trap. J. Opt. Soc. Am. B, 34:1421, 2017.

[125] N. Kiesel. Private communication.

[126] T. Emig, N. Graham, R. L. Jaffe, and M. Kardar. Casimir Forces between

Arbitrary Compact Objects. Phys. Rev. Lett., 99:170403, 2007.

[127] J. L. Garrett, D. A. T. Somers, and J. N. Munday. Measurement of the Casimir

Force between Two Spheres. Phys. Rev. Lett., 120:040401, 2018.

[128] M. Schlosshauer. Decoherence and the quantum-to-classical transition.

Springer, Berlin, 2007.

[129] H. Vahlbruch, M. Mehmet, S. Chelkowski, B. Hage, A. Franzen, N. Lastzka,

S. Goßler, K. Danzmann, and R. Schnabel. Observation of Squeezed Light
with 10-dB Quantum-Noise Reduction. Phys. Rev. Lett., 100:033602, 2008.

[130] H. Vahlbruch, M. Mehmet, K. Danzmann, and R. Schnabel. Detection of 15

dB Squeezed States of Light and their Application for the Absolute Calibration
of Photoelectric Quantum Efficiency. Phys. Rev. Lett., 117:110801, 2016.

[131] R. Schnabel. Squeezed states of light and their applications in laser interfer-
ometers. Phys. Rep., 684:1, 2017.

[132] B. P. Abbott et al. Observation of Gravitational Waves from a Binary Black

Hole Merger. Phys. Rev. Lett., 116:061102, 2016.

[133] T. Krisnanda, C. Marletto, V. Vedral, M. Paternostro, and T. Paterek. Prob-

ing quantum features of photosynthetic organisms. npj Quantum Inf., 4:60,
2018.

[134] E. Schrödinger. What is life? University Press, Cambridge, 1943.

[135] N. Bohr. Light and Life. Nature, 131:421, 1933.

[136] E. Wigner. The Probability of a Self-Reproducing Unit. Symmetries & Reflec-

tions. Indiana University Press, Bloomington, 1967.

150
[137] S. Gerlich, S. Eibenberger, M. Tomandl, S. Nimmrichter, K. Hornberger, P. J.
Fagan, J. Tüxen, M. Mayor, and M. Arndt. Quantum interference of large
organic molecules. Nat. Commun., 2:263, 2011.

[138] C. Marletto, D. M. Coles, T. Farrow, and V. Vedral. Entanglement between

living bacteria and quantized light witnessed by Rabi splitting. J. Phys. Com-
mun., 2:101001, 2018.

[139] D. Coles, L. C. Flatten, T. Sydney, E. Hounslow, S. K. Saikin, A. Aspuru-

Guzik, V. Vedral, J. K.-H. Tang, R. A. Taylor, J. M. Smith, and D. G. Lidzey.
A Nanophotonic Structure Containing Living Photosynthetic Bacteria. Small,
13:1701777, 2017.

[140] H. Amerongen, L. Valkunas, and R. Grondelle. Photosynthetic excitons. World

Scientific, Singapore, 2000.

[141] H. Lee, Y.-C. Cheng, and G. R. Fleming. Coherence Dynamics in Photosyn-

thesis: Protein Protection of Excitonic Coherence. Science, 316:1462, 2007.

[142] G. S. Engel, T. R. Calhoun, E. L. Read, T.-K. Ahn, T. Mančal, Y.-C. Cheng,

R. E. Blankenship, and G. R. Fleming. Evidence for wavelike energy transfer
through quantum coherence in photosynthetic systems. Nature, 446:782, 2007.

[143] M. Sarovar, A. Ishizaki, G. R. Fleming, and K. B. Whaley. Quantum en-

tanglement in photosynthetic light-harvesting complexes. Nat. Phys., 6:462,
2010.

[144] E. Collini, C. Y. Wong, K. E. Wilk, P. M. G. Curmi, P. Brumer, and G. D.

Scholes. Coherently wired light-harvesting in photosynthetic marine algae at
ambient temperature. Nature, 463:644, 2010.

[145] G. Panitchayangkoon, D. Hayes, K. A. Fransted, J. R. Caram, E. Harel,

J. Wen, R. E. Blankenship, and G. S. Engel. Long-lived quantum coher-
ence in photosynthetic complexes at physiological temperature. Proc. Natl.
Acad. Sci. U.S.A., 107:12766, 2010.

[146] M. M. Wilde, J. M. McCracken, and A. Mizel. Could light harvesting com-

plexes exhibit non-classical effects at room temperature? Proc. R. Soc. A,
446:1347, 2009.

[147] C.-M. Li, N. Lambert, Y.-N. Chen, G.-Y. Chen, and F. Nori. Witnessing
Quantum Coherence: from solid-state to biological systems. Sci. Rep., 2:885,
2012.

151
[148] R. Hildner, D. Brinks, J. B. Nieder, R. J. Cogdell, and N. F. Hulst. Quantum
Coherent Energy Transfer over Varying Pathways in Single Light-Harvesting
Complexes. Science, 340:1448, 2013.

[149] N. Lambert, Y.-N. Chen, Y.-C. Cheng, C.-M. Li, G.-Y. Chen, and F. Nori.
Quantum biology. Nat. Phys., 9:10, 2013.

[150] G. D. Scholes, G. R. Fleming, L. X. Chen, A. Aspuru-Guzik, A. Buchleitner,

D. F. Coker, G. S. Engel, R. Van Grondelle, A. Ishizaki, D. M. Jonas, J. S.
Lundeen, J. K. McCusker, S. Mukamel, J. P. Ogilvie, A. Olaya-Castro, M. A.
Ratner, F. C. Spano, K. B. Whaley, and X. Zhu. Using coherence to enhance
function in chemical and biophysical systems. Nature, 543:647, 2017.

[151] Y. Zhu, D. J. Gauthier, S. E. Morin, Q. Wu, H. J. Carmichael, and T. W.

Mossberg. Vacuum Rabi splitting as a feature of linear-dispersion theory:
Analysis and experimental observations. Phys. Rev. Lett., 64:2499, 1990.

[152] M. Fox. Quantum Optics-An Introduction. Oxford University Press, New

York, 2006.

[153] R. E. Blankenship, J. M. Olson, and M. Miller. Anoxygenic photosynthetic

bacteria. Springer, Dordrecht, 1995.

[154] D. M. Coles, Y. Yang, Y. Wang, R. T. Grant, R. A. Taylor, S. K. Saikin,

A. Aspuru-Guzik, D. G. Lidzey, J. K.-H. Tang, and J. M. Smith. Strong cou-
pling between chlorosomes of photosynthetic bacteria and a confined optical
cavity mode. Nat. Commun., 5:5561, 2014.

[155] T. Holstein and H. Primakoff. Field Dependence of the Intrinsic Domain

Magnetization of a Ferromagnet. Phys. Rev., 58:1098, 1940.

[156] C. Genes, D. Vitali, and P. Tombesi. Emergence of atom-light-mirror entan-

glement inside an optical cavity. Phys. Rev. A, 77:050307, 2008.

[157] C.-H. Bai, D.-Y. Wang, H.-F. Wang, A.-D. Zhu, and S. Zhang. Robust en-
tanglement between a movable mirror and atomic ensemble and entanglement
transfer in coupled optomechanical system. Sci. Rep., 6:33404, 2016.

[158] C. W. Gardiner and P. Zoller. Quantum noise. Springer Science & Business
Media, Berlin, 2004.

[159] D. F. Walls and G. J. Milburn. Quantum optics. Springer Science & Business
Media, Berlin, 2007.

152
[160] D. Bajoni. Polariton lasers. Hybrid light–matter lasers without inversion. J.
Phys. D: Appl. Phys., 45:313001, 2012.

[161] K. Ujihara. Spontaneous Emission and the Concept of Effective Area in a

Very Short Optical Cavity with Plane-Parallel Dielectric Mirrors. Jpn. J.
Appl. Phys., 30:L901, 1991.

[162] C. Houssier and K. Sauer. Circular dichroism and magnetic circular dichroism
of the chlorophyll and protochlorophyll pigments. J. Am. Chem. Soc., 92:779,
1970.

[163] M. Paternostro, D. Vitali, S. Gigan, M. S. Kim, C. Brukner, J. Eisert, and

M. Aspelmeyer. Creating and Probing Multipartite Macroscopic Entangle-
ment with Light. Phys. Rev. Lett., 99:250401, 2007.

[164] D. Vitali, S. Gigan, A. Ferreira, H. R. Böhm, P. Tombesi, A. Guerreiro, V. Ve-

dral, A. Zeilinger, and M. Aspelmeyer. Optomechanical Entanglement between
a Movable Mirror and a Cavity Field. Phys. Rev. Lett., 98:030405, 2007.

[165] M. Abdi, S. Pirandola, P. Tombesi, and D. Vitali. Entanglement Swapping

with Local Certification: Application to Remote Micromechanical Resonators.
Phys. Rev. Lett., 109:143601, 2012.

[166] M. Abdi, S. Pirandola, P. Tombesi, and D. Vitali. Continuous-variable-

entanglement swapping and its local certification: Entangling distant mechan-
ical modes. Phys. Rev. A, 89:022331, 2014.

[167] S. Gröblacher, K. Hammerer, M. R. Vanner, and M. Aspelmeyer. Observation

of strong coupling between a micromechanical resonator and an optical cavity
field. Nature, 460:724, 2009.

[168] H.-P. Breuer and F. Petruccione. The theory of open quantum systems. Oxford
University Press, 2002.

[169] E.-M. Laine, J. Piilo, and H.-P. Breuer. Witness for initial system-environment
correlations in open-system dynamics. Europhys. Lett., 92:60010, 2011.

[170] A. Smirne, H.-P. Breuer, J. Piilo, and B. Vacchini. Initial correlations in open-
systems dynamics: The Jaynes-Cummings model. Phys. Rev. A, 82:062114,
2010.

[171] J. Dajka and J. Luczka. Distance growth of quantum states due to initial
system-environment correlations. Phys. Rev. A, 82:012341, 2010.

153
[172] J. Dajka, J. Luczka, and P. Hänggi. Distance between quantum states in the
presence of initial qubit-environment correlations: A comparative study. Phys.
Rev. A, 84:032120, 2011.

[173] S. Wißmann, B. Leggio, and H.-P. Breuer. Detecting initial system-

environment correlations: Performance of various distance measures for quan-
tum states. Phys. Rev. A, 88:022108, 2013.

[174] G. Kimura, H. Ohno, and H. Hayashi. How to detect a possible correlation

from the information of a subsystem in quantum-mechanical systems. Phys.
Rev. A, 76:042123, 2007.

[175] D. Z. Rossatto, T. Werlang, L. K. Castelano, C. J. Villas-Boas, and F. F.

Fanchini. Purity as a witness for initial system-environment correlations in
open-system dynamics. Phys. Rev. A, 84:042113, 2011.

[176] A. Smirne, D. Brivio, S. Cialdi, B. Vacchini, and M. G. A. Paris. Experimen-

tal investigation of initial system-environment correlations via trace-distance
evolution. Phys. Rev. A, 84:032112, 2011.

[177] C.-F. Li, J.-S. Tang, Y.-L. Li, and G.-C. Guo. Experimentally witnessing
the initial correlation between an open quantum system and its environment.
Phys. Rev. A, 83:064102, 2011.

[178] C. A. Rodrı́guez-Rosario, K. Modi, A.-M. Kuah, A. Shaji, and E. C. G. Su-

darshan. Completely positive maps and classical correlations. J. Phys. A, 41:
205301, 2008.

[179] M. Gessner and H.-P. Breuer. Detecting Nonclassical System-Environment

Correlations by Local Operations. Phys. Rev. Lett., 107:180402, 2011.

[180] M. Gessner and H.-P. Breuer. Local witness for bipartite quantum discord.
Phys. Rev. A, 87:042107, 2013.

[181] M. Gessner, M. Ramm, T. Pruttivarasin, A. Buchleitner, H.-P. Breuer, and

H. Häffner. Local detection of quantum correlations with a single trapped ion.
Nat. Phys., 10:105, 2014.

[182] J.-S. Tang, Y.-T. Wang, G. Chen, Y. Zou, C.-F. Li, G.-C. Guo, Y. Yu, M.-F.
Li, G.-W. Zha, H.-Q. Ni, Z.-C. Niu, M. Gressner, and H.-P. Breuer. Experi-
mental detection of polarization-frequency quantum correlations in a photonic
quantum channel by local operations. Optica, 2:1014, 2015.

154
[183] S. Cialdi, A. Smirne, M. G. A. Paris, S. Olivares, and B. Vacchini. Two-step
procedure to discriminate discordant from classical correlated or factorized
states. Phys. Rev. A, 90:050301, 2014.

[184] A. Messina. A single atom-based generation of Bell states of two cavities. Eur.
Phys. J. D, 18:379, 2002.

[185] D. E. Browne and M. B. Plenio. Robust generation of entanglement between

two cavities mediated by short interactions with an atom. Phys. Rev. A, 67:
012325, 2003.

[186] I. Devetak and A. Winter. Distillation of secret key and entanglement from
quantum states. Proc. R. Soc. A, 461:207, 2005.

[187] M. Horodecki, P. Horodecki, and R. Horodecki. Limits for Entanglement

Measures. Phys. Rev. Lett., 84:2014, 2000.

[188] M. O. Scully and M. S. Zubairy. Quantum optics. Cambridge University Press,

Cambridge, UK, 1997.

[189] N. Brunner, S. Pironio, A. Acin, N. Gisin, A. A. Méthot, and V. Scarani.

Testing the dimension of hilbert spaces. Phys. Rev. Lett., 100:210503, 2008.

[190] M. Hendrych, R. Gallego, M. Mičuda, N. Brunner, A. Acı́n, and J. P. Torres.

Experimental estimation of the dimension of classical and quantum systems.
Nat. Phys., 8:588, 2012.

[191] J. Ahrens, P. Badziag, A. Cabello, and M. Bourennane. Experimental device-

independent tests of classical and quantum dimensions. Nat. Phys., 8:592,
2012.

[192] A. Ferraro, L. Aolita, D. Cavalcanti, F. M. Cucchietti, and A. Acin. Almost

all quantum states have nonclassical correlations. Phys. Rev. A, 81:052318,
2010.

[193] H. F. Trotter. On the Product of Semi-Groups of Operators. Proc. Am. Math.

Soc., 10:545, 1959.

155