Wigner-Weyl Quantum Entanglement

Miller Mateo Murillo Mejı́a

Universidad Nacional de Colombia

Facultad de ciencias, Departamento de Fı́sica
Bogotá, Colombia
2022
Wigner-Weyl Quantum Entanglement

Miller Mateo Murillo Mejı́a

Tesis presentada como requisito parcial para optar al tı́tulo de:

Magister en Ciencias: Fı́sica

Director:
Carlos Leonardo Viviescas Ramı́rez

Lı́nea de Investigación:
Teorı́a del entrelazamiento cuántico
Grupo de Investigación:
Caos y Complejidad

Universidad Nacional de Colombia

Facultad de Ciencias, Departamento de Fı́sica
Bogotá, Colombia
2022
Make your life a masterpiece; imagine no limitations
on what you can be, have or do.

Brian Tracy
I would like to thank Professor Carlos Viviescas for being not just my director, but also a partner
in the development of this thesis and the parallel learning process implied in it. Thanks for always
reminding me that we are more capable than we sometimes think that we are, and the different
spaces that you offer me to discuss elements of the present work inspired me to look forward and
advance on it.
I also want to thank to the members and partners of the Caos and Complejidad research group for
listening and criticizing my advances in the present work to always improve it. Moreover, thanks
for your attitude in the research of physics that always motivates me to push myself to be better
at it also.
In addition, I would like to thank the Grupos de Estudio Autonomo of the university because, in
all the development of the present work, I could be part of you, and you always appreciate my
different skills as a physicist.
Finally, and specially, I want to thank my family for always offering me their support and company
in the development of this thesis. Your constant advice helps me to work harder and harder to
finish the present work.
 ix

Abstract
Wigner-Weyl Quantum Entanglement
In this thesis, we describe the entanglement dynamics of a system of two particles interacting via a
Toda potential in the Wigner-Weyl phase space representation of quantum mechanics. Firstly, we
present the principal elements of the Wigner-Weyl representation of quantum mechanics, followed
by the identification and quantification of the entanglement of the two particle Toda system in the
wave representation. Next, we consider some relevant approaches to classical descriptions of the
entanglement dynamics, from which we inspire to describe the chaotic and regular dynamics of the
Toda system in the phase space via the quantum Poincaré surface of section employing stationary
Wigner functions, getting results in good agreement with the classical equivalent. Finally, we pre-
sent the analysis and interpretation of entanglement dynamics measured with two entanglement
measures in phase space. The first one is the Wigner Separability Entropy, which hasn’t been em-
ployed in a system as relevant as the Toda model, and we show that it is able to reproduce results
in agreement with another entanglement measure, the Von Neumann entropy. The second one is
the linear entropy in the Wigner-Weyl representation. This entanglement measure allowed us to
analyze and describe elements of the entanglement dynamics as periodic behaviors based on the
interpretation of the reduced Wigner function of the Toda system.
Keywords: quantum phase space, quantum entanglement, Toda model, Wigner fun-
ction and quantum Poincaré surface of section

Resumen
Entrelazamiento cuántico de Wigner-Weyl
En esta tesis, describimos la dinámica de entrelazamiento de un sistema de dos partı́culas interac-
tuando bajo un potencial tipo Toda en la representación de Wigner-Weyl de la mecánica cuántica.
Para comenzar, se presentan los elementos principales de la representación de Wigner-Weyl de la
mecánica cuántica, seguido por la identificación y cuantificación del entrelazamiento de las dos
partı́culas en el sistema Toda en la representación ondulatoria. Posteriormente, se consideran apro-
ximaciones relevantes a la descripción clásica de la dinámica de entrelazamiento, de donde nos
inspiramos para describir la dinámica caótica y regular del sistema Toda en el espacio de fase a
partir de las superficies de Poincaré cuánticas empleando funciones de Wigner estacionarias, obte-
niendo resultados comparables con los equivalentes clásicos. Finalmente, se presenta el análisis e
interpretación de la dinámica de entrelazamiento medida empleando dos medidas de entrelazamien-
to en el espacio de fase. La primera medida es la entropı́a de separabilidad de Wigner, la cual no ha
sido empleada en un sistema tan relevante como el modelo Toda, con la cual nosotros mostramos
que es capaz de reproducir resultados comparables con otra medida de entrelazamiento, la entropı́a
de Von Neumann. La segunda medida es la entropı́a lineal en la representación de Wigner-Weyl.
Esta medida de entrelazamiento nos permitió analizar y describir elementos de la dinámica de en-
trelazamiento tales como comportamientos periódicos basándonos en la interpretación de la función
de Wigner reducida del sistema Toda.
Palabras clave: espacio de fase cuántico, entrelazamiento cuántico, modelo Toda, fun-
ción de Wigner y secciones de Poincaré cuánticas
Content

Abstract IX

1. Introduction 3

2. Phase space representation 7

2.1. The Wigner function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.1.1. Wigner function properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.1.2. Dynamics of Wigner function . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
2.1.3. Wigner function examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10

3. Quantum entanglement 15
3.1. Entanglement for mixed and pure states . . . . . . . . . . . . . . . . . . . . . . . . . 15
3.2. Entanglement witness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
3.2.1. Quantum Conditional Entropy . . . . . . . . . . . . . . . . . . . . . . . . . . 17
3.3. Entanglement measures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
3.3.1. Local and Global Operations . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
3.3.2. Entropy and entanglement (entropy and correlations) . . . . . . . . . . . . . 18
3.3.3. Von Neumann entropy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
3.3.4. Linear entropy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
3.3.5. Uncertainty measures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

4. Classical dynamics of quantum entanglement 21

4.0.1. Toda system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
4.0.2. Quantum dynamics Toda system . . . . . . . . . . . . . . . . . . . . . . . . . 23
4.0.3. Classical dynamics Toda system . . . . . . . . . . . . . . . . . . . . . . . . . 24
4.0.4. Regular vs chaotic dynamics . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
4.0.5. Integrable systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
4.0.6. Poincare surface of section . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
4.0.7. Toda system Poincare surfaces . . . . . . . . . . . . . . . . . . . . . . . . . . 27
4.0.8. Shanon and Von Neumann entropies dynamics . . . . . . . . . . . . . . . . . 28

5. Entanglement dynamics Wigner-Weyl representation 39

5.1. Wigner function Toda system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
5.1.1. Reduced Wigner function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
5.2. Quantum Poincare surfaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
5.2.1. Quantum Poincare section for the Toda system . . . . . . . . . . . . . . . . . 42
Content 1

5.3. Entanglement in Wigner-Weyl representation . . . . . . . . . . . . . . . . . . . . . . 45

5.3.1. Wigner Separability Entropy (WSE) . . . . . . . . . . . . . . . . . . . . . . . 46
5.3.2. Wigner Separability Entropy for the baker’s and cat maps . . . . . . . . . . . 47
5.3.3. Separability entropy Toda system . . . . . . . . . . . . . . . . . . . . . . . . . 52
5.3.4. Linear entropy Wigner-Weyl representation . . . . . . . . . . . . . . . . . . . 54

6. Conclusions and perspectives 61

6.1. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
6.2. Perspectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62

A. Coherent states 63

B. Dispersion initial classical state 65

C. Weyl transformation non-diagonal harmonic oscillator elements 67

D. Wave function expansion in harmonic oscillator basis 69

E. Amplitude coefficients system state at time t 71

F. Elements of the energy matrix 73

G. Oscillation frequencies 77

H. Numerical aspects entropies calculations 79

I. Numerical implementation Wigner Separability Entropy chaotic maps 81

J. Phase space integral squared reduced Wigner function 83

Bibliography 84
1. Introduction

In recent years, advancements in technology have enabled scientists to study the behavior of nature
on previously unimaginable scales, including the subatomic level. Through the use of technologies
such as optical traps, lasers, and electromagnetic fields, researchers are not only able to study
these systems but also to control them [1, 2]. This has led to the development of new technologies,
particularly in the field of quantum computing, which are based on the control of physical systems
such as two-level systems with atoms and squeezed photons [3, 4]. The interest in these technologies
stems from their theoretical advantages over classical computers, which are made possible by one
property of nature: entanglement [5].
Entanglement refers to a unique correlation existing between the degrees of freedom within a system
and has no classical analog [6]. In addition, it is impossible to describe the entangled degrees of
freedom separately. In the case of photons, for instance, entanglement can arise between polarization
states of individual photons or ensembles of photons. The generation of such entanglement arises
from the interactions among different parts of the system, and the amount of entanglement in
the degrees of freedom can be different [7]. However, when the system undergoes unrestricted
interactions, the resulting entanglement may not be useful for specific applications [8]. This is
because performing an action on the system can yield different outcomes that cannot be reproduced
consistently. In contrast, to have entanglement useful for applications, it is necessary to avoid
interactions that, instead of stimulating entanglement production, reduce it, leading to the loss
of quantum characteristics, resulting in classical behaviors. This phenomenon is referred to as
decoherence [7].
It is imperative that our quantum systems maintain a sufficient level of entanglement to fulfill
their tasks. The quantification of entanglement within our system can be achieved by utilizing
entanglement measures, which are defined using tools such as density matrices or Wigner functions
[9, 10]. These measures can be experimentally determined through the application of for instance
tomographic technology [11]. Entanglement measures are quantities that are defined based on the
specific characteristics of our physical system and its corresponding quantum states. One notable
class of well-defined entanglement measures pertains to pure states. Among these widely used
measures, the Von Neumann entropy stands out as a prominent example. It can be found as follows
[9]
SV (ρr ) = −Tr[ρr ln (ρr )], (1-1)
where ρr is the reduced density matrix of our system state. Conversely, in the case of mixed states,
quantifying the amount of entanglement presents a challenging task. The challenge does not stem
from a lack of knowledge regarding measurement techniques; rather, the difficulty lies in the fact
that the choice of measurement method can yield varying measurements of entanglement for the
 4 1 Introduction

same system state [12]. To mitigate this uncertainty, maximization or minimization procedures
are employed to establish lower and upper bounds for the amount of entanglement [13]. These
procedures, which depend on the system’s size, constitute the challenging aspect of quantification.

Due to the aforementioned challenges associated with entanglement characterization, several ap-
proaches have emerged to approximate the dynamics of entanglement in a system state. These
approaches often employ semiclassical approximations and classical frameworks, such as the phase
space, in order to reduce computational requirements by avoiding the full utilization of quantum in-
formation [14, 15]. Notably, it has been discovered that the dynamics of certain classical quantities,
such as the Shannon and the Separability Entropy, closely replicate global and specific patterns of
the entanglement dynamics in particular systems [16, 17],

The previously mentioned finding offers the prospect of describing and quantifying entanglement
using classical quantities and behaviors. For instance, rather than utilizing the Von Neumann
entropy, a coarse-grained Shannon entropy [16] can be employed as follows
X
S=− pi ln (pi ), (1-2)
i

to describe the dynamics of entanglement in a quantum system, where pi represents the probability
that points (corresponding to states in phase space) originating from an initial classical distribution
reside within specific cells of the discretized phase space. The size of these cells is determined by
the chosen discretization scheme.

The relationship between quantum and classical behavior allows us not only to explore the descrip-
tion of a quantum system seeking a classical analog but also to investigate the influence of different
classical dynamics (regular, chaotic, and mixed) on the entanglement dynamics. This study can
be conducted in the phase space using established tools such as the Poincare surface of section in
classical cases [18]. In the quantum domain, we can associate the Poincare surface of section with
our quantum dynamics, but since we can not have trajectories in the phase space, we can employ
phase space representations of quantum mechanics such as Husimi and Wigner representations [19].

In the quantification of entanglement, as mentioned previously, both completely classical and com-
pletely quantum quantities can be employed. In the case of quantifying entanglement using comple-
tely quantum quantities, certain measures exhibit a strong connection to the classical realm. These
measures are defined on the Wigner function. Examples of such measures include the linear entropy
and the Wigner separability entropy [13]. While these measures have been applied to study the
behavior of dynamical maps like Baker and cat map [20, 21, 22, 17], their application to systems
with more complex Hamiltonian remains unexplored.

In line with these ideas, this work employs the Wigner-Weyl representation, a full phase space
representation of quantum mechanics, to describe the entanglement dynamics of a particle system
that has not been previously studied as far as we know using this representation - the Toda system.
This extends the application of entanglement measures defined in the phase space to new frontiers
and allows readers to explore its properties in a physical system closely connected to oscillating
systems. As the Wigner-Weyl representation is a faithful representation of quantum states, the
 5

entanglement dynamics presented in this work is purely quantum, obtained in the classical frame-
work of the phase space. Notably, as highlighted by [23], classical mechanics allows for a deeper
understanding of quantum dynamics and demonstrates the quantum-classical correspondence.
This work is organized as follows: Chapter 2 introduces the Wigner function as a phase space
representation of quantum mechanics, highlighting its key properties through simple examples such
as the harmonic oscillator and cat states. Chapter 3 explores the concept of quantum entanglement,
with particular emphasis on its quantification in the case of pure states. In Chapter 4, we introduce
the Toda model as a physical system of interest, where classical dynamics are calculated using a
symplectic propagation method. Additionally, quantum propagation is performed by solving the
Schrodinger equation and expanding quantum states in position representation. Using classical
dynamics as a foundation, we investigate the entanglement dynamics under various conditions.
Chapter 5 presents a novel analysis of the quantum dynamics of the Toda system using the quantum
Poincare surface of section defined over the Wigner function. This analysis focuses on studying the
influence of the quantum Poincare surface of sections on entanglement dynamics.
2. Phase space representation

In the realm of quantum mechanics, the state of a physical system can be described using various
representations, such as the wave function or density matrix. Additionally, different basis expan-
sions can be employed to conveniently describe and analyze the system states based on our specific
purposes or interests. This basis may include the harmonic oscillator basis, coherent states, compu-
tational basis, position basis, momentum basis, and others. This chapter presents a particular and
complete representation of quantum mechanics known as the Wigner function, which was origi-
nally introduced by E. P. Wigner in 1932. The chapter explores the fundamental properties, and
dynamics and provides examples of Wigner functions.

2.1. The Wigner function

In classical mechanics, the notion of phase space is introduced as a vector space of coordinates and
their corresponding canonical momenta. This framework enables the comprehensive characteriza-
tion of physical systems, including properties such as energy, boundaries, dynamic stability, and
integrability. In quantum mechanics, although we also have position and momentum operators, the
Heisenberg uncertainty principle prohibits the direct use of points in phase space or trajectories, as
defined in the classical case, for defining the state of a quantum system. Unlike classical mechanics,
these classical features do not provide a complete characterization of the dynamics in the quantum
realm.
In order to establish a comparison between quantum results and their classical analogs, it is ne-
cessary to extend the formalism of quantum mechanics to phase space. This involves establishing
a link between the context of a Hilbert vector space and the functions defined over phase space,
allowing us to connect quantum operators and observables with phase space coordinates. A funda-
mental step in this direction is the transition from quantum position and momentum operators to
the corresponding phase space coordinates through the following operators

Â(q, p) = δ(q − Q̂)δ(p − P̂ ), (2-1)

known as point operators [24]. The previous name is because there is an operator per phase space
point. With these operators, we introduce the Weyl quantization of a phase space function O(q, p)
as [25] Z
Ô(Q̂, P̂ ) = O(q, p)Â(q, p)dqdp. (2-2)

The previous expression is just the expansion of the operator Ô in the set of point operators, with
O(q, p) the expansion amplitudes.
 8 2 Phase space representation

The product of delta functions in Eq. (2-1) and its expression as Fourier transform, determines
the kind of ordering that we are using and the corresponding quantization rule [26]. An example
of point operators for symmetric ordering, the one used in the Wigner function and that we will
present in detail later in this work, is the following [25]
Z
1 i
Â(q, p) = e ~ [y(q−Q̂)+ξ(p−P̂ )] dξdy, (2-3)
2π~

where ~ is the reduced Plank constant, ξ and y are position and momentum variables coming from
the representation of the delta functions in the operator  as Fourier transforms.

In [25], the procedure to define the Wigner function in one dimension is shown from the Weyl
quantization as Z
1
W (q, p) = dξe−ipξ/~ hq + ξ/2|ρ|q − ξ/2i, (2-4)
2π~
where ρ is the density operator and the Weyl transform is normalized with the phase space cell
volume 2π~.

The generalization of the Wigner function to the multidimensional case is straightforward, changing
the positions and momentum by vectors and normalizing with the phase space cell volume (2π~)D ,
where D is the number of degrees of freedom of the system.

The Wigner function allows a full description of quantum mechanics. In the next section we present
some of its most prominent properties and some relevant quantities that can be obtained from it
[27].

2.1.1. Wigner function properties

All properties of states on Hilbert space are kept under the Weyl representation using Wigner
functions. Here we review those that are most relevant to the present work.

Normalization.

For a normalized state, tr(ρ) = 1, the integral of its Wigner function over all the phase space
is unity, Z
dqdpW (q, p) = 1. (2-5)

Inner product.

This property is inherited from a general one of the Weyl transform, namely, given two density
matrices ρ and ρ0 and their corresponding Wigner function W (q, p) and W 0 (q, p), the trace
of the product of density matrices is
Z
T r(ρρ ) = 2π~ dqdpW (q, p)W 0 (q, p).
0
(2-6)

Projection.
2.1 The Wigner function 9

Given the Wigner function W (q, p) representing the quantum state of a system, the integral
over the infinite strip bounded by the two parallel lines aq + bp = c1 and aq + bp = c2 is equal
to the probability that the observable aq̂ + bp̂ will have a value between c1 and c2 .

Two special and important cases are

Z
W (q, p)dp = |ψ(q)|2 (2-7)

and Z
W (q, p)dq = |ψ(p)|2 , (2-8)

which corresponds to the probability distribution of the position and momentum, respectively.

From the previous properties, the Wigner function seems to be a really good option for a quantum
phase space distribution in phase space, however, there are a couple of its properties that do not
fully allow for this interpretation. We turn to them now.

Non-arbitrarily large values.

The Wigner function can be expressed as the inner product of two wave functions, as a result,
the Wigner function is shown to be bounded in the following way [28]

|W (q, p)| ≤ 1/π~. (2-9)

It is a behavior different from the classical distributions. The previous expression means
that given the one-dimensional Wigner function corresponding to a physical pure state, its
maximum or minimum value corresponds to ±1/π~ respectively and the Wigner function can
take values between those bounds, including negative values, as we will see next.

Negative values.

The Wigner function can display negative values. An example where this fact is evident is in
the evaluation of the inner product between two orthogonal states |ψa i and |ψb i. From Eq.
(2-6)
Z
Tr(|ψa ihψa |ψb ihψb |) = 2π~ dqdpW|ψa i (q, p)W|ψb i (q, p) = 0. (2-10)

Thus, according to the previous result, the Wigner function should take negative values in
some regions of phase space.

The presence of negative values in the Wigner function does not diminish its utility or in-
terest; on the contrary, this unique characteristic contributes to some of the most intriguing
properties associated with the Wigner function. The negativity of the Wigner function is
closely linked to the incompatibility of observables, such as p and q [29], and serves as a
signature of several features, including purity, non-classical behavior (such as non-locality)
[30], and entanglement [31].
 10 2 Phase space representation

2.1.2. Dynamics of Wigner function

Central to this work is the dynamics of the Wigner function, Eq. (2-4). We can proceed to realize
it from the Schrodinger equation [32]
∂ρ
i~ = [Ĥ, ρ], (2-11)
∂t
by following different paths, each offering unique insights. One approach involves employing the
Weyl transform of Eq. (2-11), which yields the Moyal equation [33]. In this work, we will take an
alternative route by evolving the density matrix through a unitary transformation as a solution
to the Eq. (2-11), followed by applying the Weyl transform to the evolved density matrix. This
procedure will be discussed in greater detail in the subsequent chapters, shedding light on its
significance in our investigation.

2.1.3. Wigner function examples

To illustrate the previous properties of the Wigner function, we present a couple of typical examples
of continuous Wigner functions. The first one is a cat state as an example of the case of a coherent
superposition, and the second is a number state, corresponding to an eigenstate of the harmonic
oscillator.

Wigner function cat state

The cat state is the superposition of two states opposite with respect to the origin. In quantum
optics, they correspond to coherent states (see appendix A for a brief introduction) and are denoted
by |αi. The opposite state associated with this coherent state is |−αi, which is obtained by a
reflection of |αi. In such a way, a cat state can be represented by the following superposition
|ψc i = N (α)(|αi + |−αi), where N (α) is the normalization constant. For this state, its continuous
Wigner function can be calculated by replacing ρ in Eq. (2-4) by |ψc ihψc |. The result employing the
wave function of the coherent state in the position representation (see Appendix B) is the following

W (q, p) = Wα (q, p) + W−α (q, p)

1 −2|α|2 h −(−kq−i√2αi )2 −( p +i√2αr )2 √ 2 p √ 2
i (2-12)
+ e e e ~k + e−(−kq+i 2αi ) e−( ~k −i 2αr ) ,
π~
where
1 −(−kq−√2αr )2 −( p −√2αi )2
Wα (q, p) = e e ~k
π~ (2-13)
1 −(−kq+√2αr )2 −( p +√2αi )2
W−α (q, p) = e e ~k
π~
p
are the Wigner function for the coherent state |αi and |−αi respectively, k = M ω/~ corresponds
q
Mω
to the spatial frequency with M the mass and ω the time frequency of the oscillator, αr = 2~ q0 ,
i
αi = √2M p with (q0 , p0 ) the coordinates of the center of coherent state |αi. The other two terms
ω~ 0
in Eq. (2-12) correspond to interference terms giving account of the coherences of the state.
2.1 The Wigner function 11

In Fig. 2-1 we show the surface in the phase space corresponding to the Wigner function of the cat
state Eq. (2-12). The two Gaussians with the momentum 5/2 and −5/2 correspond to the Wigner
function in Eqs. (2-13). Also, the interference terms in Eq. (2-12) correspond to the oscillating
pattern between positive and negative values of the Wigner function in the middle of the Gaussian
surfaces of the coherent states.

Wigner function harmonic oscillator

At this example, we consider the harmonic oscillator eigenstate |ψi = |2i and we calculate the
Wigner function again employing Eq. (2-4) (see Appendix C for details)
1 02 02
W2 (q 0 , p0 ) = (−1)2 e−(q +p ) L2 2(q 02 + p02 ) ,
 
(2-14)
π~
where L2 represent the second Laguerre polynomial of order zero, q 0 = 0 √p .
p mω
~ q and p = mω~
Figure 2-2 shows the Wigner function of the second excited state of the harmonic oscillator, there,
the Wigner function is centered at (q, p) = (0, 0), where the Wigner function is maximum because of
the positive parity. In addition, it shows an oscillating pattern between positive and negative values,
which justifies the oscillating pattern in the q and p projections [7]. The probability distributions
for this case show to be equivalent because of the values chosen for m = ω = ~ = 1.
 12 2 Phase space representation

| (q)|2
| (p)|2

1.00
0.75
0.50
0.25
W

0.00
0.25
0.50
0.75
1.00
6
4
2
0
p 2 3
1 2
4 0
2 1
6 3 q

Figure 2-1.: Wigner function of the cat state |ψc i. The two Gaussians correspond to the Weyl
transform of the two coherent states and the interference pattern in the middle of
them, taking negative values, displays the quantum correlations of the state. The
contour curves in the W -axis show the clear position of each coherent state in the
(1) (1) (2) (2)
coordinates (q0 , p0 ) = (0, 5/2) and (q0 , p0 ) = (0, −5/2) respectively in addition
to the presence of interference patterns with low amplitude in the q-axis. The pro-
bability distribution |ψ(p)|2 shows how the interference pattern in the q-axis doesn’t
contribute to this probability unlike the probability |ψ(q)|2 which is affected by the
quantum coherences. The maximum value of the Wigner function in this case corres-
ponds to 1/π since ~ was chosen equal to one.
2.1 The Wigner function 13

| (q)|2
| (p)|2
0.4

0.3

0.2

0.1
W

0.0

0.1

0.2

0.3
4
3
2
1
0
p 1
3 4
2 2
3 0 1
2 1
4 4 3 q

Figure 2-2.: Wigner function of the second exited state of the harmonic oscillator. The parameters
of mass, oscillation frequency, and ~ were stated to one. The Wigner function achieves
its maximum at (q, p) = (0, 0) with a value of 1/π. Since the set of parameters, the
Wigner function has a symmetric shape with respect to the origin generating similar
curves for the marginal distributions |ψ(q)|2 and |ψ(p)|2 .
3. Quantum entanglement

In this chapter, one of the most important properties of the quantum realm will be presented:
quantum entanglement. Starting with its definition from the point of view of separable systems
and its relation to the correlations present in them. Later, to determine its presence in the sys-
tems, the concept of entanglement witness will be introduced. To quantify the amount and kind of
entanglement present in the systems, the concept of entanglement measurement will be developed
in the context of LOCC (Local Operations and Classical Communications). Some typical quantum
entanglement measures will be described in detail, such as the entropic measures and entanglement
measures of phase space functions. (In the majority of this chapter, the entanglement references
are done over two subsystems because most of them are defined as relations between pairs.)

3.1. Entanglement for mixed and pure states

To understand what entanglement is, we can suppose that we have a system that, at the same
time, can be described as composed of two or more subsystems. With this in mind, a system state
is entangled if it itself is not separable. In a mathematical way, this definition in the case of pure
states can be written as follows: Given a total system whose possible states are in the Hilbert space
H = H1 ⊗ H2 , where H1 and H2 are the Hilbert spaces of two subsystems of the total system. A
pure state |ψi ∈ H isn’t entangled if it can be written as a product of subsystems states, which in
the case of two of them looks like
|ψi = |φ1 i ⊗ |φ2 i, (3-1)

where |φ1 i is the state of one subsystem in H1 and |φ2 i is the state of the other subsystem in H2 .
Now, this definition in the case of mixed states looks quite different. A mixed state is separable if
it can be written as the convex sum of a set of product states

(1) (2)
X
ρ= pi ρi ⊗ ρi , (3-2)
i

P ()
where i pi = 1 and ρi are the subsystems states. To understand a bit more about the previous
definition, we can pay attention to which kind of quantity is the entanglement and with which
classical description it is related. Entanglement is a correlation that doesn’t have classical justifica-
tion, which means that if there are classical correlations between the subsystems, the entanglement
doesn’t change. On top of that, this justifies the convex sum in the Eq. (3-2).
 16 3 Quantum entanglement

3.2. Entanglement witness

After understanding the concept of entanglement, we are interested in determining when a system
is entangled. To do this, there are some quantities and criteria that can be applied to the systems
to determine if they are pure or mixed.
In the case of a pure state, one easy way to check if it is entangled or not is to get its Schmidt
decomposition, which is in essence the singular value decomposition in the context of Hilbert
spaces. Given a state |ψi in the Hilbert space H1 ⊗ H2 , exist a basis |φi i in H1 and a basis |ϕij in
H2 such that the |ψi can be written as
Xp
|ψi = λi |φi i|ϕi i, (3-3)
i

P √
with λi the amplitudes named Schmidt coefficients satisfying the condition i λi = 1.
With the previous definition, the criteria to know if a pure state is entangled can be established as
follows: If there is just one Schmidt coefficient in Eq. (3-3) different from zero, we can affirm that
the state is separable because it can be written as in Eq. (3-1) [34].

In the case of mixed states, there are also criteria to determine if a state is separable. As an example,
we describe one based on positive maps [7], whose definition will be introduced next. A map acts
over an operator in a Hilbert space and takes it to an alternative or the same starting Hilbert space.
Now remember a property of the density matrix, which is positivity, which means its eigenvalues or
the populations are positive given that they correspond to probabilities, when the maps preserve
this positivity, we call them positive maps.

Given a positive map Λ acting over the sub Hilbert space H1 . An extended map ΛE = Λ ⊗ I
can be defined to act over the states in the Hilbert space H1 ⊗ H2 . The extended map continues
acting over the Hilbert space H1 and now, acts trivially over the Hilbert space H2 through the
identity matrix I in this subspace. The extended map modifies the states in the product space, and
the resulting states are not necessarily positive [7]. When the extended map acts over a separable
density matrix Eq. (3-2), the resulting state is positive, offering a sufficient and necessary condition
for the separability of (2 × 2) or (2 × 3)-dimensional mixed states. When the map employed to
witness the separability of the system state is the partial transpose, this criterion is known as the
Peres-Horodecki criteria [35].

Some interesting things about the Peres-Horodecki criteria is that in [35] is shown that this criteria
can be considered in the phase space via the Wigner function. In the phase space, the partial
transpose operation becomes equivalent to a reflection Λ (partial time reversal) at some of the
moments of the Wigner function

Λ
W (x1 , p1 , x2 , p2 ) −−−−−−−→ W (x1 , −p1 , x2 , p2 ). (3-4)

At this representation, if the obtained Wigner function after the reflection operation in Eq. (3-4)
corresponds to a true Wigner function, meaning that it represents a physical state in the Hilbert
space, then the system state is separable.
3.3 Entanglement measures 17

3.2.1. Quantum Conditional Entropy

The quantum conditional entropy for a bipartite system with Hilbert space H = HA ⊗ HB can be
defined as
H(A|B) = H(ρAB ) − H(ρB ), (3-5)
where H is the Von Neumann entropy and ρAB represents the full state of the system, and
ρB = T rA (ρAB ) is the reduced density matrix corresponding to the state of the subsystem B.
The quantum conditional entropy is positive for separable states and is thus an entanglement
witness.

3.3. Entanglement measures

After knowing if a system state is entangled, it is relevant to know how much entanglement it has to
be employed as a resource for quantum information processing and so on. Given that a system can
have both classical and quantum correlations, where entanglement is included, the entanglement
measures should satisfy some minimal properties to avoid quantifying these other quantities too.
Some of these properties are related to the measurement process in quantum mechanics itself, so
we will introduce some general elements of it, such as the LGM+CC.

3.3.1. Local and Global Operations

If we have a system state ρ in a Hilbert product space H1 ⊗ H2 , we can perform operations
employing operators over just one of the subsystem states and act as the identity over the other
one. We have seen one example of these operations, such as the partial trace. Let’s represent one
of these operators as Âj , then the operation described previously over the first subsystem of the
P (1) (2)
mixed state i pi ρi ⊗ ρi looks like follow

pi (Âj ⊗ I)ρi (†j ⊗ I) ⊗ ρi .

(1) (2)
X
(3-6)
i

In general, these local operators satisfy the completeness property j †j  = I. When the local
P

previous operation is applied to both subsystems at the same time, employing another set of
operators B̂k for the other subsystem, we have a local general measure over our system
X
σ= Âj ⊗ B̂k . (3-7)
jk

These local operations in the lab can be interpreted as the operations or measurements that we do
over part of the system. Besides the fact that the operators that we are using are local, the states over
which they act don’t have in general the same structure, i.e., a product structure. Because of this,
local operations are done in different labs over the same system state change, and these changes have
to be communicated between the labs to continue with more local operations. The communication
described contains information about quantum states, but the channel to communicate it can be
perfectly classical (e.g., an email or a phone call). When the measurements done are as in Eq. (3-7),
 18 3 Quantum entanglement

we say that the actions over the system state can be classically correlated, and then a process as in
Eq. (3-7) is known as LGM+CC (Local General Measurement + Classical Communications) [9].
With this deeper understanding of the measurement process and the possible induction of classical
correlation, we will proceed to present some properties of the entanglement measurements denoted
as E acting over a state ρ:
E(ρ) = 0 iff ρ is separable.
E(ρ) is invariant under Local Unitary Operations.
E(ρ) cannot increase under LGM+CC i.e. E(σρ) ≤ E(ρ).

3.3.2. Entropy and entanglement (entropy and correlations)

Entropy is related to the uncertainty in measurement processes. When a system has classical and
quantum correlations between its subsystems, the uncertainty in a measurement process changes.
The entropy S evaluated over the reduced state of the system measures the uncertainty of an
observable from the perspective of an observer that has access to part of the system. This requires
measuring conditional uncertainty via conditional entropies. In addition, entropy is a good measure
of the loss of purity and predictability of a system state [36].

3.3.3. Von Neumann entropy

This entropy measure is closely related to the classical Gibbs entropy; it is an extension of quantum
mechanics. Given a density matrix ρ in a Hilbert space H, the Von Neumann entropy is defined as

SV = −T r(ρ ln ρ), (3-8)

where T r represents the trace operation and ln is the logarithm operation that over a density
matrix is calculated employing its eigenvalues. This quantity measures the amount of information
present in a system and the number of correlations between quantum systems. In such a way, this
quantity works as a good entanglement measure for pure states. In the case of mixed states, this
measure fails to distinguish between classical and quantum correlations [9]. Von Neumann entropy
is employed in different forms; one of them is interesting for us, where the reduced density matrix
ρr is employed instead of ρ. The reduced density matrix is calculated as a function of the Hilbert
space separation or the entanglement that we want to measure.

3.3.4. Linear entropy

Directly from the previous entropic measure can be defined a more computationally convenient
version of it, the linear entropy [37]. This entanglement measure is defined as the linear approxi-
mation of the Von Neumann entropy Eq. (1-1), taking the linear approximation of the logarithm
function ln ρr ≈ ρr − 1, from where the Eq. (1-1) results in

SL (ρr ) = 1 − T r(ρ2r ). (3-9)

3.3 Entanglement measures 19

From the previous expression, we notice directly its computational advantage by leaving out the
logarithm operation and, in the same way, the diagonalization procedure. Another interesting pro-
perty of this entanglement measure is that it quantifies the purity of the reduced state, meaning
that the measure is bounded between zero and one, where zero linear entropy corresponds to a pure
reduced state with zero entanglement.

3.3.5. Uncertainty measures

As was presented previously, the conditional entropy with the Von Neumann entropy for bipartite
systems can be employed as a witness of entanglement. In addition, from uncertainty relations
of incompatible measures as the Robertson uncertainty relation [38] can be employed to build
entanglement measures or entanglement upper bounds for entangled systems built over measures
that can be developed locally in the lab. These kinds of upper bounds are constantly improved to
more accurately characterize the entanglement in a system. One example of these proposals can be
found, for instance, in [36] with the upper bound proposed as
0 0
−H(A|B) ≥ qM U − H(XA |XB ) − H(ZA |ZB ), (3-10)

where H as in the witness entanglement, represents the Von Neumann entropy, qM U quantifies the
degree of incompatibility between the measures XA and ZB (the compatibility means the viability to
find a basis that allows to develop the measures of both observables without one spoiling the other),
and XA 0 (Z 0 ) represents measures on an arbitrary basis over the subsystem A (B) respectively.
B
4. Classical dynamics of quantum entanglement

In this chapter, we delve into classical approaches to studying quantum entanglement, which offer
valuable insights and invite us to explore quantum phenomena from a classical perspective. Our
focus is on understanding entanglement dynamics within the classical framework of the phase space
and recognizing its inherent advantages.

The chapter begins by introducing the Toda system, which serves as the foundation for our sub-
sequent exploration of the Wigner-Weyl formalism. We then provide a concise overview of the
description of classical systems, distinguishing between regular and chaotic dynamics. This unders-
tanding helps us grasp the toroidal structure of the manifold that encompasses classical trajectories,
as well as the significance of constructing Poincare surfaces of sections from this framework.

Next, we delve into the dynamics of entanglement and examine classical approaches to studying
this phenomenon. We carefully compare the classical and quantum dynamics of the system state,
utilizing equivalent initial states. To quantify and assess entanglement, we calculate and analyze
the Von Neumann and Shannon entropies. Moreover, we employ the classical notion of Poincare
surfaces to discern whether the dynamics exhibit regular or chaotic behavior.

By considering these classical perspectives, we gain valuable insights into entanglement dynamics
and its manifestations in the Toda system. This chapter sets the stage for our exploration of
the Weyl-Wigner formalism in subsequent chapters, offering a comprehensive understanding of
entanglement within the phase space representation of quantum mechanics.

4.0.1. Toda system

Numerous systems have garnered attention for investigating the connection between classical beha-
vior and quantum entanglement dynamics. These systems range from dynamical maps like the cat
and baker map [17] to anharmonic oscillator models [39]. Additionally, the Toda lattice, originally
introduced in solid-state physics to describe crystal behaviors, has exhibited intriguing entangle-
ment and dynamic characteristics when applied to two-particle systems [40]. These systems serve as
valuable platforms for exploring the intricate interplay between classical and quantum phenomena.

Since its introduction in 1967, the Toda model has emerged as a valuable tool for describing various
physical phenomena and system behaviors. Currently, ongoing research and development efforts are
focused on exploring its applications in the fields of hydrodynamics [41] and optical computing. In
the realm of optical quantum computing, the Toda model serves as a lattice structure for generating
solitons and facilitating information transmission. For instance, by employing two Toda lattices,
logic gates can be constructed, where the presence or absence of solitons is assigned values of 1 and
 22 4 Classical dynamics of quantum entanglement

0 respectively. These gates utilize impurities as harmonic oscillators that introduce delays in the
soliton propagation within the Toda lattice [42]. The utilization of the Toda model in these contexts
highlights its significance and ongoing research in advancing various technological applications.

The Toda lattice model, including the two-particle case, describes an interacting chain of particles
with an exponential potential. The quantum Hamiltonian for a particular configuration of the Toda
system for two particles of masses M1 and M2 can be written as

p̂21 p̂2  
Ĥ = + 2 + V e−q̂1 /d + eq̂2 /d + e−(q̂2 −q̂1 )/d − 3V I, (4-1)
2M1 2M2

where the first two terms account for the kinetic energy of each particle, and the Toda potential
is characterized by its positive amplitude V and a characteristic distance d, for which we will take
the value of 1 in all our calculations. The last term guarantees that the minimum energy in the
system is zero. As can be seen in Fig. 4-1, for low energies, the system can be well approximated
by harmonic oscillators.

0.75

0.50

0.25

0.00
p1

0.25

0.50

0.75

1.00
1.00 0.75 0.50 0.25 0.00 0.25 0.50 0.75
q1

Figure 4-1.: Contour curves of the Toda potential for the equipotential values 0.5, 0.1, 0.01, and
0.001, where the largest equipotential value corresponds to the biggest curve and it
becomes smaller when the equipotential value is decreased. When the equipotential
value is reduced, the shape of the potential curve is closer to two coupled quantum
harmonic oscillator’s potential.
 23

4.0.2. Quantum dynamics Toda system

In this section, the quantum dynamics of the Toda system, whose Hamiltonian and potential were
presented above, are developed. The time evolution of a state under the Hamiltonian of the Toda
system Eq. (4-1), can be obtained via unitary evolution since the Hamiltonian is time-independent
and the system is closed. This time evolution as a solution of the Schrodinger equation is as follows

i
|ψ(t)i = e− ~ Ĥt |ψ(0)i, (4-2)

where |ψ(0)i is the initial state for the two particles and ~ is the Planck constant. As we mentioned
before, for low energies, the Toda potential can be well approximated by a harmonic potential;
therefore, a convenient representation for the initial state of the particles is the harmonic osci-
llator basis. Then, the initial state in the harmonic oscillator or Fock basis using the identity
P
m1 m2 |m1 m2 ihm1 m2 | in the Eq. (4-2) can be represented by

X i
|ψ(t)i = hm1 m2 |e− ~ Ĥt |ψ(0)i|m1 m2 i,
m1 ,m2
X (4-3)
= Cm1 m2 (t)|m1 m2 i,
m1 ,m2

where we use the harmonic oscillator basis {|m1 m2 i}, with |m1 i representing the Fock basis for
the first particle and |m2 i the Fock basis for the second one. In the second line, Cm1 m2 (t) =
i
hm1 m2 |e− ~ Ĥt |ψ(0)i are the amplitude coefficients, which are calculated in the appendix E. At this
point, it is relevant to mention that, though the sums over the harmonic oscillator basis should
go in principle to infinity since we are going to develop numerical calculations, we should truncate
them at some point.

Initial state

The initial state that we evolve following the unitary evolution described previously is a separable
coherent state |ψ(0)i = |α1 α2 i, where |α1 i corresponds to the first particle and |α2 i to the second
one. See appendix A for a brief introduction to the coherent states. The previous state in the
harmonic oscillator basis can be written as (see appendix B)

1 1 2 1 2
hn1 n2 |α1 α2 i = e− 2 |α1 | α1n1 e− 2 |α2 | α2n2 , (4-4)
n1 !n2 !

where
r
Mi ωi i
αi = qi (0) + √ pi (0)
2~ 2Mi ωi ~
p
specifies the coherent state |αi i by its expected values in position and momentum and ωi = 2/Mi
are the oscillation frequencies of each particle (see appendix G).
 24 4 Classical dynamics of quantum entanglement

4.0.3. Classical dynamics Toda system

Now we turn to the classical dynamics of the Toda system. The classical Hamiltonian is equivalent
to the one in Eq. (4-1), but removing the operator character of the positions and moments, i.e.

p21 p2  
H= + 2 + V e−q1 /d + eq2 /d + e−(q2 −q1 )/d − 3V. (4-5)
2M1 2M2

In this case, the initial state to propagate is equivalent to the coherent state of the quantum case Eq.
(4-4) is a gaussian distribution for each classical particle, with the standard position and momentum
p p
deviation given respectively by σqi = ~/(2Mi ωi ) and σpi = ~mi ωi /2, where the index i labels
the particles (see appendix B for details).
Though this distribution is continuous, for the simulations it is necessary to discretize the classical
√
phase space. To do it in relation to the quantum case, we choose the cell length as ~. To get the
time evolution of this initial Gaussian distribution, we use Hamilton’s equations

∂H
p˙i = − (4-6)
∂qi

∂H
q˙i = , (4-7)
∂pi
where i = 1, 2 corresponds to Hamilton’s equation for particle 1 and particle 2 respectively.

These four equations use the Hamiltonian of the Toda system Eq. (4-5) become

V h −q1 /d i
ṗ1 = e − e−(q2 −q1 )/d
d
p1
q̇1 = (4-8)
M1
V h q2 /d i
ṗ2 = − e − e−(q2 −q1 )/d
d
p2
q̇2 = .
M2

We use the Symplectic Leapfrog method of second order [43] that allows iteratively to calculate
each of the previous phase space coordinates after a time step τ , preserving the symplectic area.
The second-order Leapfrog method can be interpreted as a dynamical map acting over some initial
phase space variables q (k) and p(k) . The configuration of the map that we use is the following

p(k+1) = p(k) + bk+1 τ F (q (k) )

q (k+1) = q (k) + ak+1 τ G(p(k) ),

where F corresponds to the force function from the Toda potential and G is the velocity function
of the particles that can be identified from Eqs. (4-9). ak+1 and bk+1 are the coefficients of the map
that are different depending on the order of the method. For the second order case, the coefficients
are a1 = a2 = 0.5, b1 = 0 and b2 = 1 [44].
 25

4.0.4. Regular vs chaotic dynamics

The time evolution of a system in classical mechanics is deterministic, which means that a system’s
evolution is completely predictable when we have infinite precision of the minimum information
required in the equations of motion. Notwithstanding, one can ask about the behavior of trajectories
emerging from initial conditions with small differences. As expected, as the system evolves, these
initial differences can increase with time, yet, depending on the system, two contrasting behaviors
are identified. In most of the systems, the differences between the trajectories increase exponentially
with time, while in the remaining systems, they do so only linearly. In the first case, the system is
said to possess sensibility to initial conditions and is labeled as chaotic; in the other case, the system
is regular [18]. This sensibility to initial conditions can be quantified via the system of Lyapunov
exponents [45].
Previously, we have presented the Toda system as the interest system and the classical and quantum
dynamics of it. But given that our system has more than one degree of freedom, it can follow regular
or chaotic dynamics according to the system parameters [45]. The path that we follow in this chapter
for characterizing this behavior of the system is the following: First, we discuss what an integrable
system is and its relation to regular and chaotic dynamics; later, we present the definition of the
Poincare surface of sections as a tool to analyze different characteristics of the system’s dynamics,
including its regular and chaotic behavior.

4.0.5. Integrable systems

To provide a comprehensive understanding of integrable systems and the conditions outlined by
Liouville’s theorem, it is helpful to begin with the concepts of separability and boundedness. A
system is considered separable when its dynamics can be described by Hamilton-Jacobi equations
[46] that can be transformed into ordinary differential equations. Moreover, a system is deemed
bounded when its coordinates have finite limits or exhibit periodic behavior. These preliminary
notions lay the foundation for comprehending the conditions outlined by Liouville’s theorem for an
integrable system.
In the case where the system exhibits local integrability, it is feasible to employ canonical trans-
formations to acquire system coordinates that render the Hamiltonian independent of these trans-
formed variables. Consequently, within the Hamilton-Jacobi equations, all the canonical conjugate
momenta become constants of motion [46]. This property allows for a simplified description of the
system dynamics and provides valuable insights into the behavior of the system’s phase space.
Under these conditions, the equations of motion for the system will be something like the following
Θi (t) = vi t + Θi (0), (4-9)
where Θi (t) represents a particular coordinate of the system, vi are the frequencies of the coordinates
in the circular curves, and Θi (0) is a parameter of the system related to the initial conditions. The
coordinate Θi (t) completes an integral value of 2π each time that the corresponding circular curve
is completed; because of that, these coordinates are called Angles and the conjugated moments are
called Actions Ji . The previous description is possible for integrable systems, which are defined via
the following Liouville Integrability Theorem [18].
 26 4 Classical dynamics of quantum entanglement

Liouville Integrability theorem (LI)

A system of n-degrees of freedom is integrable if it satisfies two conditions: the first one is that it has
n-independent constants of motion Ii in involution (square brackets between them are zero), and
the second one is that the ΣI submanifolds are compact (ΣI are the intersection of submanifolds
for a set of action variables). Moreover, the LI theorem states that each ΣI is homotopic to an
n-dimensional torus T n and the existence of the action-angle variables allows for the description of
the motion.

4.0.6. Poincare surface of section

A convenient representation that offers information about the behavior of physical systems is the
phase space in dynamical systems, where the positions and moments of the system are represented
in an n-dimensional space. To take advantage of the differential geometry theory in the description
of dynamical systems, a more general frame for the phase space is the symplectic manifolds, where
the different configurations of the system can also be described. A differentiable manifold is a
set defined by a set of charts such that each point in the manifold can be represented by the
chart set. The mentioned charts are functions from Rn to the manifold. Examples of manifolds are
representations of the behavior of a set of coordinates, like curves and surfaces. Moreover, when a
manifold is extended to include the velocities or tangent vectors at each point of the manifold, we
are talking about the manifold T Q or tangent bundle.

The manifold allows us to describe different characteristics of physical systems as stability points
and to distinguish the types of trajectories of the system. As a simple but illustrative example of the
visualization of the dynamics in the manifold, we present the phase portrait of the plane pendulum
of radius R shown in Fig. 4-2 in the manifold T Q. The phase portrait is shown in Fig. 4-3. Because
of the periodic boundary conditions in the coordinate θ and the unbounded possible values for the
velocity, the manifold has a cylinder shape. This phase portrait allows for the representation of all
possible physical behaviors of the pendulum (Fig. 4-2) according to the initial energy, which in
polar coordinates can be written as

1
E = p2θ − mg cos(θ), (4-10)
2

where m is the mass of the particle in the pendulum and pθ = mRθ̇ is the momentum associated
to the angular velocity. As was mentioned above, with this representation of the physical behavior
in the manifold, we can identify stability points that, in the case of the plane pendulum, are the
points in the coordinates θ1 = pθ1 = 0 and θ2 = π with pθ2 = 0. The first point corresponds to an
elliptic point because the trajectories close to it have that shape, as can be seen in Fig. 4-3. On the
other hand, the second stability point corresponds to a hyperbolic point, and this can be identified
because the trajectories approach the point but, at some distance, they move away.

When the dimension of the manifold T Q is larger than two, the description mentioned is more
difficult to realize. In these cases, the Poincare discrete maps become a useful tool. Understanding
T Q as the space where all the possible integral trajectories of the system live, the Poincare sec-
 27

R
θ
m

Figure 4-2.: Plane pendulum of radius R and mas m. Here we can see the circular trajectory that
can follow the mass m parameterized with the angle θ regarding the vertical.

tion corresponds to a submanifold of T Q, where the trace of the previous integral trajectories is
registered.

The Poincare surface of section allows us to determine if a system is chaotic or separable under some
parameters configuration. In Fig. 4-4, a Poincare surface is shown for the Toda system. Almost all
of the plot is filled with random color points, meaning that a particular initial condition generates
a time trajectory that can be at any distance from the initial point, covering a wide region of the
phase space. On the other hand, in Fig. 4-5, we can follow more easily the integral trajectory of an
initial condition in the curves drawn by each different color, and we notice that at some regions,
the trajectories have the shape of closed curves corresponding to the torus of the integrable regions
in the Poincare surface of section. The amount of points in the Fig. 4-5 is larger than in the Fig.
4-4, but besides it, the submanifold covered is smaller due to the regular behavior.

The following section will describe in detail how to obtain the Poincare surface for the Toda system.

4.0.7. Toda system Poincare surfaces

Equations (4-8) allow to calculate the state of the Toda system at any time with a given initial
state (q1 (0), p1 (0), q2 (0), p2 (0)). The trajectories of motion in this case live in a phase space of four
degrees of freedom. To get the Poincare surface of section in this case, we are going to reduce the
degrees of freedom through two constraints: one is the energy condition, and the other is that we
are going to fix a coordinate to a constant value, which typically is zero. Let’s fix the coordinate
q2 = 0. With these two conditions, our phase space has been reduced to two degrees of freedom.
 28 4 Classical dynamics of quantum entanglement

pθ pθ

θ θ

Figure 4-3.: Phase space portrait plane pendulum picture inspired by an example in [18].

From the energy condition, it is possible to obtain the conjugate momentum of the coordinate that
we fixed q2 , i.e., we can clear p2 from the following equation

E = H(q1 , p1 , q2 = 0, p2 ). (4-11)

In addition, given that the classical Hamiltonian is quadratic in momentum, we have to choose
one of the momentum signs as a convention, since each of them gives the same information of the
system dynamics.
Figures 4-4, 4-5, 4-6 and 4-7 were obtained using the method described before. Figure 4-6 shows
a particular parameter configuration in which any initial condition generates a trajectory that lives
in a torus. Otherwise, Fig. 4-7 shows a configuration, where some torus from the regular behavior
still remains, but most of the Poincare section is chaotic, which can be identified because of the
color mixture, the increase in the points density compared with the regular case, and the reduction
in the presence of torus.
Now that we have a tool to discriminate if the dynamics of the system are regular or chaotic, in
the following section we consider how this behavior affects the entanglement dynamics of the Toda
system.

4.0.8. Shanon and Von Neumann entropies dynamics

In this section, the comparison between the quantum entanglement measures, the Von Neumann
entropy Eq. (1-1) and the classical Shannon entropy Eq. (1-2), is presented. In Eq. (1-1), we have
 29

10

0
p1

10

4 2 0 2 4
q1

Figure 4-4.: Poincare surface of section for the Toda system, where the relevant parameters are
the masses and the energy. The energy value in this case is 80 with M1 = 1, M2 =
0.54 and 64 initial points. In the plot, each different color corresponds to a different
initial condition or integral trajectory. Under these parameter conditions, most of
the trajectories are chaotic and there is just a small region where some toruses still
remain.

the function logarithm acting over the reduced density matrix operator ρr . This operation, when
P
the reduced density matrix is written in its eigenbasis ρr = i λi |ψi ihψi | is defined as applying
the logarithm operation to the eigenvalues λi . Taking into account the trace operation, the Von
Neumann entropy over the reduced density matrix in the eigenbasis can be calculated as
X
h(ρr ) = − λi ln λi . (4-12)
i

To obtain the mentioned eigenvalues of the reduced density matrix, we employed the Schmidt
decomposition over the vector state of the Toda system Eq. (4-3), which as can be seen in the
appendix H is equivalent to doing the Singular Value Decomposition over the amplitude coefficients
of Eq. (4-3).
On the other hand, to calculate the Shannon entropy Eq. (1-2) for the Toda system, we measure
the point location uncertainty as it is proposed in [16], which represents the system state at some
time t in the reduced coordinates (q1 , p1 ). To do it, the trajectories of the initial classical ensembles
 30 4 Classical dynamics of quantum entanglement

0
p1

4
2 1 0 1 2
q1

Figure 4-5.: Poincare surface of section of the Toda system with the parameters M1 = 1, M2 = 0.1,
E = 7, and an initial number of points of 72. In the figure is possible to identify
the chaotic and regular behaviour in some regions, moreover, the mixed regions of
transition between the regular and chaotic dynamics. Inside some regular regions
appear torus and so on that reflect some kind of fractal pattern in the Poincare
surface.

are calculated for a time t, applying the symplectic Leapfrog integrator as described above. The
trajectory ensemble is projected in the reduced phase space as an analogous operation to the partial
trace in the quantum case. Each time that the map of the symplectic integrator is applied, we count
√
the number of points in each phase space cell i of size ~, we note this number as wi . To determine
the probabilities of the classical system state in the phase space, we divide wi over the total number
of points in the initial ensemble N . Then the Shannon entropy, as a classical approach to the Von
Neumann entropy, can be written as [16]
X wi (t) wi (t)
S(t) = − ln . (4-13)
N N
i

As an example of the points counting procedure, we present Fig. 4-8. There, we can see an example
of the point distribution at time t corresponding to the phase space state of the Toda system
projected over the reduced phase space, which for this example case is (q, p).
In Figs. 4-9 and 4-10 the results of the Shannon and Von Neumann entropies for the Toda system
 31

0
p1

4
2 1 0 1 2
q1

Figure 4-6.: Poincare surface of section for the Classical Toda system in a regular dynamics with
the parameters M1 = 1, M2 = 1 and E = 7. This Poincare surface is a clear picture
of the stability of the system under the parameters configuration. Moreover, we see
that although we know that the manifold where the trajectories live are torus, those
don’t have elliptic cross sections, because the different torus lives in a space dimension
bigger than three.

are shown. The classical dynamics are regular in Fig. 4-9 and chaotic in Fig. 4-10. Numerical
details regarding the calculations can be found in appendix H. In both cases, the classical entropy
serves as an upper bound for the system’s entanglement, and it exhibits considerable accuracy
in reproducing the quantum entanglement behavior within certain regions. However, in regions
proximate to the initial condition, the classical description deviates from the classical behavior as
the phase space cells are insufficiently populated to accurately replicate the quantum dynamics.

The initial state of the system is separable, resulting in a Von Neumann entropy value of zero.
However, as the system evolves and the components interact, correlations are generated between
them, leading to an increase in entropy. This process can be understood in terms of decoherence,
where the initially pure states of each particle in the Toda model become entangled due to their
interactions.

It is valuable to mention that although the phase space analog to the coherent quantum initial
p
state is a Gaussian distribution with the standard deviation of σq = ~/(2mω) for the position
 32 4 Classical dynamics of quantum entanglement

0
p1

4
2 1 0 1 2
q1

Figure 4-7.: Poincare surface of section for the classical Toda system in chaotic dynamics with
the parameters M1 = 1, M2 = 0.54 and E = 7. For this case, the decreasing mass
of the second particle in comparison with the regular case reduced the integrability
in several regions, as illustrated by the mixing of the point’s colors in the Poincare
section.

√
coordinate and σp = ~mω/2 for the momentum, given that the initial quantum state is a pure
state, the classical parallel is a delta distribution. Therefore, the classical description is more accu-
rate than the quantum one when the classical initial distribution is almost all inside a one-phase
space cell.

When a physical system is integrable locally, the entanglement grows logarithmically on time [47],
[48]. From Fig. 4-9, we notice that the classical curve is closer to the quantum results at larger
times compared with the chaotic case at Fig. 4-10, showing that when the global dynamics of
the Toda system are regular, the classical entanglement dynamics has a better reproduction of the
quantum entanglement dynamics. Moreover, both in the regular and chaotic cases, the amount
of entanglement increases when the value of ~ is decreased. Classically, it is because the possible
positions in the phase space are increased, increasing the uncertainty in the system state. In Fig. 4-
11 is presented the overlap between the Von Neumann entropy under regular and chaotic dynamics
for larger times. It is more clear here that the entanglement of the system achieves a saturation
value, which is different for each kind of dynamics. This saturation value is given by the maximal
 33

Figure 4-8.: Examples of point counting to calculate the coarse-grained Shannon entropy Eq.
(4-13). On the graph, we find one hundred different color points corresponding to
different initial conditions. The cells highlighted illustrate different situations to con-
sider in the point counting. In cases like the red cell, we can find one or more points
inside the cell, but as it is shown in the blue cell with the same size as the red, we can
also have cases where there isn’t any point inside it. In these cases, cell counting is
not taken into account to avoid problems with the logarithm operation in the Shan-
non entropy calculation. The green cell reflects the situation where the phase space
cell size is big enough to contain all the points at some time, which is the condition
expected at time zero for the Toda system.

uncertainty of the system state, and it can be estimated with the following general bound for the
entanglement entropy for pure states [49, 50]

S ≤ ln min{dim(H A ), dim(H B )}, (4-14)

where H A and H B represent the Hilbert space of each part of the bipartite system. In our case,
the dimension of the Hilbert space of the reduced system corresponds to the size of the basis in Eq.
(4-3), which is N = 90, therefore, the saturation value of the entropy corresponds to ln(90). In the
next chapter, we will come back to this saturation limit to include some particular modifications.
In order to gain insight into the influence of regular and chaotic dynamics on the entanglement
dynamics of the Toda system, we analyze the behavior of the Von Neumann entropy under these
conditions, as depicted in Fig. 4-12. Interestingly, we observe that the overall difference in entan-
glement dynamics between regular and chaotic cases is not substantial. However, it is noteworthy
 34 4 Classical dynamics of quantum entanglement

3
S

1
Shannon
Von Neumann =0.2
Von Neumann =0.1
0
0 10 20 30 40 50
t

Figure 4-9.: Comparison between the classical Shanon entropy and the Von Neumann entropy in
regular dynamics with parameters M1 = 1, M2 = 1 and N = 90 for different values
√
of ~, with the phase space discretization of ~ in the classical case. The initial phase
√
space positions for each particle that determine α1 and α2 were taken as p1 = p2 = E
and q1 = q2 = 0.

that the initial growth of entanglement is typically more rapid under chaotic dynamics, as reported
in previous studies [51, 16, 52]. This can be attributed to the increased dispersion of the distribu-
tion function in the chaotic regime, which introduces greater uncertainty in the determination of
particle states within the phase space.
 35

3
S

1 Shannon
Von Neumann =0.2
Von Neumann =0.1
0
0 10 20 30 40 50
t

Figure 4-10.: Comparison between the classical Shanon entropy and the Von Neumann entropy
in chaotic dynamics with parameters M1 = 1, M2 = 0.54 and N = 90 for different
√
values of ~, with the phase space discretization of ~ in the classical case. For
concentrating most of the initial distribution in a one-phase space cell, the standard
√
deviation employed for q and p was ~/4. In this case, the initial distributions were
√ √
in the phase space coordinates p1 = E, p2 = M2 E and q1 = q2 = 0.
 36 4 Classical dynamics of quantum entanglement

3.0

2.5

2.0

1.5
S

1.0

0.5
Chaotic
Regular
0.0
0 25 50 75 100 125 150 175 200
t

Figure 4-11.: Von Neumann entropy overlapping for the regular and chaotic global dynamics of
the Toda system with a size basis of N = 60 and ~ = 0.2. At both dynamics, the
entanglement increases due to the particle interaction until a saturation value is
reached. The oscillation pattern continues after saturation because of the fluctua-
tions of the finite statistics.
 37

2.5

2.0

1.5
S

1.0

0.5
Chaotic
Regular
0.0
0 10 20 30 40 50
t

Figure 4-12.: Von Neumann entropy under regular and chaotic classical dynamics: here, we notice
that the entanglement dynamics under chaotic or regular dynamics have similar
general behavior, with the starting value increasing due to the interaction and the
saturation value difference because of the available phase space. At the saturation
value, we notice the presence of some extra oscillations due to the finite statistics.
The parameters for this plot are N = 90 and ~ = 0.2.
5. Entanglement dynamics Wigner-Weyl
representation

In this chapter, we present the key findings and analysis derived from our research. Our investigation
centers around the application of entanglement in phase space via the Wigner function using the
Toda system as a primary case study. The Toda system’s compatibility with a harmonic oscillator
expansion makes it an ideal choice for implementing phase space formalism. Additionally, the Toda
system’s technological relevance [42] further enhances the significance of exploring our ideas in this
context.

Previously, entanglement dynamics studies using measures such as the Wigner Separability Entropy
(WSE) have primarily focused on dynamical maps such as the baker’s map or the coupled cats map
[20, 21, 22]. However, our research extends the application of these entanglement measures to the
Toda system, providing novel insights into the dynamics of entanglement in a system of practical
interest.

The chapter is organized as follows: The first part focuses on utilizing the Wigner-Weyl formalism
to depict the quantum dynamics of the Toda system states. Subsequently, the quantum Poincare
surface of section is introduced and computed to analyze the dynamics of the corresponding Wigner
function. Lastly, entanglement measures, including the Wigner Separability Entropy (WSE) and
the linear entropy, are introduced and applied to characterize the entanglement dynamics in the
phase space.

5.1. Wigner function Toda system

The entanglement dynamics of the two-particle Toda system has been previously investigated in
the position representation [16]. In order to establish a stronger link with the classical dynamics,
we will utilize the Wigner-Weyl representation of quantum mechanics to describe the quantum
dynamics in the phase space.

The state of the Toda system at any given time can be expressed as shown in Eq. (4-3). From it,
we can evaluate its corresponding density matrix
X
∗ 0 0
ρ(t) = Cm1 m2 (t)Cm 0 m0 (t)|m1 m2 ihm1 m2 |. (5-1)
1 2
m1 ,m2 ,m01 ,m02

Now, given that the Wigner function is defined as the Weyl transform of the density matrix, the
 40 5 Entanglement dynamics Wigner-Weyl representation

Wigner function of the Toda system with the density matrix Eq. (5-1) can be written as
1 X
∗
W (q1 , p1 , q2 , p2 ) = Cm1 m2 (t)Cm 0 m0 (t)
(2π~)2 1 2
m1 ,m2 ,m01 ,m02
Z Z
× dξ1 dξ2 e−p1 ξ1 /~ e−p2 ξ2 /~ hq1 + ξ1 /2, q2 + ξ2 /2|m1 m2 ihm01 m02 |q1 − ξ1 /2, q2 − ξ2 /2i.

For each particle, we can identify the contributions to the Wigner function of the different elements
of the basis,
X
∗
W (q1 , p1 , q2 , p2 ) = Cm1 m2 (t)Cm 0 m0 (t) WT 1 (q1 , p1 )m1 m0 WT 2 (q2 , p2 )m2 m0 , (5-2)
1 2 1 2
m1 ,m2 ,m01 ,m02

where Z
1
WT i (qi , pi )mi ,m0i = dξi e−pi ξi /~ hqi + ξi /2|mi ihm0i |qi − ξi /2i (5-3)
(2π~)
is the Weyl transform of the non-diagonal
q harmonic oscillator elements |mi ihm0i | of the i-th particle.
Using the re-scaled position qi0 = Mi ωi
and momentum p0i = √ pi the WT i are evaluated to
~ qi Mi ωi ~
,
(see appendix C)
0
(−1)min(mi ,mi )
q 1
02 02 0 0 0 0
WT i (qi0 , p0i )mi ,m0i = mi !m0i !e−q −p [2(q 02 + p02 )] 2 |mi −mi | ei(mi −mi ) arctan(p /q )
π~ max(mi , m0i )!
|m −m0 | 02
i
× Lmin(m i
0 [2(q
i ,m )
+ p02 )] (5-4)
i

i |m −m0 | 0
0 are the associate Laguerre polynomials of order |mi − mi |.
i
where Lmin(mi ,m ) i

The numerical evaluation of the Wigner function Eq. (5-2) can be simplified. Notice that each
factor depends on two indices; hence, each factor can be organized as a matrix, and we can take
advantage of the matrix operations such as the matrix product and the trace. Identifying with C
the matrix of expansion coefficients, WT 1 and WT 2 as the matrices for the phase space function
Eq. (5-4) per each particle, Eq. (5-2) simplifies to

W (q1 , p1 , q2 , p2 ) = T r[C † WT 1 (q1 , p1 ) C WT 2 (q2 , p2 )]. (5-5)

5.1.1. Reduced Wigner function

Since in this work, we are mainly interested in entanglement, we consider the reduced states of
each particle in the Toda system; their purity is a hallmark of entanglement [53]. As it turns out,
the Wigner function of reduced states, or the reduced Wigner function, as we shall also call it, is
an excellent indicator of the features of entanglement in the system.
The reduced density matrix of each particle is obtained by tracing over the degrees of freedom of
the other particle (or subsystem). Hence, taking the full density matrix, Eq. (5-1), and tracing over,
let’s say, the second system, we get the reduced density matrix for particle one,
X X
∗ 0
ρ1 = hl2 |ρ|l2 i = Cm1 m2 (t)Cm 0 m (t)|m1 ihm1 |,
2
(5-6)
1
l2 m1 ,m2 ,m01
5.2 Quantum Poincare surfaces 41

where |l2 i is the harmonic oscillator basis of the Hilbert space corresponding to the second particle.
The reduced density matrix of the second particle, ρ2 , can be evaluated in a similar fashion. At this
point, we do not give relevance to which subsystem we choose since, as is shown in the appendix
H, the entanglement determination is independent of which subsystem one chooses.

Now that we have the reduced density matrix for the system, we can calculate the corresponding
reduced Wigner function. Replacing Eq. (5-6) in Eq. (2-4), we obtain
Z
1 X
Wr (q1 , p1 ) = Cm1 m2 (t)Cm0 m2 (t) dξe−ip1 ξ/~ hq1 + ξ/2|m1 ihm01 |q1 − ξ/2i,
∗
(5-7)
2π~ 0
1
m1 ,m2 ,m1

which can be written as

X
∗
Wr (q1 , p1 ) = Cm1 m2 (t)Cm 0 m (t)WT 1 (q1 , p1 )m1 ,m0
2 1
(5-8)
1
m1 ,m2 ,m01

after using Eq. (5-4). For its numerical evaluation, we can use the convenient expression

Wr (q1 , p1 ) = T r[C (WT 1 (q1 , p1 ) C ∗ )> ]. (5-9)

At this point, we can illustrate our formalism with the initial states that we shall be using, ρ0 =
|α1 α2 ihα2 α1 |, e.g. the state product of two coherent states, one for each particle. In this case, the
reduced Wigner function of just one subsystem is given by Wα in Eq. (2-13), and examples of it
are plotted in Figs. 5-4a, 5-5a and 5-8a.

5.2. Quantum Poincare surfaces

As we showed before in Sec. (4.0.6), the Poincare surface of section is a very efficient tool to analyze
the dynamics of classical systems, allowing a clear distinction between integrable and non-integrable
behavior. Here we show how to extend this useful tool to the quantum regime.

Let us start with a brief discussion of what can be understood by the regular and chaotic dynamics
of a quantum system. In Sec. (4.0.4), we mentioned the utility of focusing on the trajectory behavior
to determine the system stability in classical mechanics. In a quantum description, however, the
idea of trajectory disappears because of the Heisenberg uncertainty principle, and the distinction
between regular and chaotic behavior needs different criteria.

In a similar fashion to the classical case, to analyze the quantum stability of a system, a good starting
point is to understand how an integrable quantum system is defined. A quantum system with N
degrees of freedom is integrable if there exist N globally defined (its mean non local) operators
Iˆm (p̂1 ...., p̂N , q̂1 , ..., q̂N ) that commute between them [54]. The relation between a classical integrable
system and the quantized version of it is evident. In Ref. [54] some examples of systems where it is
possible for classical integrable systems to find quantum operators whose Moyal brackets commute
with the Hamiltonian of the system, satisfying the integrability condition in the quantum case, are
shown. Incidentally, some of these examples correspond precisely to Toda systems.
 42 5 Entanglement dynamics Wigner-Weyl representation

Chaotic quantum systems, on the other hand, are identified by the fluctuations of their energy
spectrum. The statistical properties of a chaotic quantum system follow the same statistics as the
eigenvalues of ensembles of random matrices of the proper symmetry class [55, 56]. Clear signatures
of the implications of this statement can be illustrated with the quantum Poincare sections.
We take the eigenstates |ψE i at energy E of the quantum Hamiltonian as the starting point to
construct the quantum Poincare section. If we now calculate the Wigner function of an |ψE i, we
get a stationary distribution in the phase space. Over this stationary distribution, we can start to
search for a reduced distribution in a similar way to the classical surface of section: We first fix
one coordinate, let us say, q2 equals zero, and evaluate the conjugate momentum from the classical
energy condition, p2 = p2 (E, p1 , q1 , q2 = 0), where E is the energy of the considered eigenstate. The
quantum surface of section is then defined as [57]

Sψ (p1 , q1 ) = Wψ (p1 , q1 , p2 (E, p1 , q1 , q2 = 0), q2 = 0). (5-10)

Alternatively, and equivalent to the previous definition, instead of using the energy condition, one
can integrate the Wigner function over the conjugated momentum of the fixed coordinate [57],
Z
0
Sψ (p1 , q1 ) = dp2 Wψ (q1 , p1 , q2 = 0, p2 ). (5-11)

This is so because the Wigner function on the energy shell in the semiclassical limit is sharply
picked, from where the integration just picks up contributions very close to the energy surface.

5.2.1. Quantum Poincare section for the Toda system

To our knowledge, the quantum Poincare surface of section for the Toda system has not been
obtained before employing the Wigner function. In this section, we address this issue and use
our results to compare with the classical features presented in Sec. (4.0.7), and to analyze the
new quantum features that emerge in this image, e.g. fringes patterns with information related to
quantum correlations.
A stationary state of the Toda system, on the basis of the harmonic oscillator, is given by
X
l
|φl i = C̄m 1 m2
|m1 m2 i, (5-12)
m1 m2

where the amplitudes C̄m l = hm1 m2 |φl i are time independent and are evaluated in appendix E.
1 m2
With this wave function, we use the Weyl transform to get the following stationary Wigner function
X
Wl (q1 , p1 , q2 , p2 ) = l
C̄m C̄ ∗l0 0 WT 1 (q1 , p1 )m1 ,m01 WT 2 (q2 , p2 )m2 ,m02 .
1 ,m2 m ,m
(5-13)
1 2
m1 ,m2 m01 m02

We now proceed following Eq. 5-10 to get the Poincare surface of section of Figs. 5-1, 5-2 and 5-3.

Figure 5-1 presents the stationary Wigner function of the Toda system for an energy value of E = 3.
The left panel depicts the quantum Poincare section, while the right panel illustrates a comparison
5.2 Quantum Poincare surfaces 43

0.6
2 2
0.4
1 1
0.2

0 0.0 0
p1

p1
0.2
1 1
0.4
2 2
0.6
2 1 0 1 2 1 0 1
q1 q1
(a) Poincare surface of section of the Wigner fun- (b) Correspondence between the quantum (con-
ction. The region of the biggest amplitude lo- tour lines) and classical Poincare surface of
calizes around the fixed points [58] of the clas- sections. The quantum Poincare surface of
sical dynamics, meaning more stable regions. sections shows two characteristic behaviors,
In the bigger white region, we notice some one corresponding to torus that matches the
soft negative regions of the Wigner function classical ones and a transversal closed curve
that correspond to oscillation patterns with to the other regions that are also Lyapunov
small amplitudes compared with those in the stable and orbital stable.
more stable region.

Figure 5-1.: Quantum Poincare surface of sections for the Wigner function of the Toda system
under regular dynamics and their comparison with the classical Poincare surfaces
of sections. The state energy is E = 3, with q2 = 0 and p2 given by the energy
condition. The Wigner function shows a similar behavior to the classical Poincare
surface of sections and reproduces the presence of the torus in the phase space.

between the quantum representation (contour lines) and the classical Poincare surface of section. In
Fig. 5-1, it is evident that the Wigner function is concentrated around the more stable torus, which
is determined by the elliptic fixed points of the classical dynamics. The Wigner function exhibits
oscillations between positive and negative amplitudes. Conversely, in the other island of stability,
the amplitude of the Wigner function is small, yet it displays oscillatory behavior, as observed in
Fig. 5-1b.

Figure 5-1b reveals a correspondence between the contour lines of the Wigner function and the cur-
ves of the classical Poincare surface of section, specifically at the more stable torus. This agreement
between the quantum and classical representation indicates a consistent behavior in this region. In
contrast, the Wigner function exhibits small oscillations in the other stability region, indicating its
presence in the lower part of the plot. In classical terms, this region is characterized as Lyapunov
stable, where points in close proximity to an equilibrium point remain close, and orbital stable,
 44 5 Entanglement dynamics Wigner-Weyl representation

where the curves originating from different initial points remain close to each other.

0.75 3
3

2 0.50 2

1 0.25 1

0 0.00 0

p1
p1

1 0.25 1

2 2
0.50
3 3
0.75
4 4
3 2 1 0 1 2 3 2 1 0 1 2
q1 q1
(a) The Wigner function is localized around the (b) Notice how the contour lines coincide with
torus of major stability in the phase space. the classical structure of phase space beneath
them.

Figure 5-2.: Quantum Poincare surface of sections for the Toda system under regular dynamics
(M1 = M2 = 1) and their comparison with the classical Poincare surfaces of sections.
The eigenstate is considered to have energy close to 7 and we have taken ~ = 0.1.

In the subsequent analysis, we examine the behavior of the Wigner function and its quantum
Poincare surface of section for an eigenstate with energy E = 7, demonstrating regular dynamics in
the system. Figure 5-2 presents the quantum Poincare surface of section for this state, along with
its comparison to the classical Poincare surface of section. Similar to the eigenstate with E = 3
(refer to Fig. 5-1), the Wigner function is localized around the fixed point located at the center
of the stability island. This distinctive pattern of the Wigner function stands in contrast to the
structure observed in the quantum state at a higher energy level (E = 20) under chaotic dynamics,
with the particle masses set M1 = 0.6 and M2 = 1, as depicted in Fig. 5-3. In this case, the Wigner
function exhibits a uniform distribution across the phase space, as evident in Fig. 5-3a. Notably,
Fig. 5-3b highlights that the Wigner function predominantly overlaps with the chaotic region of
the classical Poincare surface o section, featuring minimal oscillations in the regions occupied with
regular islands.

Comparing the Wigner functions of all the three states considered, we observe that their contour
lines allow us to clearly identify the regular and chaotic regions of the underlying classical dynamics.
5.3 Entanglement in Wigner-Weyl representation 45

4 0.4 4

2 0.2 2

0 0.0 0
p1

p1
2 0.2 2

4 0.4 4

2 0 2 2 0 2
q1 q1
(a) Wigner function for a parameters configura- (b) Overlap between the classical Poincare sec-
tion of the Toda system with irregular beha- tions and the contour lines of the stationary
vior in almost all regions of the phase space. Wigner function. The chaotic sea provides
The Wigner functions spread uniformly over support for the Wigner function.
almost the whole accessible phase space.

Figure 5-3.: Quantum Poincare surface of sections for chaotic dynamics (M1 = 0.6 and M2 = 1)
and its comparison with the corresponding classical Poincare surfaces of section. The
energy of the state was fixed at 20 with ~ = 0.2.

5.3. Entanglement in Wigner-Weyl representation

So far, the Wigner-Weyl representation of quantum states has allowed us to describe quantum
results from classical characteristics, such as regular and chaotic dynamics, without the full proper-
ties of a probability density. In fact, the negative regions of the Wigner function for non-gaussian
states have shown interesting behaviors, such as in the Toda system, where these regions lay over
the more stable islands in the torus (see Sec. (5.2.1)). Another interesting characteristic associated
with this representation is the correlations that do not have a classical explanation. A system that
has negative regions in its Wigner function can present quantum correlations that can be of the
kind of entanglement, non-locality, and quantum steering [59].

We will now embark on the exploration of entanglement dynamics in a quantum system. As we

will demonstrate, these dynamics can be comprehensively characterized within the phase space
framework, encompassing the system’s initial state and its evolution over time. To quantitatively
assess entanglement throughout the state dynamics, it becomes necessary to introduce appropriate
entanglement measures. Recently, a phase space-based entanglement measure known as the Wigner
Separability Entropy (WSE) was proposed [20]. As mentioned in the introductory section of this
work, the WSE bears close relevance to classical density separability entropy and Von Neumann
 46 5 Entanglement dynamics Wigner-Weyl representation

entropy.
In the upcoming sections, we will proceed by defining the Wigner Separability Entropy based on the
concept of partitioning the phase space of a system. Subsequently, we will examine its application
through the analysis of two dynamical maps. This will serve as a preliminary investigation before
delving into the entanglement dynamics within the Toda system. We will compare our findings
with previous results obtained using the Von Neumann entropy. Additionally, we will utilize the
linear entropy as an alternative measure to assess entanglement in the phase space of Toda system
states. We will compare the results obtained using the linear entropy with those obtained using the
Wigner Separability Entropy, emphasizing the significance of the reduced Wigner function in this
measurement and elucidating its relationship with entanglement dynamics.

5.3.1. Wigner Separability Entropy (WSE)

We notice that the Wigner function that describes a quantum state is defined over a quantum
phase space Ω. This 2D-dimensional phase space can be decomposed arbitrarily into two subsets,
Ω = Ω1 ⊕ Ω2 , where Ωi is a subset of coordinates. With x representing the set of coordinates in Ω1
and y of Ω2 , we can decompose the Wigner function as follows:
X
W (x, y) = µ0ij u0i (x)vj0 (y), (5-14)
i,j

where {u0i (x)} and {vj0 (y)} are basis of square integrable functions of Ω1 and Ω2 respectively, and
µ0ij is a matrix with the amplitude coefficients. On the left-hand side of Eq. (5-14), we have the
vectorial phase space, where the Wigner function is an application of R2 −→ R. After the splitting
of the phase space on the right-hand side, we have applications on the subsets of the phase space
(the basis). Clearly, Eq. (5-14) suggests following the Schmidt decomposition procedure to find a
unique function decomposition given the selected phase space separation. Using the singular value
decomposition of the matrix µ0ij , the Wigner function can be written on a diagonal basis
X
W (x, y) = µn un (x)vn (y), (5-15)
n

where the µi are the Schmidt coefficients of the Wigner function. At this point, we notice that the
invariant integral in the Moyal equation, W 2 (x, y)dxdy, employing the expression in Eq. (5-15),
R

is equivalent to the sum of the square of the Schmidt coefficients, n µ2n .

P

We may now define the WSE of the Wigner function, h(W ), as [20]
X
h(W ) = − µ̃2n ln µ̃2n (5-16)
n

µn
with µ̃n = √R the normalized Schmidt coefficients.
dxdyW 2 (x,y)

The most relevant characteristics of the WSE are:

The WSE does not have problems with the negative regions of the Wigner function because
the Schmidt coefficients are positive.
5.3 Entanglement in Wigner-Weyl representation 47

For pure states, the WSE is two times the Von Neumann entropy [20], h(W ) = 2SV (ρr ).

Given that we can define the Wigner function of a mixed state and realize the Schmidt decom-
position of it with a selected phase space separation, as opposed to the Schmidt decomposition
in the state space, the WSE can be applied to quantum mixed states.

Like the operator separability entropy, the WSE is invariant under change of coordinates [17].

The WSE allows to exploration of the separability of the Wigner function for different kinds
of separations of the phase space. One example of a possible separation is between positions
and moments.

As we will demonstrate in the following, evaluating the WSE of a quantum state expanded
on a harmonic oscillator basis is quite simple.

5.3.2. Wigner Separability Entropy for the baker’s and cat maps
Some common systems used to investigate the characteristics of chaotic dynamics are discrete maps.
These dynamical systems evolve in discrete time steps, offering valuable insights into the behavior
of chaotic systems. In this context, we introduce the quantum counterparts of two well-known
discrete maps, namely the baker’s map and the cat map. We provide a brief overview of their
classical versions and proceed to evaluate the time evolution of the Wigner Separability Entropy
within these quantum representations.

The first discrete map we consider is the baker’s map [45]. In it, the phase space is a rectangle with
length Q in the position direction and length P in the momentum direction, and its area, QP , is
preserved during the time evolution. In the classical case, each time step of the evolution of the
system is determined by Eq. (5-17). Given a phase space point (qt , pt ) at some time t, the baker’s
map doubles the position coordinate of it and takes out half of the momentum, then rearranges the
result to recover the rectangular phase space of size QP via the module operation [20],
(
(2qt , 12 pt ) if 0 ≤ qt ≤ 1/2
(qt+1 , pt+1 ) = , (5-17)
(2qt − 1, 21 pt + 21 ) if 12 < qt < 1.

where we have taken Q = P = 1.

To introduce the quantum version of a classical map, we need to understand the quantum corres-
pondence for a discrete map. A quantum map is a suitable unitary transformation describing the
evolution of a quantum state during one discrete time step [60]. Hence, we assign B̂ to the quantum
baker’s map operator and ψ(n) to the quantum state, where n represent discrete positions that
come from the discretization q = n/N , with N the number of divisions of the position space. The
inverse of the dimension N of the Hilbert space fixes the value of the effective Plank constant
1
~eff = 2πN . A quantized version of the baker’s map in the position representation can be defined
as [20]
N
X −1
ψ(n0 ) = Bn0 n ψ(n), (5-18)
n=0
 48 5 Entanglement dynamics Wigner-Weyl representation

where

!
−1 G(N/2) 0
B = G(N ) (5-19)
0 G(N/2)

with

1
G(N )lj = √ e−2πijl/N (5-20)
N

the discrete Fourier transformation on N sites.

The unitary transformation in Eq. (5-18) of the quantum state ψ(n) imitates stepwise the classical
deformation described by Eq. (5-17) [60]. To picture the system dynamics in phase space, we employ
the discrete Wigner function defined in [61]. Details about the numerical implementation of this
particular discrete Wigner function can be found in [62].

Figure 5-4 shows the dynamics induced by the quantum baker’s map over an initial coherent state
√
center at (q, p) = (0.5, 0.5) with σ = ~. The Wigner function of the initial state corresponds to
the Gaussian shown in Fig. 5-4a. The quantum map starts the evolution by moving the position of
the Gaussian from the center to the edge of the phase space and cutting it according to Eq. (5-17).
5.3 Entanglement in Wigner-Weyl representation 49

1.0 1.0 1.0

0.8 0.8 0.8
0.6 0.6 0.6
p

p
0.4 0.4 0.4
0.2 0.2 0.2
0.00.00 0.25 0.50 0.75 1.00 0.00.00 0.25 0.50 0.75 1.00 0.00.00 0.25 0.50 0.75 1.00
q q q
(a) step 0 (b) step 1 (c) step 2
1.0 1.0 1.0
0.8 0.8 0.8
0.6 0.6 0.6
p

p
0.4 0.4 0.4
0.2 0.2 0.2
0.00.00 0.25 0.50 0.75 1.00 0.00.00 0.25 0.50 0.75 1.00 0.00.00 0.25 0.50 0.75 1.00
q q q
(d) step 3 (e) step 4 (f) step 5
1.0 1.0 1.0
0.8 0.8 0.8
0.6 0.6 0.6
p

0.4 0.4 0.4

0.2 0.2 0.2
0.00.00 0.25 0.50 0.75 1.00 0.00.00 0.25 0.50 0.75 1.00 0.00.00 0.25 0.50 0.75 1.00
q q q
(g) step 6 (h) step 7 (i) step 8
1.0 1.0 1.0
0.8 0.8 0.8
0.6 0.6 0.6
p

0.4 0.4 0.4

0.2 0.2 0.2
0.00.00 0.25 0.50 0.75 1.00 0.00.00 0.25 0.50 0.75 1.00 0.00.00 0.25 0.50 0.75 1.00
q q q
(j) step 10 (k) step 30 (l) step 60

Figure 5-4.: Time evolution of the Wigner function of an initial coherent state in the quantum
baker’s map. At each time step, we notice the map stretches the Wigner function
in the q coordinate and compresses it in the p coordinate, in an equivalent way to
the classical baker’s map. As expected for chaotic systems, for long times the Wigner
function uniformly distributes over the whole phase space.
 50 5 Entanglement dynamics Wigner-Weyl representation

The Wigner function of the quantum system exhibits distinct characteristics that reflect both
its classical behavior and quantum nature. The large positive amplitudes of the Wigner function
align with the expected classical behavior dictated by the map. However, the quantum aspect
becomes evident through the presence of interference patterns observed between the deformed
and cut initial Gaussian states. As the map is unitary, the normalization of the Wigner function is
maintained at each time step. However, the maximum (minimum) amplitude of the Wigner function
gradually decreases (increases) to compensate for the spreading of the initial distribution across
the available phase space. This gradual redistribution ultimately leads to a uniform distribution of
the Wigner function throughout the entire phase space, as illustrated in Fig. 5-4l. This behavior
is a characteristic feature of chaotic maps and confirms the expected behavior of the system.
For the cat map, we proceed in a similar fashion. As with the baker’s map, the cat map deforms
the initial rectangular phase space, but this time, following a matrix of scale factors M, it recovers
the rectangular area QP with the module operation. A particular cat map version with Q = P = 1
is [20] ! !
qt+1 qt
=M , (5-21)
pt+1 pt + (qt )
with !
2 1 K
M= and (qt ) = − sin (2πqt ). (5-22)
3 2 2π
Here  is a perturbation and K is the perturbation strength parameter.
A quantum version of the cat map Eq. (5-21) can be established with the N × N unitary matrix
M [20],  
πi 2 2
Mlj = A exp (M11 l − 2lj + M22 j ) + F (l) (5-23)
N M12
with  1/2
1 ikN
A= and F (l) = cos(2πl/N ). (5-24)
iN M12 2π
The exponential with the function F (l) in Eq. (5-23) corresponds to the nonperiodic part of the
propagator [58] associated with the classical perturbation  in the classical cat map Eq. (5-22).
Figure 5-5 shows the cat map time evolution of an initial coherent state. The Wigner function of
the initial state is a Gaussian (see Fig. 5-5a). The cat map stretches and folds the Wigner function
until it is spread over the whole phase space, as is depicted in Fig. 5-5i.
To calculate the WSE for states in both the quantum baker’s and quantum cat maps, we proceed
from Eq. (5-14) choosing the phase space separation for the Wigner function. As in these maps,
the phase space corresponds to the direct product of q and p, there is a unique separation of the
phase space: we define the first basis as the position basis and the second one as the momentum
basis: x = (q) and y = (p). For a discrete Wigner function, the WSE is obtained by diagonalizing
the state, noticing that its eigenvalues correspond to its Schmidt coefficients (see appendix I). The
time evolution of the WSE for the quantum baker’s map and the quantum cat map are shown in
Fig. 5-6a and Fig.5-6b respectively. There, we notice some similar behavior in the dynamics of the
WSE for the two chaotic maps, which we list next:
5.3 Entanglement in Wigner-Weyl representation 51

1.0 1.0 1.0

0.8 0.8 0.8
0.6 0.6 0.6
p

p
0.4 0.4 0.4
0.2 0.2 0.2
0.00.00 0.25 0.50 0.75 1.00 0.00.00 0.25 0.50 0.75 1.00 0.00.00 0.25 0.50 0.75 1.00
q q q
(a) step 0 (b) step 1 (c) step 2
1.0 1.0 1.0
0.8 0.8 0.8
0.6 0.6 0.6
p

p
0.4 0.4 0.4
0.2 0.2 0.2
0.00.00 0.25 0.50 0.75 1.00 0.00.00 0.25 0.50 0.75 1.00 0.00.00 0.25 0.50 0.75 1.00
q q q
(d) step 3 (e) step 4 (f) step 5
1.0 1.0 1.0
0.8 0.8 0.8
0.6 0.6 0.6
p

0.4 0.4 0.4

0.2 0.2 0.2
0.00.00 0.25 0.50 0.75 1.00 0.00.00 0.25 0.50 0.75 1.00 0.00.00 0.25 0.50 0.75 1.00
q q q
(g) step 6 (h) step 7 (i) step 8

Figure 5-5.: Time evolution of the Wigner function of an initial coherent state in a cat map. From
step 2 onwards, we notice that the classical deformation of the initial Gaussian state
is in the regions with a big positive Wigner function, while the negative regions with
small amplitudes are due to quantum interference. Quite quickly, the Wigner function
covers the whole phase space, as expected for a chaotic map.

The WSE at time zero is close to zero. This result is expected because the Wigner function
of the initial state |ψ(0)i = |α1 α2 i is factorized for the given phase space separation since its
Wigner function is simply the product Wα1 (q1 , p1 ) · Wα2 (q2 , p2 ), where Wα (q, p) is the Wigner
function of a coherent state presented in Eq. (2-13). In the appendix I is shown as an example
of the WSE calculation.
 52 5 Entanglement dynamics Wigner-Weyl representation

In Fig. 5-6 the WSE for both maps increases from zero when the time evolves, which according
to Figs. 5-4 and 5-5 is due to the interference between the Wigner function patterns produced
in each time map step, as we can notice from the time step 1 Fig. 5-4b in the case of the
baker’s map and in Fig. 5-5c in the case of the cat map.
The WSE for the chaotic maps shows an asymptotic saturation value (see dotted black line in
Fig. 5-6), which has been calculated theoretically in the context of Random Matrix Theory
[49] and corresponds to ln (0.6N ).
It is relevant to mention that in this case, because of the separation of the phase space, we are
not measuring any kind of entanglement since we don’t have more than one degree of freedom;
instead, we are measuring the separability of the reduced Wigner function for a given phase space
separation.

6 6
5 5
4 4
WSE

WSE

3 3
2 2
1 1
0 0
0 10 20 30 40 50 60 0 2 4 6 8 10
Time (map steps) Time (map steps)
(a) WSE dynamics of an initial coherent state (b) WSE dynamics of an initial coherent state
evolving following the dynamics generated by following the dynamics generated by a quan-
a quantum baker’s map. The increasing WSE tum cat map. The WSE achieves its satu-
means that the Wigner function becomes mo- ration value at time step 2, showing almost
re disorganized when the time steps increase. constant behavior from this point.

Figure 5-6.: WSE for the quantum dynamical systems, the baker’s and cat maps. The number of
divisions of the phase space per each coordinate is of N = 29 . The saturation value
represented with the doted black straight line in the time interval of measurement is
achieved faster for the quantum cat map than for the quantum baker’s, which means
a more disorganized effect of the phase space in the cat map at initial stages, as can
be seen comparing Figs. 5-4 and 5-5.

5.3.3. Separability entropy Toda system

As was mentioned before, the phase space separation in subsets Ω1 and Ω2 is arbitrary. Here, we
want to analyze the entanglement between the two particles of the Toda system in the phase space,
because of that, we choose x = (q1 , p1 ) and y = (q2 , p2 ) as the subsets of coordinates to decompose
the phase space. With this phase space decomposition, the function in Eq. (5-4) can be chosen as
5.3 Entanglement in Wigner-Weyl representation 53

the L2 basis for each particle. Using the previous basis selection, the Eq. (5-2) can be rewritten in
the following way:
X
W (q1 , p1 , q2 , p2 ) = DJK (t)WT 1 (q1 , p1 )J WT 2 (q2 , p2 )K , (5-25)
J,K

where DJK (t) = Dm1 m01 m2 m02 (t) = Cm1 m2 (t)C ∗ m02 m01 (t) and the summations in Eq. (5-2) have been
contract over just two indices J and K. To compute the WSE Eq. (5-15) for the Toda system, the
Schmidt decomposition procedure is applied over the DJK (t) coefficient matrix.

Figure 5-7 presents the results of the Wigner Separability Entropy (WSE) for the Toda system,
alongside the corresponding Von Neumann entropy. The time-dependent behavior of the WSE
closely mirrors the entanglement dynamics captured by the Von Neumann entropy, exhibiting a
consistent pattern with a magnitude difference of two factors. This precise correspondence can be
attributed to the utilization of the same harmonic oscillator expansion of the quantum state in both
measures. Furthermore, the coefficients involved in Eq. (5-16) for pure states are directly connected
to those in Eq. (1-1), further reinforcing the alignment between the WSE and the Von Neumann
entropy in capturing the entanglement characteristics of the system [20].

It is worth noting that the Wigner Separability Entropy (WSE) accurately captures the absence
of entanglement at the initial time, as expected for a separable initial state. As the particles in-
teract and entanglement emerges, the WSE exhibits a corresponding increase, correctly reflecting
the growth of entanglement in the system. The significance of the WSE lies in its ability to be
computed within the classical framework of phase space, making it accessible when the Wigner
function of the system is available. This feature enables a direct connection between the quantum
entanglement dynamics and the classical phase space description, enhancing our understanding of
the entanglement behavior in a comprehensible and measurable manner.

Given that the Von Neumann entropy measures entanglement based on information from the re-
duced density matrix, while the Wigner Separability Entropy (WSE) relies on the reduced Wigner
function, it is intriguing to examine the evolution of the reduced Wigner function over time, as
depicted in Figs. 5-8 and 5-9. From these snapshots, we observe that in both regular and chaotic
cases, the initial coherent state at t = 0 undergoes a transformation and changes its shape. In the
chaotic case, this transformation occurs more rapidly compared to the regular case, aligning with
the initial faster increase of entanglement under chaotic dynamics. Furthermore, we note that the
amplitude of the Wigner function decreases as time progresses, as evidenced by the color bars. This
behavior is consistent with the forthcoming discussion on the linear entropy of the Toda system in
Figs. 5-10b and 5-11b, where the mixture of the reduced state becomes apparent.

From Figs. 5-8 and 5-9, we notice that the reduced Wigner function becomes more dispersed when
the time evolves, and then the uncertainty of the position of the particles in phase space increase,
justifying the increase of the entanglement in Fig. 5-7.

In the next section, we come back to the density amplitude of the reduced Wigner function and its
relation with the entanglement dynamics.
 54 5 Entanglement dynamics Wigner-Weyl representation

6
5
5
4
4
3
3
S

S
2 2

1 Von Neumann 1 Von Neumann

WSE WSE
0 0
0 10 20 30 40 50 0 10 20 30 40 50
t t
(a) Classical dynamics. Parameters N = 90, (b) Chaotic dynamics. Parameters N = 90,
M1 = 1, M2 = 1, E = 7 and ~ = 0.2. M1 = 1, M2 = 0.54, E = 7 and ~ = 0.2.

Figure 5-7.: Comparison of the Wigner Separability Entropy (yellow curve) (WSE) with the entan-
glement measure Von Neumann entropy (blue curve) for regular and chaotic classical
dynamics. The WSE for both cases shows the same dynamical behavior of the Von
Neumann entropy with a difference in amplitude of two factors.

5.3.4. Linear entropy Wigner-Weyl representation

Another entanglement measurement that can be calculated with the Wigner function is the linear
entropy Eq. (3-9). This function has the particularity that when the Weyl transformation is applied
over it, its functional shape doesn’t change, and we get it as a function of the Wigner function as
follows: Z
SL = 1 − (2π~)D dzWr2 (z), (5-26)

where Wr (z) is the reduced Wigner function, z represent the phase space variables of the system
that haven’t been reduced, and D is the number of degrees of freedom of the system. Equation
(5-26) corresponds to the continuous linear entropy in the Wigner-Weyl representation. A discrete
version of it can be defined in a phase space with cells of size ∆q∆p for one degree of freedom in
the following way [10] X
SL = 1 − ∆q∆p Wr2 (qi , pi ; t)∆q∆p, (5-27)
i
where the summation in principle is over all the phase space. The product ∆q∆p at the front of
the summation in Eq. (5-27) satisfies the uncertainty principle.
With the introduction that we have done previously of a discrete linear entropy in terms of the
Wigner function and the explicit expression for the reduced Wigner function, Eq. (5-8), we can
calculate this entanglement measure for the Toda system in the easy way described next.
The reduced Wigner function is calculated in the harmonic oscillator basis employing the Eq.
(5-9) and it is evaluated over the phase space discretization chosen of size ∆q (in our case, we
take ∆p=∆q). As we mentioned before, although the summation in Eq. (5-27) is over all the
phase space, as we can see in Figs. 5-8 and 5-9, the amplitude of the reduced Wigner function is
5.3 Entanglement in Wigner-Weyl representation 55

1.5 1.0
1.0 1.0
2 2 2 0.5
0.5 0.5
0 0.0 0 0.0 0 0.0
p1

p1

p1
0.5 0.5 0.5
2 2 2
1.0 1.0
1.5 1.0
4 2.5 0.0 2.5 4 2.5 0.0 2.5 4 2.5 0.0 2.5
q1 q1 q1
(a) t=0 (b) t=1 (c) t=2
1.5 0.4
1.0 0.3 1.0
2 2 0.2 2
0.5 0.1 0.5
0 0.0 0 0.0 0 0.0

p1
p1

p1

0.5 0.1 0.5

2 2 0.2 2
1.0 0.3 1.0
4 2.5 1.5 4 2.5 0.4 4 2.5
0.0 2.5 0.0 2.5 0.0 2.5
q1 q1 q1
(d) t=3 (e) t=5 (f) t=10
0.3
0.3 0.2
2 0.2 0.2 2
2 0.1
0.1 0.1
0 0.0 0 0.0 0 0.0
p1

p1
p1

0.1 0.1
2 2 2 0.1
0.2 0.2
0.3 0.2
4 2.5 4 2.5 0.3 4 2.5
0.0 2.5 0.0 2.5 0.0 2.5
q1 q1 q1
(g) t=20 (h) t=30 (i) t=50

Figure 5-8.: Snapshots of the reduced Wigner function of the particle one of the Toda system
evolving in time under regular dynamics at several times with the parameters N = 90,
M1 = 1, M2 = 1, E = 7 and ~ = 0.2.

relevant in the regions bounded by the energy condition, so it is enough to evaluate the reduced
Wigner function at that region. From Fig. 5-10b, we notice that the linear entropy, besides being
a first approximation of the Von Neumann entropy is capable of reproducing the behavior of the
entanglement shown by the WSE Fig. 5-10a. The principal difference that we notice between the
two entanglement measures in Fig. 5-10 is the amplitude, whereas, for the linear entropy, its range
is bounded between zero and one because this quantity measures the purity of the reduced state,
where zero linear entropy corresponds to the entanglement in the initial separable state employed
in the Toda system in this case.
 56 5 Entanglement dynamics Wigner-Weyl representation

1.5
1.0 1.0 1.0
2 2 2
0.5 0.5 0.5
0 0.0 0 0.0 0 0.0

p1

p1
p1

0.5 0.5 0.5

2 2 2
1.0 1.0 1.0
4 2.5 1.5 4 4
0.0 2.5 2 0 2 2 0 2
q1 q1 q1
t=0 t=1 t=2
1.0 0.4
0.75
2 0.50 2 0.5 2 0.2
0.25
0 0.00 0 0.0 0 0.0
p1

p1
p1

0.25
2 0.50 2 0.5 2 0.2
0.75 1.0 0.4
4 2 0 2 4 4
2 0 2 2 0 2
q1 q1 q1
t=3 t=5 t=10

0.4 0.3
0.2 0.2
2 0.2 2 2
0.1 0.1
0 0.0 0 0.0 0 0.0
p1

p1

p1

0.2 0.1 0.1

2 2 0.2 2
0.2
0.4 0.3
4 2 0 2 4 2 0 2 4 2 0 2
q1 q1 q1
t=20 t=30 t=50

Figure 5-9.: Snapshots of the reduced Wigner function of the particle one of the Toda system
evolving in time under chaotic dynamics at several times with the parameters N = 60,
M1 = 1, M2 = 0.54, E = 7 and ~ = 0.2.

The results in Figs. 5-10 and 5-11 show that the oscillating patterns in the entanglement dynamics
inside the Ehrenfest time for different phase space discretizations are the same, meaning that at
this regime they are most affected by the caustics presence instead of the numerical discretization.

In the following analysis, we will delve into the insights provided by the phase space representation
of quantum entanglement measurements and the linear entropy, focusing on their connection to
the classical dynamics of entanglement and the distinct oscillating pattern observed in our previous
findings. As mentioned earlier, as time progresses, the reduced Wigner function exhibits regular
5.3 Entanglement in Wigner-Weyl representation 57

1.0
5
0.8
4
0.6
3
WSE

SL
2 0.4

1 0.2

0 0.0
0 10 20 30 40 50 0 10 20 30 40 50
t t
(a) Wigner Separability entropy for the Toda (b) Linear entropy calculated with the reduced
system Wigner function of the Toda system. The dis-
cretization employed was ∆q = ∆p = 0.05.

Figure 5-10.: Comparison between the entanglement measures in the Wigner Weyl representation,
the Wigner Separability Entropy, and the linear entropy for the Toda system under
regular dynamics. The parameters employed were E = 7, N = 90 and ~ = 0.2.

5 1.0

4 0.8

3 0.6
WSE

SL

2 0.4

1 0.2

0 0.0
0 10 20 30 40 50 0 10 20 30 40 50
t t
(a) WSE for the Toda system under chaotic dy- (b) Linear entropy in the Weyl representation
namics. for the Toda system under chaotic dynamics.
The discretization employed was ∆q = ∆p =
0.05.

Figure 5-11.: Parallel between the WSE and the Weyl representation of the linear entropy follo-
wing chaotic dynamics. The parameters employed were E = 7, N = 90 and ~ = 0.2.

dynamics, as illustrated in Fig. 5-8, displaying a periodic behavior that becomes more apparent
from t = 0 until t = 10. This periodicity can be associated with the frequency at which classical
trajectories traverse the torus along the submanifold of interest, in our case, corresponding to the
first particle submanifold. To investigate the relationship between this periodic behavior and the
oscillating pattern observed in Figs. 5-10 and 5-11, we present Figs. 5-13 and 5-14. Figure 5-13
 58 5 Entanglement dynamics Wigner-Weyl representation

0
p1

2 1 0 1 2 3
q1

Figure 5-12.: Projection of the classical trajectories over the manifold (q1 , p1 ) for 5 points initially
distributed in a Gaussian distribution evolving for a time of 300 s. The two regions
where the points start to overlap are the caustics, regions where the wave and ray
descriptions of different physical systems have had divergence problems [63]. The
same distribution of the caustics can be seen in the chaotic case of the remaining
torus.

showcases the reduced Wigner function at several time points where the entanglement dynamics of
the Toda system reach a local maximum. Notably, during these maximum points of the oscillating
pattern, the reduced Wigner function predominantly exhibits positive values throughout the phase
space, reminiscent of the dynamics of a classical Liouville distribution. Conversely, in Fig. 5-14,
we present the reduced Wigner function at various time steps when the entanglement reaches a
local minimum. Here, the reduced Wigner function primarily localizes in specific regions on the left
side of the phase space, displaying significant negative values. To characterize these regions where
the reduced Wigner function exhibits localized behavior during the local minimum entanglement,
we present the corresponding classical result in Fig. 5-12, which depicts the projected trajectory
of initial points following the dynamics dictated by the classical Hamiltonian of the Toda system
Eq. (4-5). Analogous to optical systems, when classical trajectories move tangentially to the torus,
certain regions on the submanifold experience a concentration of classical trajectory movement,
known as caustics [63]. In Fig. 5-12, these caustics can be identified as the regions with a higher
density of points. From the aforementioned observations, we can conclude that the minima in
5.3 Entanglement in Wigner-Weyl representation 59

1.00 0.3 0.2

0.75 0.2
2 0.50 2 2 0.1
0.25 0.1
0 0.00 0 0.0 0 0.0

p1

p1
p1

0.25 0.1
2 0.50 2 2 0.1
0.75 0.2
1.00 0.3 0.2
4 2.5 0.0 2.5 4 2.5 4 2.5
0.0 2.5 0.0 2.5
q1 q1 q1
(a) t=1.9 (b) t=4.8 (c) t=7.8
0.20 0.20
0.4 0.15 0.15
2 0.2 2 0.10 2 0.10
0.05 0.05
0 0.0 0 0.00 0 0.00
p1

p1

p1
0.05 0.05
2 0.2 2 0.10 2 0.10
0.4 0.15 0.15
4 2.5 4 2.5 0.20 4 2.5 0.20
0.0 2.5 0.0 2.5 0.0 2.5
q1 q1 q1
(d) t=10.8 (e) t=17.4 (f) t=20.4

Figure 5-13.: Reduced Wigner function at several times where the linear entropy in Fig. 5-10b
is maximum. The reduced Wigner function shows itself to be mostly positive in the
time steps displayed, indicating that the same dynamics can be obtained from a
classical distribution.

the entanglement dynamics correspond to time steps when the reduced Wigner function traverses
the classical location of the caustics, exhibiting strong oscillatory behavior and significant negative
amplitudes. This correlation between the dynamics of the reduced Wigner function near the caustics
and the reduction in the entanglement makes sense since the uncertainty of the quantum state
decreases when the reduced Wigner function becomes concentrated in a smaller region of the phase
space.

To finish this section, we present an analytical result for the linear entropy in the Wigner-Weyl
representation and we set down the numerical advantage of its implementation. The continuous
linear entropy for the Toda system can be obtained by replacing the reduced Wigner function of
the Toda system Eq. (5-8) in Eq. (5-26)
X X
∗ ∗
SL = 1 − 2π~ Cm1 m2 (t)Cm 0 m (t)Cl1 l2 (t)Cl0 l (t)
2 2
1 1
m1 ,m2 ,m01 l1 ,l2 ,l10
Z (5-28)
× WT 1 (q1 , p1 )m1 ,m01 WT 1 (q1 , p1 )l1 ,l10 dq1 dp1 ,

solving the integral over the squared reduced Wigner function (see appendix J for details), we
get the following expression to implement numerically the continuous linear entropy for the Toda
 60 5 Entanglement dynamics Wigner-Weyl representation

1.0 1.0 0.75

2 2 2 0.50
0.5 0.5 0.25
0 0.0 0 0.0 0 0.00
p1

p1

p1
0.5 0.5 0.25
2 2 2 0.50
1.0 1.0 0.75
4 2.5 4 2.5 4 2.5 0.0 2.5
0.0 2.5 0.0 2.5
q1 q1 q1
(a) t=3.7 (b) t=6.7 (c) t=9
0.6
0.4 0.4
2 2 0.4 2
0.2 0.2 0.2
0 0.0 0 0.0 0 0.0
p1

p1
p1

0.2 0.2 0.2

2 2 2
0.4 0.4
0.4
4 2.5 4 2.5 0.6 4 2.5
0.0 2.5 0.0 2.5 0.0 2.5
q1 q1 q1
(d) t=11.7 (e) t=16.1 (f) t=18.8

Figure 5-14.: Reduced Wigner function at several times where the linear entropy in Fig. 5-10b is
minimum. Here, we notice two different regions where the reduced Wigner function
concentrates, which coincide with the same two caustics in Fig. 5-12.

system
X X
∗ ∗
SL = 1 − 2π~ Cm1 m2 (t)Cm 0 m (t)Cl1 l2 (t)C(m −m0 +l )l (t)
2 1 1 2
(5-29)
1 1
m1 ,m2 ,m01 l1 ,l2

π Γ[min(m1 , m01 ) + |m1 − m01 | + 1]

· dm1 m01 dl1 (m1 −m01 +l1 ) .
2 min(m1 , m01 )!

The previous expression has the advantage over the discrete case Eq. (5-27) that we don’t need to
evaluate the Wigner function over each point of the phase space given that the integral could be
solved analytically thanks to the Laguerre polynomials properties.
In both cases, continuous and discrete, the linear entropy reveals its advantage over the Von Neu-
mann and WSE not involving diagonalization procedures.
6. Conclusions and perspectives

6.1. Conclusions
In our immersion in descriptions of the entanglement dynamics in the Wigner-Weyl representation,
systems that are favorites to realize these descriptions have been toy models as dynamical maps.
In those descriptions, different challenges have been solved such as the necessity of the formulation
of a discrete version of the Wigner function since in the simulations we do not work on a conti-
nuous position and momentum basis and the necessity of considering the concept of integrability
at a quantum level to research its influence in quantum dynamics. In relation to entanglement
measurements in this representation, the proposals that we found are measurements derived from
the family of entropic measurements because of the close relation that we can establish between
the uncertainty of the output of a measurement in a system and the mixture of the quantum sta-
te itself. These measurements consider the negative regions of the Wigner function as an issue if
you try to understand the Wigner function as a probability distribution and apply monotonically
functions over it as the logarithm operation and directly as a quantifier of the quantumness or the
entanglement of the system via the integration over these negative regions, for instance. Following
this track to find a way to describe and measure the entanglement of a quantum system from the
classical concept of the phase space via the Wigner function, we found two entanglement measu-
rement proposals in this century that are suitable for the Toda system, which is a physical system
that we chose because have been broadly studied and employed to find a classical description of
the entanglement.

In this work, via the WSE and the linear entropy in the Wigner-Weyl representation, we were able
to reproduce the quantum and classical approaches to the entanglement dynamics in the Toda
system reported in [16]. To achieve this goal, we simulated in the Python programming language
the quantum dynamics of the Toda system thanks to the method of harmonic oscillator state
expansion, surpassing the implementation challenge of employing a considerable size of basis. In
parallel, we also got the classical dynamics of the system state under dynamics generated by a
classical Hamiltonian developing a symplectic propagation. We characterized the similarities and
differences between the quantum measurement employed in [16] in the wave function representation
and for us in the Wigner-Weyl representation. In this comparison, we were able to show:

The simplicity with which, via the harmonic oscillator expansion of a quantum state, we can
move from the wave representation to the phase space representation employing well-known
bases such as the Hermite and Laguerre polynomials.

The proportionality relation between the Von Neumann entropy and the WSE.
 62 6 Conclusions and perspectives

The rich and direct classical interpretation that we can do from the Wigner function in
relation to classical properties of the physical systems, in particular the Toda model and the
entanglement dynamics from tools that we explored as the stationary Wigner function, the
quantum Poincare surface of section, and the reduced Wigner function at some submanifold.
The interpretation of the oscillating pattern of the quantum entanglement dynamics, where we
found that the minimum corresponds to relevant amplitudes of the reduced Wigner function
with significant negative values in the localization of the caustics.
One element relevant in the description of the entanglement dynamics that we found is the integra-
bility of the system in which we are interested because, as we mentioned, typically, the integrability
of the system affects the uncertainty of the system state at some point. This characterization for the
Toda system was done via the employment of the concept of quantum Poincare surface of sections
defined in the Wigner-Weyl representation for stationary distributions. In addition, the obtained
stationary reduced Wigner distribution is able to show characteristic behaviors to differentiate re-
gular and chaotic behaviors via particular patterns that show the presence of classical torus via big
negative or positive Wigner function amplitudes as well as small fringes testifying the destruction
of torus and the loss of integrability.

6.2. Perspectives
After the results and the work presented in this thesis, we find points of the entanglement dynamics
in the phase space that can be approached with it as a starting point, such as:
The characterization of the quantum chaos from the eigenvalues of the density matrix of the
Toda system in comparison with the results of the Random Matrix Theory.
The dynamical entanglement saturation of the WSE in the quantum case and its quantifica-
tion and interpretation in the phase space.
The contribution of the negative regions of the reduced Wigner function in the quantification
of entanglement.
The comparison of the quantum results with alternative classical approaches to the entangle-
ment dynamics, such as the Separability entropy or the truncated Wigner approximation.
The numerical implementation done in the realization of this thesis can be parallelized to
improve the processing speed, achieve higher basis numbers, and explore higher energy condi-
tions where the classical dynamics of the Toda system are less mixed and more fully chaotic.
The WSE is an entanglement measure that can be employed on mixed states, but there are
no works as far as we know where it measures the entanglement of these quantum states.
Due to the arbitrariness of the coordinate set selection in the WSE, this opens the possibility
to explore different phase space separations and the chance to measure another kind of en-
tanglement. This possibility has been noted previously in [7], but with the present work and
the presented framework, it is worth checking.
A. Coherent states

In classical mechanics, the system state can be represented by a point or a distribution in the
phase space. In quantum mechanics, there are some states with similar properties to the classical
states called coherent states [64]. The coherent states represented by the Hilbert vector |αi, are the
superposition of Fock states, with the property that they don’t change more than one phase by the
bosonic annihilation operator
â|αi = α|αi, (A-1)
where α is a complex number and â = (2m~ω)−1/2 (mω q̂ + p̂) is the annihilation operator with m
and ω parameters of a quantum harmonic oscillator configuration. Given that a coherent state is
a superposition of Fock states, it physically corresponds to a reservoir of an undetermined photon
number. Other properties that characterize the coherent states are listed next:
They can be obtained from a translation from the ground state of the harmonic oscillator,
such as is shown in the appendix B.
They saturate the uncertainty principle, which means that for a coherent state the relation
σq · σp ≤ ~/2 with equality is satisfied, where σq and σp are the position and momentum
dispersion, respectively.
They are overcomplete, which means that there are redundant elements in a chosen basis,
given that a representation in this basis is not unique.
The coherent state basis is not orthogonal.
It satisfies a closure relation Z
1
I= d2 α|αihα|, (A-2)
π
where the integration is over the real and imaginary parts of α.
B. Dispersion initial classical state

In this appendix, we present some additional elements related to the coherent states introduced
in appendix A, which include the obtainment of them from the translation operator and some
useful representation in terms of the Fock basis. We also present the position and momentum
representation of a coherent state.
† −α∗ â
A coherent separable state comes from applying the Weyl-Heisenberg translation operator eαâ
over the ground state of the harmonic oscillator, such as is shown in the following equation

† −α∗ â
|αi = eαâ |0i, (B-1)
with r
mω i
α= hqi + √ hpi.
2~ 2mω~
The previous result using Baker Campbell Hausdorff formula (BCH) [65] and the completeness for
P
the Fock basis n |nihn| can be expressed as

2 /2
X αn
|αi = e−|α| √ |ni, (B-2)
n n!

from where
αn 2 /2
hn|αi = e−|α| √ . (B-3)
n!
In the case of a two-dimensional separable coherent state, the previous coefficient is just

1 1 2 1 2
hn1 n2 |α1 α2 i = e− 2 |α1 | α1n1 e− 2 |α2 | α2n2 (B-4)
n1 !n2 !

In addition, by replacing the harmonic oscillator wave function Eq. (D-1) in the Eq. (B-2) we get
X αn
2 /2
hq|αi = e−|α| √
n n!
 1/4 r !
1 Mω M ωq 2 M ω
× √ e− 2~ Hn q
2n n! π~ ~
M ω 1/4 − M ωq2 −|α|2 /2
 
= e 2~ e
π~
r !
X αn 1 Mω
× √ √ Hn q , (B-5)
n n! 2n n! ~
 66 B Dispersion initial classical state

using the following generating function for the Hermite polynomials:

∞
X Zn 2 +2ξz
Hn (ξ) = e−z (B-6)
n!
n=0

we get
k 1/2 −k2 q2 /2 −|α|2 /2 −α2 /2+√2kαq
hq|αi = e e e (B-7)
π 1/4
and  √ 2
k −k2 q− 2α r
hq|αihα|qi = e k
(B-8)
π 1/2
p
where αr represents the real part of α and was used k = M ω/~. From the previous expression,
comparing it with a Gaussian distribution, we can get the dispersion in the position of the coherent
√
state σq = 1/( 2k). As the coherent state satisfies the uncertainty principle, we have σq · σp = ~/2,
√
from where the dispersion in the p direction corresponds to σp = ~k/ 2.
C. Weyl transformation non-diagonal harmonic
oscillator elements

In this annex, we calculate the Weyl transform Eq. (5-3) for the non-diagonal elements that we can
identify with the Fock elements |nihm|.
Z
1
WT (q, p)n,m = dξ e−ipξ/~ hq + ξ/2|nihm|q − ξ/2i. (C-1)
2π~

Now, to start solving the previous integral, we use Eq. D-1 to express the wave function of the Fock
elements in the position representation in terms of the Hermite polynomials. Replacing Eq. D-1 in
Eq. C-1

1  mω 1/2
WT (q, p)n,m = √
2π~ 2n+m n!m! π~
Z r  r 
mω(q+ξ/2)2 mω(q−ξ/2)2 mω mω
−ipξ/~ − −
× dξ e e 2~ e 2~ Hn (q + ξ/2) Hm (q − ξ/2) .
~ ~

For convenience, we are going to do the next change of variables q 0 = 0

p mω p mω
~ q, ξ = ~ ξ and
0 p
p = mω~ , then
√

1
WT (q 0 , p0 )n,m = √
2π 3/2 ~
Z 2n+m n!m!
0 0 1 0 0 2 1 0 0 2
× dξ0 e−ip ξ e− 2 (q +ξ /2) e− 2 (q −ξ /2) Hn (q 0 + ξ 0 /2)Hm (q 0 − ξ 0 /2)
1 02 02
= √ e−q −p
3/2 n+m n!m!
Z2π ~ 2
0 0 2
× dξ0 e−(ξ /2+ip ) Hn (q 0 + ξ 0 /2)Hm (q 0 − ξ 0 /2),

making the substitution, the previous expression becomes u = ξ 0 /2 + p0

(−1)m 02 02
WT (q 0 , p0 )n,m = √ e−q −p
3/2 n+m
Zπ ~ 2 n!m!
2
× dξ0 e−u Hn [u + (q 0 − ip0 )]Hm [u − (q 0 + ip0 )]
(−1)m 02 02
= √ e−q −p
3/2 n+m n!m!
Zπ ~ 2
2
× dξ0 e−u Hn (u + y1 )Hm (u − y2 )
 68 C Weyl transformation non-diagonal harmonic oscillator elements

with y1 = q 0 + ip0 and y2 = q 0 − ip0 . Now using the expression D-4, we have

(−1)m 02 02
WT (q 0 , p0 )n,m = √ e−q −p
3/2 n+m
π ~ 2 ! n!m!!
n X m Z
X n m m−k0 2
× 0 (2y1 )n−k
(−2y2 ) dξ0 e−u Hk (u)Hk0 (u)
0
k k
k k
(−1)m 02 02
= √ e−q −p
π~ 2 n+m n!m!
min(n,m)
! !
X
k n m
× 2 k! (2y1 )n−k (−2y2 )m−k .
k k
k

To simplify the previous result, we are going to use the Rodriguez formula for the associated
Laguerre polynomials
n
k
X (n + k)!
Ln (x) = (−1)m xm . (C-2)
(n − m)!(k + m)!m!
m=0

For that we are going to organize the expression of Wn,m (q 0 , p0 ) in the following way
1 √ 02 02
WT (q 0 , p0 )n,m = n!m!e−q −p
π~
min(n,m)
1 X (−1k )
× 2 2 (n+m) 2−k (q 0 + ip0 )n−k (q 0 − ip0 )m−k
(n − k)!(m − k)!k!
k
1 √ −q 02 −p02
= n!m!e
π~
min(n,m)
02 02 21 (n+m) i(n−m) arctan(p0 /q 0 )
X (−1)k
× [2(q + p )] e [2(q 02 + p02 )]−k ,
(n − k)!(m − k)!k!
k

making the substitution k = min(n, m) − l

(−1)min(n,m) √ 02 02 1 0 0
WT (q 0 , p0 )n,m = n!m!e−q −p [2(q 02 + p02 )] 2 |n−m| ei(n−m) arctan(p /q )
π~ max(n, m)!
min(n,m)
X (−1)l max(n, m)!
× [2(q 02 + p02 )]l
(min(n, m) − l)!(|m − n| + l)!l!
l

from where
(−1)min(n,m) √ 02 02 1 0 0
WT (q 0 , p0 )n,m = n!m!e−q −p [2(q 02 + p02 )] 2 |n−m| ei(n−m) arctan(p /q )
π~ max(n, m)!
|n−m|
× Lmin(n,m) [2(q 02 + p02 )]. (C-3)

The particular case when n = m gives the following Wigner function expression for the Fock basis
1 02 02
Wn (q 0 , p0 ) = (−1)n e−(q +p ) Ln 2(q 02 + p02 ) .
 
(C-4)
π~
D. Wave function expansion in harmonic
oscillator basis

Next are presented some expressions used in quantum dynamics following the harmonic oscillator
basis method.
The harmonic oscillator eigenbasis in the position representation in terms of the Hermite
polynomials !
M ω 1/4 − M ωx2
  r
1 Mω
ψn (x) = √ e 2~ Hn x , (D-1)
2n n! π~ ~
in an equivalent way
 1/4
1 Mω ξ2
ψn (ξ) = √ e− 2 Hn (ξ) , (D-2)
2n n!π 1/2 ~

p
where ξ = M ω/~x and
2 d −ξ2
Hn (ξ) = (−1)n eξ (e ).
dξ n

Orthogonality relationship of the Hermite polynomials

Z ∞
2
dξHn (ξ)Hm (ξ)e−ξ = 2n n!π 1/2 δnm . (D-3)
−∞

Hermite polynomials of a binomial term using the Taylor expansion

n
!
X n
Hn (ξ + y) = Hk (ξ)(2y)n−k . (D-4)
k
k

Some useful derivatives

   
d 2
− ξ2
2
− ξ2 1
e Hn (ξ) = e nHn−1 (ξ) − Hn+1 (ξ) ,
dξ 2

d2
   
2
− ξ2
2
− ξ2 1 (2n + 1)
e Hn (ξ) = e Hn+2 (ξ) − Hn (ξ) + n(n − 1)Hn−2 (ξ) .
dξ 2 4 2
Using the before derivatives, the derivatives of the harmonic oscillator eigenfunctions are
given by
 70 D Wave function expansion in harmonic oscillator basis

r r
n n+1
ψn0 (ξ) = ψn−1 (ξ) − ψn+1 (ξ),
2 2
r r
00 n(n − 1) (2n + 1) (n + 1)(n + 2)
ψn (ξ) = ψn−2 (ξ) − ψn (ξ) + ψn+2 (ξ) (D-5)
4 2 4
E. Amplitude coefficients system state at time t

To calculate the amplitude coefficients in Eq. (4-3) we use the energy basis of the system in the
following way:

Ĥ|φl i = El |φl i
hn1 n2 |Ĥ|φl i = El hn1 n2 |φl i
X
hn1 n2 |Ĥ|m1 m2 ihm1 m2 |φl i = El hn1 n2 |φl i,
m1 m2

where we have used the energy eigestates |φl i. Moreover, as El is an eigenvalue of the Hamiltonian,
the number of them is limited by the dimension of the energy matrix hn1 n2 |Ĥ|m1 m2 i and it isn’t
a value that we can choose. Now labeling
l
hm1 m2 |φl i = C̄m 1 m2
, (E-1)

we have for example for n1 = n2 = 0, 1 and l = 0, the following system equations

     
E0000 E0001 E0010 E0011 0
C̄00 0
C̄00
E   0 C̄ 0 
 0100 E0101 E0110 E0111  C̄01  01 
· = E0 0 .

   0
E1000 E1001 E1010 E1011  C̄10  C̄10 
E1100 E1101 E1110 E1111 0
C̄11 0
C̄11

with En1 n2 m1 m2 = hn1 n2 |Ĥ|m1 m2 i the energy matrix elements presented in appendix F.
Then, the vector of coefficients for each value of l is the l eigenvector of the energy matrix. As a
consequence, the state in Eq. 4-3 can be written as
X X
|ψ(t)i = e−iEl t/~ hm1 m2 |φl ihφl |n1 n2 ihn1 n2 |ψ(0)i|m1 m2 i (E-2)
m1 m2 n1 n2 l

with X
Cm1 m2 (t) = e−iEl t/~ hm1 m2 |φl ihφl |n1 n2 ihn1 n2 |ψ(0)i. (E-3)
n1 n2 l
F. Elements of the energy matrix

Using the harmonic oscillator eigenbasis, the Hamiltonian elements can be expressed as

hn1 n2 |Ĥ(q̂1 , q̂2 , p̂1 , p̂2 )|m1 m2 i =

p̂21 p̂22
q q q q
M1 ω1 M 2 ω2 M1 ω1 M2 ω2
− q̂1 q̂2 q̂1 − q̂2
hn1 n2 | + +e ~ +e ~ +e ~ ~ − 3|m1 m2 i.
2M1 2M2

We can separate the previous calculation in the following way:

d2
Z
2
hn1 n2 |pˆ1 |m1 m2 i = −~ 2
dq1 ψn∗ 1 (q1 ) 2 ψm1 (q1 )δn2 ,m2
dq1
 Z
M1 ω1
= −~2 dq1 ψn∗ 1 (q1 )
~
"r r #
m1 (m1 − 1) (2m1 + 1) (m1 + 1)(m1 + 2)
× ψm1 −2 (q1 ) − ψm1 (q1 ) + ψm1 +2 (q1 ) δn2 ,m2 .
4 2 4

~2 M1 ω1
 
hn1 n2 |pˆ1 2 |m1 m2 i = −
2 ~
hp p i
× m1 (m1 − 1)δn1 ,m1 −2 − (2m1 + 1)δn1 ,m1 + (m1 + 1)(m1 + 2)δn1 ,m1 +2 δn2 ,m2 . (F-1)

d2 ~2 M2 ω2
Z  
2
hn1 n2 |pˆ2 |m1 m2 i = −~ 2
dq2 ψn∗ 2 (x2 )
ψm (q2 )δn1 ,m1 = −
dq22 2 2 ~
hp p i
× m2 (m2 − 1)δn2 ,m2 −2 − (2m2 + 1)δn2 ,m2 + (m2 + 1)(m2 + 2)δn2 ,m2 +2 δn1 ,m1 . (F-2)
 74 F Elements of the energy matrix

Z
−q̂1 /d
V hn1 |e dq1 ψn∗ 1 (q1 )e−q1 /d ψm1 (q1 )δn2 ,m2
|m1 i = V
Z q
1 1 −ξ 2 − M ~ω ξ/d
= V p ·p dξe 1 1 Hn1 (ξ) Hm1 (ξ) δn2 ,m2
2n1 n1 !π 1/2 2m1 m1 !π 1/2
Z  q 2
1 1 − ξ+ 2d1 ~ 1
+ 4d ~
= V p ·p dξe M1 ω1 M1 ω1
Hn1 (ξ) Hm1 (ξ) δn2 ,m2
2n1 n1 !π 1/2 2m1 m1 !π 1/2
Z r ! r !
1 1 1 ~ 2 1 ~ 1 ~
= V p ·p e 4d M1 ω1 dξe−u Hn1 u − Hm1 u − δn2 ,m2
n
2 n1 !π
1 1/2 m
2 m1 !π
1 1/2 2d M1 ω1 2d M1 ω1
1 1 1 ~
= V p ·p e 4d2 M1 ω1
2n1 n1 !π 1/2 2m1 m1 !π 1/2
n1 Xm1 r !n1 +m1 −k−l ! !Z
X 1 ~ n1 m1 2
× V − dξe−u Hk (u) Hl (u) δn2 ,m2
d M1 ω1 k l
k l
1 1 1 ~
= √ ·√ e 4d2 M1 ω1
2n1 n1 !
2m1 m1 !
n1 X
m1 r !n1 +m1 −k−l ! !
X 1 ~ n1 m1
× − (2k k!)δk,l δn2 ,m2 .
d M1 ω1 k l
k l

r !n1 +m1
−q̂1 /d 1 1 1 ~ 1 ~
V hn1 |e |m1 i = V √ ·√ m e 4d2 M1 ω1 −
2n1 n1 ! 2 1 m1 ! d M1 ω1
min(n1 ,m1 )  k ! !
X 2M1 ω1 d2 n1 m1
× k! δn2 ,m2 . (F-3)
~ k k
k

r !n2 +m2
1 1 1 ~ 1 ~
V hn2 |eq̂2 /d |m1 i = V √ ·√ m e 4d2 M2 ω2
2n2 n2 ! 2 2 m2 ! d M2 ω2
min(n2 ,m2 )  k ! !
X 2M2 ω2 d2 n2 m2
× k! δn1 ,m1 . (F-4)
~ k k
k

r !n1 +m1
q̂1 /d−q̂2 /d 1 1 1 ~ 1 ~
V hn1 n2 |e |m1 m2 i = V √ ·√ m e 4d2 M1 ω1 −
2n1 n1 ! 2 1 m1 ! d M1 ω1
min(n1 ,m1 )  k ! !
X 2M1 ω1 d2 n1 m1
× k!
~ k k
k
r !n2 +m2
1 1 1 ~ 1 ~
× √ n ·√ m 2
e 4d 2 2
M ω
2 2 n2 ! 2 2 m2 ! d M2 ω2
min(n2 ,m2 )  l ! !
X 2M2 ω2 d2 n2 m2
× l! . (F-5)
~ l l
l
 75

Then, the Hamiltonian elements are given by

hn1 n2 |H(q̂1 , q̂2 , p̂1 , p̂2 )|m1 m2 i

pˆ1 2 pˆ2 2
= hn1 | |m1 i + hn2 | |m2 i
2M1 2M2
+ V hn1 |e−q̂1 /d |m1 i + V hn2 |eq̂2 /d |m2 i
+ V hn1 n2 |eq̂1 /d−q̂2 /d |m1 m2 i − 3hn1 n2 |m1 m2 i. (F-6)
G. Oscillation frequencies

For the simulations and to get the results in [16], we have to choose the free parameters carefully.
These parameters are the frequencies of the harmonic oscillators used to do the representation ω1
and ω2 .
To select the optimal frequencies, we can start doing small approximations over the positions in
the Hamiltonian (Eq. (4-1)) to get

p21 V p2 V
H= + 2 x21 + 2 + 2 x22 − 2x1 x2 , (G-1)
2M1 d 2M2 d
from here, we can do the correspondence with the simple harmonic oscillator, from where

V Mi ωi2
= , (G-2)
d2 2
where the sub-index i identifies the first or second harmonic oscillator. Taking the same parameter
p
values as in [16] (V = d = 1), we get ωi = 2/Mi .
H. Numerical aspects entropies calculations

Von Neumann entropy

To calculate the entropy, we use the Singular Value Decomposition over the coefficient matrix Eq.
E-3 of dimension d.

Cm1 m2 (t) = U ΛV †
d p
X
= λl Um1 l Vm∗ 2 l
l=1

where U and V are unitary matrices built on the basis of the Hilbert spaces in which the Hilbert
space of the system state is factorized. With the expression for the coefficients, the state of the
system in a time t is given by the Schmidt decomposition
d p
X
|ψ(t)i = λl |ul i ⊗ |vl i (H-1)
l=1
P P
where |ul i = m1 Um1 l |φm1 i and |vl i = m1 Vm1 l |φm1 i.
Now that the partial trace is the same for both subsystems, the Von Newman entropy is given by
d
X
S=− λi ln λi (H-2)
i

Shannon entropy

To calculate a classical entropy, in [16] a coarse-grained entropy is proposed, where the probability
that points come from the dynamics of an initial distribution moving in the phase space is employed.
This probability is assigned according to the point in each cell of a proposed discretization process.
As we want to compare with the quantum case, this discretization of the phase space is done with
p
cells of size ~ef f ective (with ~ef f ective taking values as 0.1, 0.2, ) in the plane of the phase space
(x1 , p1 ). The following expression was used to calculate the coarse-grained Shannon entropy
X
Scl = − pi ln(pi ), (H-3)

where pi is the number of points in the i cell over the total number of points.
I. Numerical implementation Wigner Separability
Entropy chaotic maps

In Sec. (5.3.2) the separation of the phase space chosen to calculate the WSE for the quantum Baker
and quantum cat maps was x = (q) and y = (p). With that separation and the basis selection, the
coefficients in Eq. (5-14) correspond to the amplitudes of the Wigner function in Fig. 5-4 for the
quantum Baker map and Fig. 5-5 for the quantum cat map.
As an example of the calculation of the Schmidt coefficients, let’s take the Wigner function in Fig.
5-4a. We calculate the N = 29 eigenvalues of it and normalize them

~ = [9.99×10−1 , 6.25e×10−3 , 1.30×10−15 , 1.22×10−15 , ..., 8.99×10−17 , ..., 1.64×10−17 , 1.34×10−18 ].

µ̃
(I-1)
From the previous list of eigenvalues, we notice that the first eigenvalue is almost one and the
others are at least two orders of magnitude smaller, meaning that in the diagonal representation of
the Wigner function Eq. (5-15) we get

W (x, y) ≈ u1 (x)vn (y), (I-2)

where µ̃1 was approximated as one and the other eigenvalues to zero for the purpose of the analysis.
From the previous result, since Eq. (5-16) measure the separability of the Wigner function given
the phase space separation, we get that the WSE for Fig. 5-4a is equal to zero as is shown in Fig.
5-6a. The previous result is an expected result since the Wigner function of the coherent state in
Fig. 5-4a corresponds to Wα in Eq. (2-13) which can be written as in Eq. (I-2).
J. Phase space integral squared reduced Wigner
function

The Integral Z
WT 1 (q10 , p01 )m1 ,m01 WT 1 (q10 , p01 )l1 ,l10 dq10 dp01 (J-1)

after replacing the function W1 using the Eq. (5-4)

Z
WT 1 (q10 , p01 )m1 ,m01 WT 1 (q10 , p01 )l1 ,l10 dq10 dp01
Z
02 02 1 0 0
= dm1 m1 dl1 l1 dq 0 dp0 e−2(q +p ) [2(q 02 + p02 )] 2 (|α|+|β|) ei(α+β) arctan(p /q )
0 0 (J-2)

· L|α| 02 02 |β| 02 02
µ [2(q + p )]Lν [2(q + p )],

where 0
(−1)min(m1 ,m1 )
q
dm1 m01 = m1 !m01 !, (J-3)
π~ max(m1 , m01 )!
0
(−1)min(l1 ,l1 )
q
dl1 l10 = l1 !l10 !, (J-4)
π~ max(l1 , l10 )!

α = m1 − m01 , β = l1 − l10 , µ = min(m1 , m01 ) and ν = min(l1 , l10 ).

Using polar coordinates in Eq. (J-2)
Z
WT 1 (q10 , p01 )m1 ,m01 WT 1 (q10 , p01 )l1 ,l10 dq10 dp01 (J-5)
Z
2 1
= dm1 m01 dl1 l10 dθdrre−2r (2r2 ) 2 (|α|+|β|) ei(α+β)θ L|α| 2 |β| 2
µ (2r )Lν (2r )
Z
1
= 2πδα,−β dm1 m01 dl1 l10 due−u u|α| L|α| |α|
µ (u)Lν (u)
4
1 δµ,ν
= 2πδα,−β dm1 m01 dl1 l10 Γ(µ + |α| + 1),
4 µ!

where in the third line of the previous expression was used the substitution u = 2r2 .
Bibliografı́a

[1] M. Mazzanti, R. X. Schüssler, J. D. Arias Espinoza, Z. Wu, R. Gerritsma, and A. Safavi-Naini.

Trapped ion quantum computing using optical tweezers and electric fields. Phys. Rev. Lett.,
127(260502), 2021.

[2] X. Pan, Y. Zhou and H. Yuan. Engineering superconducting qubits to reduce quasiparticles
and charge noise. Nat Commun, 13(7196), 2021.

[3] L.S. Madsen, F. Laudenbach, M.F. Askarani, et al. Quantum computational advantage with
a programmable photonic processor. Nature, 606:75–81, 2022.

[4] D. Bluvstein, H. Levine, G. Semeghini, et al. A quantum processor based on coherent transport
of entangled atom arrays. Nature, 604:451–456, 2022.

[5] A. Y. Khrennikov. Echoing the recent google success: Foundational roots of quantum supre-
macy. ArXiv, abs/1911.10337, 2019.

[6] J.J. Sakurai. Modern Quantum Mechanics. Addison-Wesley, 1994.

[7] A. Buchleitner, C. Viviescas and M. Tiersch. Entanglement and Decoherence: Foundations

and Modern Trends. Springer, Berlin Heidelberg, 2009.

[8] D. Gross, S. T. Flammia, and J. Eisert. Most quantum states are too entangled to be useful
as computational resources. Phys. Rev. Lett., 102(190501), 2022.

[9] V. Vedral, M. B. Plenio, M. A. Rippin, and P. L. Knght. Quantifying entanglement. Phys.

Rev. Lett., 78(2275), 1997.

[10] G. Manfredi & M. R. Feix. Entropy and wigner functions. Phys. Rev. E., 62(4), 2000.

[11] D. James, P. Kwiat, W. Munro, and A. White. Measurement of qubits. Phys. Rev. A.,
64(052312), 2001.

[12] G. Sentı́s, C. Eltschka, O. Gühne, M. Huber, and J. Siewert. Quantifying entanglement of

maximal dimension in bipartite mixed states. Phys. Rev. Lett., 117(190502), 2016.

[13] C. Zhang Y. Huang C. Li G. Guo S. Xie, Y. Zhao and J. H. Eberly. Experimental examination
of entanglement estimates. Phys. Rev. Lett., 130(150801), 2023.

[14] J. S. Moreno. Semiclassical propagators of wigner function: a comparative study, 2020.

[15] T. Dittrich. Semiclassical propagator of the wigner function. Phys. Rev. Lett., 96(070403),
2006.
 86 Bibliografı́a

[16] G. Casati, I. Guarneri, & J. Reslen. Classical dynamics of quantum entanglement. Phys. Rev.
E., 85(036208), 2012.

[17] T. Prosen. Complexity and nonseparability of classical liouvillian dynamics. Phys. Rev. E.,
83(031124), 2011.

[18] J. V. José & E. J. Saletan. Classical Dynamics a contemporary approach. Cambridge University
Press, 1998.

[19] W. P. Schleich. Quantum Optics in Phase Space. Wiley-VCH., Germany, 2001.

[20] G. Benenti, G. G. Carlo and T. Prosen. Wigner separability entropy and complexity of quan-
tum dynamics. Phys. Rev. E., 85(051129), 2012.

[21] P. D. Bergamasco, G. G. Carlo and A. M. F. Rivas. Quantum and classical complexity in

coupled maps. Phys. Rev. E., 96(062144), 2017.

[22] P. D. Bergamasco, G. G. Carlo and A. M. F. Rivas. Out-of-time ordered correlators, complexity,

and entropy in bipartite systems. Physical Review Research, 1(033044), 2019.

[23] J. Gong & P. Brumer. Intrinsic decoherence dynamics in smooth hamiltonian systems:
Quantum-classical correspondence. Phys. Rev. A., 68(022101), 2003.

[24] C. Miquel, J. Paz and M. Saraceno. Quantum computers in phase space. Phys. Rev. A.,
65(062309), 2002.

[25] S. Keppeler & M. Fysik. Introduction to Wigner-Weyl calculus. 2004.

[26] D.F. Walls & G. J. Milburn. Quantum Optics. Springer, Germany, 2008.

[27] W. K. Wootters. A wigner function formulation of finite state quantum mechanics. Annals of
Physics, 176:1–21, 1987.

[28] W. B. Case. Wigner functions and weyl transforms for pedestrians. Am. J. Phys., 76(937),
2008.

[29] R. L. Hudson. When is the wigner quasi-probability non-negative? Reports On Mathematical

Physics, 6(2), 1974.

[30] A. Kenfack & K. Życzkowski. Negativity of the wigner function as an indicator of non-
classicality. Nature, 6:396–404, 2004.

[31] I. I. Arkhipov, A. Barasiński and J. Svozilı́k. Negativity volume of the generalized wigner
function as an entanglement witness for hybrid bipartite states. Nature, 8(16955), 2018.

[32] L. E. Ballentine. Quantum Mechanics A modern development. World Scientific Publishing Co.
Pte. Ltd., Singapore, 2000.

[33] R. Cabrera, D. I. Bondar, K. Jacobs, and H. A. Rabitz. Efficient method to generate time
evolution of the wigner function for open quantum systems. Phys. Rev. A., 92(042122), 2015.
Bibliografı́a 87

[34] A. Ekert & P. L. Knight. Entangled quantum systems and the schmidt decomposition. Am.
J. Phys., 63(415), 1995.

[35] R. Simon. Peres horodecki separability criterion for continuous variables systems. Phys. Rev.
Lett., 84(12), 2000.

[36] B. Bergh & M. Gärttner. Entanglement detection in quantum many-body systems using
entropic uncertainty relations. Phys. Rev. A., 103(052412), 2021.

[37] W. H. Zurek, S. Habib, and J. P. Paz. Coherent states via decoherence. Phys. Rev. Lett.,
70(9), 1993.

[38] H. P. Robertson. The uncertainty principle. Phys. Rev., 34(163), 1929.

[39] L. Y. Chew & N. N. Chung. The quantum signature of chaos through the dynamics of
entanglement in classically regular and chaotic systems. Acta Physica Polonica A, 120(6-
A), 2011.

[40] M. Gutzwiller. The quantum mechanical toda lattice, ii. Annals of Physics, 133(2), 1981.

[41] B. Doyon. Generalized hydrodynamics of the classical toda system. Journal of Mathematical
Physics, 60(073302), 2019.

[42] M. Cusveller. Soliton computing in the toda lattice: controllable delay and logic gates, 2021.

[43] S. K. Gray, D. W. Noid and B. G. Sumpter. Symplectic integrators for large scale molecular
dynamics simulations: A comparison of several explicit methods. The Journal of Chemical
Physics, 101(4062), 1994.

[44] H. Yoshida. Construction of higher order symplectic integrators. Phys. Lett. A., 150(5,6,7),
1990.

[45] E. Ott. Chaos in Dynamical Systems. Cambridge University Press, 2002.

[46] H. Goldstein, C. Poole and J. Safko. Classical Mechanics. Addison Wesley, 2001.

[47] W. W. Ho and D. A. Abanin. Entanglement dynamics in quantum many-body systems. Phys.

Rev. B., 95(094302), 2017.

[48] A. Lerose & S. Pappalardi. Bridging entanglement dynamics and chaos in semiclassical systems.
Phys. Rev. A., 102(032404), 2020.

[49] J. N. Bandyopadhyay & A. Lakshminarayan. Testing statistical bounds on entanglement using

quantum chaos. Phys. Rev. Lett., 89(6), 2002.

[50] T. Nishioka. Entanglement entropy: Holography and renormalization group. Reviews of Mo-
dern Physics, 90, 2018.

[51] F. Xu, C. C. Martens and Y. Zheng. Entanglement dynamics with a trajectory-based formu-
lation. Phys. Rev. A., 96(022138), 2017.
 88 Bibliografı́a

[52] Z. Zhou J. Tan, Y. Luo and W. Hai. Combined effect of classical chaos and quantum resonance
on entanglement dynamics. Chin. Phys. Lett, 33(7), 2016.
[53] M. A. Nielsen & I. L. Chuang. Quantum Computation and Quantum Information. Cambridge
University Press, United Kingdom, 2000.
[54] L. E. Reichl. The transition to chaos: conservative classical systems and quantum manifesta-
tions. Springer, 2004.
[55] G. Casati and F. Haake. Encyclopedia of Condensed Matter Physics: Nonlinear Dynamics and
Nonlinear Dynamical Systems. Elsevier, 2005.
[56] F. Haake. Quantum Signatures of Chaos. Springer, 2010.
[57] P. Leboeuf & M. Saraceno. Eigenfunctions of non-integrable systems in generalised phase
spaces. J. Phys. A: Math. Gen., 23:1745–1764, 1990.
[58] P. A. Dando & T. S. Monteiro. Quantum surfaces of section for the diamagnetic hydrogen
atom: Husimi functions versus wigner functions. J. Phys. B: At. Mol. Opt. Phys, 27:2681–2692,
1994.
[59] H. S. Qureshi, S. Ullah and F. Ghafoor. Hierarchy of quantum correlations using a linear beam
splitter. Scientific Reports, 8(16288), 2018.
[60] N. L. Balazs & A. Voros. The quantized baker’s transformation. Annals of Physics, 190:1–31,
1989.
[61] A. Argüelles & T. Dittrich. Wigner function for discrete phase space: Exorcising ghost images.
Physica A: Statistical Mechanics and its Applications, 356(1):72–77, 2005.
[62] A. Argüelles. Construcción de la función de wigner para espacios de fases discretos, 2004.
[63] Y. A. Kravtsov & Y. I. Orlov. Caustics, Catastrophes and Wave Fields. Springer, 1993.
[64] J. P. Gazeau. Coherent States in Quantum Physics. Wiley-VCH, 2009.
[65] D. J. Tannor. Introduction to Quantum Mechanics. University Science Books, United Sates of
America, 2007.