Distorted stability pattern and chaotic features for quantized

prey-predator-like dynamics

A. E. Bernardini1, ∗ and O. Bertolami1, †

1
Departamento de Fı́sica e Astronomia,
Faculdade de Ciências da Universidade do Porto,
Rua do Campo Alegre 687, 4169-007, Porto, Portugal.
(Dated: March 20, 2023)
arXiv:2303.09622v1 [quant-ph] 16 Mar 2023

Abstract
Non-equilibrium and instability features of prey-predator-like systems associated to topologi-
cal quantum domains emerging from a quantum phase-space description are investigated in the
framework of the Weyl-Wigner quantum mechanics. Reporting about the generalized Wigner
flow for one-dimensional Hamiltonian systems, H(x, k), constrained by ∂ 2 H/∂x ∂k = 0, the prey-
predator dynamics driven by Lotka-Volterra (LV) equations is mapped onto the Heisenberg-Weyl
non-commutative algebra, [x, k] = i, where the canonical variables x and k are related to the
two-dimensional LV parameters, y = e−x and z = e−k . From the non-Liouvillian pattern driven
by the associated Wigner currents, hyperbolic equilibrium and stability parameters for the prey-
predator-like dynamics are then shown to be affected by quantum distortions over the classical
background, in correspondence with non-stationarity and non-Liouvillianity properties quantified
in terms of Wigner currents and Gaussian ensemble parameters. As an extension, considering the
hypothesis of discretizing the time parameter, non-hyperbolic bifurcation regimes are identified
and quantified in terms of z − y anisotropy and Gaussian parameters. The bifurcation diagrams
exhibit, for quantum regimes, chaotic patterns highly dependent on Gaussian localization. Besides
exemplifying a broad range of applications of the generalized Wigner information flow framework,
our results extend, from the continuous (hyperbolic regime) to discrete (chaotic regime) domains,
the procedure for quantifying the influence of quantum fluctuations over equilibrium and stability
scenarios of LV driven systems.

∗
Electronic address: [email protected]; On leave of absence from Departamento de Fı́sica, Universidade
Federal de São Carlos, PO Box 676, 13565-905, São Carlos, SP, Brasil.
†
Electronic address: [email protected]; Also at Centro de Fı́sica do Porto, Rua do Campo Alegre
687, 4169-007, Porto, Portugal.

1
I. INTRODUCTION

The Lotka-Volterra (LV) equations for prey-predator systems [1, 2] were originally pro-
posed as the setup for a classical deterministic description for the ecological equilibrium of
competitive populations. Due to their accessible mathematical properties, species coexis-
tence chains could be straightforwardly described by phase-space closed orbits parameterized
by a nonlinear Hamiltonian expressed in terms of the dimensionless one-dimensional x − k
phase-space by [3]
H(x, k) = a x + k + a e−x + e−k = , (1)

where x and k variables are correlated with the numbers of prey and predator species, y
and z, by y = e−x and z = e−k , respectively, and  is an arbitrary constant. On the
phenomenological perspective, while noticing that natural environments generally consist
of heterogeneous domains which affect the behavior of microscopic bio-systems, homoge-
neous prey-predator-like distributions in the domain space are just hypothetical. In Nature,
macroscopic and microscopic species are closely linked to their environment conditions,
which define growth directives to the categories of populations. Equivalent rate of growth
of species can coexist with different strategies related to additional degrees of freedom, even
if, for instance, the environment is homogeneous. Such aspects, once expressed in terms
of phenomenological parameters, can turn the stable coexistence of species into unstable
scenarios.
Generically, a stable classical setup as described by the Hamiltonian Eq. (1) can be
modified in order to encompass more complex prey-predator-like configurations which in-
clude, for instance, extinction and revival mechanisms [3, 4], perpetual coexistence [5],
competition-induced chaos [6–8] and microscopic molecular dynamics for symbiotic syn-
chronization [9, 10].
Specifically for microscopic systems, mechanisms that drive both quantum fluctuations
and non-linear effects may be admitted, even if a theoretical connection with crude bio-
chemical and biological evolutionary modeling is still unveiled. In fact, prey-predator-like
oscillations, competition-induced chaos, and symbiotic synchronization are examples of such
microscopic behavior which have been identified experimentally [6, 8–11]. They can be
regarded as a motivation for considering the quantization of LV systems and possibly to
respond how and at which scales classical macroscopic and quantum microscopic evolution

2
coexist, and if quantum effects arise at measurable scales. The answer for these questions
can also be relevant, for instance, in the investigation of the above quoted stability crite-
ria for microbiological communities [11, 12], or even in the analysis of stochastic system
dynamics [13–15] and in the description of phase transitions in finite microscopic systems
[16]. Therefore, despite the classical background driven by LV equations, the inclusion of
both quantum-like and instability triggers must be considered, in particular, in the scope of
equilibrium and stability analysis
For macroscopic (non-quantized) ecosystems [4, 5, 17, 18], a quantitative description,
either in terms of the hyperbolic equilibrium description [19–21], for a presumed continuous
variable dynamics, or in terms of chaos pattern classification, by the Hopf bifurcation analysis
(if the predator-prey dynamics is not covered by the hyperbolic equilibrium regime) is already
admitted. Otherwise, for microscopic organisms or bio-systems, the understanding of the
transition between classical and quantum regimes demands for the inclusion of systematic
and operative procedures which are still incipient [3, 22, 23].
In the framework here considered, non-equilibrium and instability properties are associ-
ated to topological quantum domains which emerge from a quantum phase-space descrip-
tion of the prey-predator-like dynamics [22]. Generically, for one-dimensional Hamiltonian
systems (cf. Eq. (1)), quantum features associated to phase-space patterns, stationary be-
havior, and information fluxes are straightforwardly quantified in terms of Wigner currents
and related properties [23–29].
The classical dynamics is given by the equations of motion obtained from Hamiltonian
Eq. (1), i.e.

dx/dτ = {x, H}P B = 1 − e−k , (2)

dk/dτ = {k, H}P B = e−x − 1, (3)

where τ is the dimensionless time, and the prey-predator z − y system mapped into z 7→
e−k and y 7→ e−x can be exactly evaluated in terms of phase-space coordinates through
the Weyl-Wigner (WW) framework [3]. This procedure allows for an effective quantum
description for the system [22]. Equilibrium and stability conditions affected by quantum
distortions over the classical background can then be theoretically connected to stationarity
and Liouvillianity drivers [23–29]. These quantum features can then be quantified in terms
of Wigner currents and ensemble parameters.

3
 Conceptually, the collective behavior depicted from phase-space effective quantum tra-
jectories are interpreted as averaged-out results of the space-time evolution of the species
distributions. Quantum deviations from classical patterns and their effects on the time
evolution of the prey-predator number of species can then be evaluated. In fact, the WW
framework allows for identifying how classical and quantum evolution coexists at different
scales and how prey-predator quantum analogue effects emerge [3, 30].
Considering that the prey-predator equilibrium regime is driven by an autonomous system
of ordinary differential equations, our analysis here is focused on verifying if the conditions for
equilibrium and stability criteria are met. As it will be verified, the quantum distortions over
the classical pattern, once convoluted by Gaussian ensembles [3], produce evident hyperbolic
and non-hyperbolic equilibrium and instability patterns which can all be quantified with
tools of the WW framework.
Having set the goals of our work, the outline of the manuscript is as follows. Sec. II is
concerned with the foundations of the generalized WW framework [22], which result in the
quantum driven phase-space trajectories given in terms of Wigner currents. Stationarity
and Liouvillianity quantifying operators obtained in the context of the extended Wigner
framework for non-linear Hamiltonians, H(x, k) = K(x) + V(k), are recovered [22] and the
mathematical structure for obtaining the corresponding statistically convoluted Gaussian
ensemble exact solutions is re-examined. Non-equilibrium and instability features of prey-
predator-like competition systems associated to topological quantum domains emerging from
such a quantum phase-space description are investigated in Sec. III. Since the equilibrium
regime is generated by an autonomous system of ordinary differential equations from the
WW phase-space, the quantitative correspondence of stationarity and Liouvillianity with
stability properties in terms of hyperbolic equilibrium parameters is discussed. Equilib-
rium and stability quantifiers can thus be obtained for quantum Gaussian ensembles when
they are dynamically driven by the corresponding Winger currents. Furthermore, emergent
topological phases due to the diffusive appearance of unstable vortices and saddle-points in
the phase-space, as an effect of quantum distortions over the classical background, are also
identified. Finally, considering the hypothesis of discretizing the time parameter, beyond
the hyperbolic regime, non-hyperbolic bifurcation patterns are identified in Sec. IV, being
described in terms of species anisotropy and Gaussian localization parameters. Our conclu-
sions are drawn in Sec. V, where the bridges between classical and quantum descriptions of

4
prey-predator-like systems are evinced, and a consistent interpretation on the meaning of
quantum distortions is provided.

II. QUANTUM DRIVEN PHASE-SPACE TRAJECTORIES

The WW phase-space framework [24–26] encompasses all the quantum features (see the
Appendix I) through a quasi-probability distribution function of canonical coordinates of
position, x, and momentum, k, through the so-called Wigner function, W(x, k). Since
it is associated with the quantum density matrix operator, ρ̂ = |ψihψ|, through its Weyl
transform one has
Z +∞
−1
ρ̂ → W(x, k) = π dw exp [2 i k w] ψ(x − w) ψ ∗ (x + w), (4)
−∞

written in a dimensionless form, where the Planck constant, ~, has been set equal to 1.
Correspondently, the probability flux evolves according to the continuity equation written
as [23, 25–27, 29]
∂τ W + ∂x Jx + ∂k Jk = ∂τ W + ∇ξ · J = 0, (5)

where the time, τ , is also dimensionless. For generic Hamiltonians in the form of H(x, k) =
K(k) + V(x), the associated Wigner currents are straightforwardly given by [22]
∞  2η
X i 1  2η+1
∂k K(k) ∂x2η W(x, k; τ ),

Jx (x, k; τ ) = + (6)
η=0
2 (2η + 1)!
∞  2η
X i 1
∂x V(x) ∂k2η W(x, k; τ ),
 2η+1 
Jk (x, k; τ ) = − (7)
η=0
2 (2η + 1)!

so to account for the quantum back reaction through the η > 1 contributions from the
corresponding series expansions. Once truncated at η = 0, currents Eqs. (6) and (7) yield
the classical Liouvillian regime [25, 26].
For phase-space coordinates identified by ξ = ξx x̂+ξk k̂, stationarity and Liouvillianity can
be locally quantified by divergence operators given by, ∇ξ ·J and ∇ξ ·w, respectively [23, 27,
29]. These are mutually connected by a parametric definition of a quantum-analog velocity,
w, implicitly given in terms of the vector currents, J = w W 1 ,which has the classical limit,
w → vξ(C) ,consistently described by the vector velocity vξ(C) = ξ̇ = (ẋ, k̇) ≡ (∂k H, −∂x H).

1
From which it can be noticed that ∇ξ · J = W ∇ξ · w + w · ∇ξ W.

5
From those definitions, the associated stationarity and Liouvillianity divergence operators
are given respectively by
∞
X (−1)η
∂x V(x) ∂k2η+1 W − ∂k2η+1 K(k) ∂x2η+1 W ,
 2η+1   
∇ξ · J = 2η
(8)
η=0
2 (2η + 1)!

and
∞
(−1)η
    
X  2η+1  1 2η  2η+1  1 2η
∇ξ · w = ∂k K(k) ∂x ∂ W − ∂x V(x) ∂k ∂ W ,
η=0
22η (2η + 1)! W x W k
(9)
from which, stationary and classical patterns are recovered by ∇ξ · J = 0 and ∇ξ · w = 0,
respectively [27].
From this point, replacing the Wigner distribution, W(x, k; τ ), by a Gaussian one2 ,

α2
exp −α2 x2 + k 2 ,


Gα (x(τ ), k(τ )) ≡ Gα (x, k) = (10)
π

into Eqs. (6) and (7), allows for writing [22]

∂χ2η+1 Gα (x, k) = (−1)2η+1 α2η+1 h2η+1 (αχ) Gα (x, k), for χ = x, k, (11)

where hn (αχ) are the Hermite polynomials of order n.

Also considering the auxiliary derivatives for K(k) and V(x) obtained from Eq. (1) as

∂x2η+1 K(k) = δη0 − e−k , (12)

∂k2η+1 V(x) = a δη0 − e−x ,


(13)

the Gaussian convoluted prey-predator dynamics can thus be described in terms of Wigner
currents written in the form of [3],
h
α2 i
∂x Jxα (x, k) = −2 α2 x − sin α2 x e 4 −k Gα (x, k), (14)
h
α2 i
∂k Jkα (x, k) = +2a α2 k − sin α2 k e 4 −x Gα (x, k), (15)

2
For the classical to quantum transition viewed by the statistical point of view, the classical analog of the
second-order moments of position and momentum coordinates, which define the Heisenberg uncertainty
principle, should consistently satisfy the same constraints of position and momentum quantum observables.
This procedure is isomorphic to the implementation of Gaussian ensembles (or states) [25, 36] in the sense
that, in both cases, the same statistics is reproduced.

6
which correspond to the convergent expression for the series expansions, Eqs. (6) and (7),
since the convergent contributions
∞
X s2η+1
h2η+1 (αχ) = sinh(2s αχ) exp[−s2 ] (16)
η=0
(2η + 1)!

have been found [3, 22]. Finally, from the above results, and after a straightforward inte-
gration, the quantum analog Wigner velocity components are given by
√
α i π −(k−α2 x2 )
wx (x, k) = 1 − e {Erf [α(x − i/2)] − Erf [α(x + i/2)]} , (17)
2α √ 
α i π −(x−α2 k2 )
wk (x, k) = −a 1 − e {Erf [α(k − i/2)] − Erf [α(k + i/2)]} , (18)
2α
written in terms of Gaussian error functions, Erf[. . . ].
The above results for the Wigner flow pattern can all be constrained by the form of J α =
w Gα (x, k), with w = (wαx , wαk ), which, besides yielding analytical results for components
of the Wigner current, J , bring up the topological properties of the Wigner flow and their
consequences to the prey-predator-like dynamics, as will be discussed in the following.

A. Results for Gaussian ensembles

The phase-space prey-predator Gaussian ensemble Wigner flow pattern evaluated in terms
of the Gaussian parameter, α, from α2 = 1/2 to 10 is depicted in Fig. 1. The plots are for
the stationarity quantifier (first column), ∇ξ · J , for the Wigner flux (second column),
J , and for the Liouvillian quantifier (third column), ∇ξ · w, which is superposed by the
normalized quantum velocity field (red arrows), w/|w|. In a kind of diffusive behavior,
discretized quantum domains emerge for (highly) peaked Gaussian envelops, which in Fig. 1
is evaluated from α2 = 10  1 (first row) to α2 = 1/2 & 0 (last row). Quantum features
are indeed suppressed for smoothly peaked Gaussian envelops. For α2 . 1, the quasi-
stationary quantum pattern approaches the classical regime, and the quantum effects are
seen as a small displacement of the equilibrium coordinates, (x, k) ∼ (0, 0), for quasi-closed
orbits. The greenyellow color scheme qualitatively depicts the intensity of the Wigner current
from maximal (yellow) to minimum (green) values. In fact, it shows that the probability
flux pushes the equilibrium points away from (x, k) ∼ (0, 0) as α increases. In both first
and second columns, additional green and orange contour lines are respectively evinced for
Jx = 0 and Jk = 0, as to depict the boundaries for the reversal of the Wigner flow in

7
the x and k direction, respectively. For increasing values of α, which correspond to more
peaked Gaussian distributions (from lower to upper rows), the quantum distortions emerge
as (anti)clockwise vortices (with winding number equals to (−) + 1), as well as separatrix
intersections and saddle points (with winding number equals to 0), i.e. flow stagnation points
(2) (2)
which are all identified by orange-blue crossing lines, where Jx = Jk = 0. The contra-flux
fringes (bounded by blue/orange lines) destroy the closed orbit patterns since they emerge to
compensate the retarded evolution of the quantum flux, and introduce discretized quantum
domains. These can be identified by the Liouvillian quantifier (third column), which clearly
depicts Wigner quantum distorted velocities (red arrows), w. The Wigner function and
current patterns define a kind of phase-transition driven by α as the definitive imprint of
the classical to quantum transition. The phase-transition is identified by the emergence of
(2) (2)
multiple stagnation points, with Jx = Jk = 0, for multiple quantum domains connected
by the topological features of the Wigner flow.

III. HYPERBOLIC STABILITY

The stagnation points which are identified in the phase-space Wigner flow (cf. Fig. 1)
can be evaluated in terms of the Gaussian parameter, α, which implicitly accounts for the
contributions due to the quantum fluctuations over the classical background, according to
the magnitude of the Gaussian localization, which, for α = 0, reproduces the classical limit
of Eqs. (17) and (18). The effects depicted in Fig. 1, which emerge from increasing α values,
can be identified through the phase-space evolution of additional quantum mechanically
emerging equilibrium points which replace the unique classical equilibrium point, (x, k) ∼
(0, 0). The rows of in Fig. 2 depict three different frame views for the behavior of the
phase-space stagnation points in terms of α, for 0 . α . 3.2.
The equilibrium point (flux) surrounding envelop is defined by |w| < 0.006, and the
plots are exhibited for a = 4 (first row), a = 1 (second row) and a = 1/4 (third row).
As it will be explained in the following, according to the so-called hyperbolic equilibrium
criteria, for 0 < α . 1.825, the anisotropy parameter, a, drives the stability regime for the
attractor focus and node points. For a > 1, one has stable regimes, as it appears for the
convergent behavior of the equilibrium points for a = 4 (first row). Conversely, for a < 1,
one has unstable regimes, as it appears for the diffusive behavior of the equilibrium points

8
FIG. 1: (Color online) First column: Stationarity quantifier, ∇ξ · J for the Gaussian ensemble. The color
scheme depicts quasi-stable (darker regions) and unstable (lighter) regions. Increasing values of α localizes
the quantum distortions, which result into evinced non-stationarity (unstable) domains. Blue and orange
contour lines are for Jx = 0 and Jk = 0, respectively. Second column: Corresponding Wigner flux, J . The
vector plot scheme is modulated by |J |1/4 and the greenyellow color scheme depicts the intensity of the
Wigner current from maximal (yellow) to minimum (green). The quantum critical points are identified by
(2) (2)
orange-blue crossing lines, for Jx = Jk = 0. Third column: Liouvillianity quantifier, ∇ξ · w, superposed
by the normalized quantum velocity field (red arrows), w/|w|. The divergence values, ∇ξ · w, are depicted
through the background color scheme, from darker regions, ∇ξ · w ∼ 0, to lighter regions, ∇ξ · w > 0. All
the results are for α2 = 10 (first row), 8, 6, 4, 2, 1 and 1/2 (last row) and for the x (horizontal axis) and
k (vertical axis) isotropic case (a = 1), with k, x ∈ [−3/2, +3/2] (unnecessary print values are suppressed
from the axis as to have clearer visual effect). Classical trajectories are shown as a collection of background
black lines.
9
a)

b)

c)
FIG. 2: (Color online) Region plot scheme for the phase-space evolution of stagnation points in the phase-
space, x − k. The plots show how the attractors (blue regions) are affected by the magnitude of the Gaussian
spreading parameter α, for 0 . α . 3.2. The equilibrium point (flux) surrounding envelop is defined by
|w| < 0.006, and the plots are exhibited for: a) a = 4 (first row), b) a = 1 (second row) and c) a = 1/4
(third row). for 0 < α . 1.825, the anisotropy parameter, a, drives the stability regime for attractor focus
and node points. For a > 1 one shall have stable regimes, as it appears for the convergent behavior of the
equilibrium points for a = 4 (first row). Conversely, for a < 1 one shall unstable regimes, as it appears for
the diffusive behavior of the equilibrium points for a = 1/4 (third row). Results are for the Wigner flow
in correspondence with Fig. 1. Local effects compensate each other when sliced views of the Wigner flux
for fixed α are considered, i.e. either when two vortices of opposite winding numbers match each other or
when saddle points (white to red regions) mutually annihilates each other. The portraits are the same for
different angle views.

10
for a = 1/4 (third row). The spreading behavior of the Gaussian ensemble, which runs from
red bubble saddle-point islands to the blue enveloped focus, corresponding to decreasing
values of α, recovers a quasi classical pattern, for which the quantum imprint is found only
for the small displacement (from (x, k) ∼ (0, 0)) of the attractor equilibrium points. The
blue region corresponds to the hyperbolic equilibrium regime, and can be described, for
instance, through a perturbative analysis [30].
In fact, for clarifying the above statements, one should turn attention to the definition
of the equilibrium points of a two-dimensional dynamical system. In the context of the
classical phase-space dynamics, they correspond to the solutions of a system of ordinary
differential equations with stationary behavior. Therefore, the equilibrium is geometrically
defined by ξ̇ = 0 (i.e. vx(C) = vk(C) = 0), whichhas a straightforward quantum correspon-
dence expressed in terms of the quantum velocity, w, by wx = wk = 0. The equilibrium
point solutions, in both classical and quantum descriptions, correspond to the phase-space
stagnation points. For a continuous system reduced to an equivalent discretized iterative
system, they correspond to the so-called fixed points of the equation system. Considering
the effective quantum dynamics described by the behavior of the phase-space velocities as

wx = Jx (x, k; t)/W(x, k, ; t) ≡ f (x, k),

wk = Jx (x, k; t)/W(x, k, ; t) ≡ g(x, k), (19)

the so-called hyperbolic stability of the equilibrium points is established by the features of
the Jacobian matrix identified by
 
∂x f (x, k) ∂k f (x, k)
j(x, k) =  , (20)
∂x g(x, k) ∂k g(x, k)

which defines an approximated linear stability by means of its eigenvalues so that equilibrium
and stability conditions can be naturally stratified into subclassifications, through the trace,
T r[. . . ], and the determinant, Det[. . . ], of j(x, k), when all derivatives are evaluated at the
equilibrium point, ξ e = (xe , ke ), obtained from f (xe , ke ) = g(xe , ke ) = 0.
To summarize, j(xe , ke ) has all the eigenvalues with negative real parts for asymptotically
stable systems. For unstable systems, at least one eigenvalue of j(xe , ke ) has a positive real
part. Likewise, the Jacobian matrix establishes the conditions for the so-called hyperbolic
equilibrium if all their corresponding eigenvalues have non-zero real parts. Finally, if at

11
least one eigenvalue of the Jacobian matrix at equilibrium points, j(xe , ke ), has a zero
real part, then the equilibrium is not hyperbolic, and the robustness of equilibrium and
stability conditions require another criterion [20]. Topologically, through the observation
of the vector field distribution, w(x, k), atξ e = (xe , ke ), the hyperbolic equilibrium and
stability features are as follows: saddle points, which correspond to unstable configurations,
for real eigenvalues with opposite signs; divergent (unstable) nodes, for both real eingenvalues
with positive signs; convergent (stable) nodes, for both real eingenvalues with negative
signs; divergent (unstable) focus, for both complex eingenvalues with positive real parts;
and convergent (stable) focus, for both complex eingenvalues with negative real parts.
According to the above criteria, one notices that focus and node stabilities are defined
by trace properties as

T r[j(xe , ke )] = ∇ξ · w|ξ > 0 → instability,

e

T r[j(xe , ke )] = ∇ξ · w|ξ < 0 → stability, (21)

e

when

Det[j(xe , ke )] = ∂x f |ξ ∂k g|ξ − ∂k f |ξ ∂x g|ξ > 0 → for focus and nodes, (22)

e e e e

and

Det[j(xe , ke )] = ∂x f |ξ ∂k g|ξ − ∂k f |ξ ∂x g|ξ < 0 → for saddle points. (23)

e e e e

Focus and nodes are also separated by the threshold value, ∆[j] = T r[j]2 − 4Det[j] = 0,
with

∆[j(xe , ke )] > 0 for nodes,

∆[j(xe , ke )] < 0 for focus. (24)

Since the hyperbolic equilibrium admits small linear perturbations over the system of
equations, the phase-space portrait qualitatively does not deviate from the steady state
configuration. Hence, the local phase portrait of a nonlinear system can be perturbatively
mapped by its linearized version, which equivalently accounts for eventual short displace-
ments of the fixed points (cf. theHartman-Grobman theorem [19]). Conversely, several types
of non-hyperbolic equilibrium patterns result into local bifurcations, which may change sta-
bility, suppress the fixed point features, or even split them into several equilibrium points,
as it is qualitatively depicted in Fig. 2.

12
 Turning back to the results from Fig. 2, which correspond to Eqs. (17) and (18), the
equilibrium point classifications are covered by the hyperbolic equilibrium criterium only
for α . 1.825, where short displacements from classical configurations (α ∼ 0) are evinced
by the blue region. Despite approaching classical-like closed orbits, they are perturbed by a
quantum vortex distortion which emerges from surrounding values of x and k and destroys
the (classical) closed orbit pattern. These features can be noticed from the Poincaré maps
for time intervals set as a quarter the orbit period, T /4, as depicted in Fig. 3 for α = 0 (first
row), α = 1/2 (second row), and α = 3/4 (third row), all under the hyperbolic regime. The
Poincaré maps of any continuous dynamical system show successive points of the phase-
space trajectory computed by means of the integration of the original continuous system.
From Fig. 3, it can be noticed that the relevant deviations from the closed orbit regime
emerge due to tiny anisotropic perturbations, which were introduced by setting a = 1.01
(first column) and a = 0.99 (third column). For the isotropic configuration, with a = 1,
since T r[j(xe , ke )] = 0 (cf. Eqs. (17) and (18)), classical (α = 0) and quantum (α 6= 0)
closed orbits are not affected. Correspondently, stable and unstable attractor behaviors are
noticed from the plots for a = 1.01 > 1 (first column) and a = 0.99 < 1 (third column),
respectively.
The complete hyperbolic equilibrium pattern is summarized by the results from Fig. 4:
phase-space saddle points are identified by α & 1.825, obtained from Det[j(xe , ke )] < 0; fo-
cus and node points are identified by α . 1.825, from Det[j(xe , ke )] > 0; and the attractor
regimes, for ∆[j(xe , ke )] > 0, are all defined and identified in terms of the anisotropy factor,
a, which also sets the threshold for close orbit stability at a = 1. For increasing values of
α, the saddle-points which emerge from the lighter white patterns indicated in Fig. 2 natu-
rally contributes to the subsequent diffusive appearance of unstable vortices and additional
saddle-points that completely annihilate the classical pattern, where the red bubble regions
correspond to the quantum drivers for the pattern depicted in the first rows of Fig. 1. How-
ever, in this case, the hyperbolic equilibrium classification corresponds to an extrapolation
from the perturbative regime.
Interestingly, out of the continuous hyperbolic equilibrium domain, the multifocal vortex
and saddle-point patterns of the quantum regime depicted by Fig. 2 have a chaotic counter-
part in the discrete domain. If discreteness is admitted for the driver of prey-predator oscil-
lation evolution, turning the LV system into a iterative discrete map, such chaotic features

13
 ��� α=� ��� α=� ��� α=�

��� ��� ���

��� ��� ���

��� ��� ���

�(�)

�(�)

�(�)
��� ��� ���

��� ��� ���

��� ��� ���

a) ������ ��� ��� ��� ��� ��� ��� ���

���
��� ��� ��� ��� ��� ��� ��� ���
���
��� ��� ��� ��� ��� ��� ��� ���
�(�) �(�) �(�)
��� α = �/� ��� α = �/� ��� α = �/�

��� ��� ���

��� ��� ���

��� ��� ���

�(�)

�(�)

�(�)
��� ��� ���

��� ��� ���

��� ��� ���

b) ���
��� ��� ��� ��� ��� ��� ��� ���
���
��� ��� ��� ��� ��� ��� ��� ���
���
��� ��� ��� ��� ��� ��� ��� ���
�(�) �(�) �(�)
��� α = �/� ��� α = �/� ��� α = �/�

��� ��� ���

��� ��� ���

��� ��� ���

�(�)

�(�)

�(�)
��� ��� ���

��� ��� ���

��� ��� ���

c) ������ ��� ��� ��� ��� ��� ��� ���

���
��� ��� ��� ��� ��� ��� ��� ���
���
��� ��� ��� ��� ��� ��� ��� ���
�(�) �(�) �(�)
��� α=� ��� α=� ��� α=�

��� ��� ���

��� ��� ���

��� ��� ���

�(�)

�(�)

�(�)

��� ��� ���

��� ��� ���

��� ��� ���

d) ������ ��� ��� ��� ��� ��� ��� ���

���
��� ��� ��� ��� ��� ��� ��� ���
���
��� ��� ��� ��� ��� ��� ��� ���
�(�) �(�) �(�)

FIG. 3: (Color online) Poincaré maps for: a) α = 0 (first row); b) α = 1/2 (second row); c) α = 3/4 (third
row); d) α = 1 (forth row), and for a = 1.01 (first column), a = 1 (second column) and a = 0.99 (third
column).

emerge from a single bifurcation pattern of the classical background, which is chaotically
affected by quantum fluctuations, as it shall be discussed in the next section.

IV. DISCRETIZED SYSTEMS – BIFURCATION MAPS AND CHAOTIC FEA-

TURES

Deterministic chaos is associated to non-linearity in Hamiltonian systems and its pattern

emerges when a dynamical system depends, in a sensitive way, on its previous state. Cir-
cumstantially, it is manifested by the discretization of the time evolving coordinate. The

14
FIG. 4: (Color online) Extrapolated hyperbolic pattern for effective quantum prey-predator system from
Eqs. (17) and (18) as function of the α and the anisotropy parameter a.

minimum complexity of a chaotic system can be described by the bifurcation theory, through
which changes onto the topological structure of a dynamical system can be addressed [19–21].
That is the case of the phase-space prey-predator-like dynamical system here investigated,
when one considers that time evolves through discretized steps.
According to the established theoretical grounds [19–21], a local bifurcation occurs when
a parameter change causes the stability of an equilibrium (or fixed point) to change. In
continuous systems, it is driven by the real part of an eigenvalue of an equilibrium configu-

15
ration passing through zero. It results into the bifurcation point, for which the equilibrium
is non-hyperbolic. In a discrete version of both the classical and the quantum modified
Gaussian convoluted LV equations, one can read {x(t), k(t)} 7→ {xn , kn } for t 7→ n/∆,
with ∆ arbitrarily large. Considering the results obtained in the previous section, and the
natural deviations from the hyperbolic domain, Fig. 5 depicts the bifurcation map for the
prey-predator classical pattern as function of the anisotropy parameter a.

� �

�=ⅇ -�� �=ⅇ -�

� �
�� �

� �

-� -�

-� -�
��� ��� ��� ��� ��� ��� ��� ��� ��� ���
a) � b) �

FIG. 5: (Color online) a) (Left) Bifurcation pattern for the discrete classical evolution of the mapped
prey-predator chain, x and k, as function of the anisotropy localization parameter a. Red points are for
the evolution of the equilibrium points for the x coordinate, and blue points are for the evolution of the
equilibrium points for the k coordinate. Results are for t 7→ n/∆ with ∆ = 5000, and n = 1, 2, . . . , 5 × 105 .
b) (Right) Once mapped into the prey-predator scheme,z 7→ e−k and y 7→ e−x , the discretized pattern
introduces the possibility of the prey(predator) extinction hypothesis, as the blue curve approaches zero for
increasing values of a(> 1).

One notices that the bifurcation emerge for a = 1 such that for a > 1 a two-fold degener-
acy of the equilibrium point can be identified. Accordingly, under the influence of quantum
fluctuations driven by the Gaussian parameter α, the bifurcation at a = 1 can be correlated
with the results for the Poincaré map from Fig. 3, as depicted by Fig. 6. For increasing
values of α, the behavior of the stability points for the anisotropy parameter a = 0.99 (< 1)
and a = 1.01 (> 1) exhibit a two-dimensional chaotic pattern which can be indistinctly ex-
tended for any value of a > 1 and a < 1, where disorder increases with increasing values of
α. Conversely, for α . 1, in the first and second rows of Fig. 6 (a . 1), just a tiny deviation
from the classical equilibrium point, i.e., (xe , ke ) ∼ (0, 0), is noticed. Likewise, in the third
and forth rows of Fig. 6 (a & 1), the classical bifurcation pattern is slightly affected by
quantum corrections at α . 1.
The topological changes in the phase-space portrait of the system define the classification
of bifurcations, which by itself is extensively analyzed in the literature [19–21] and is out of
the scope of such a preliminary analysis. From the disorder pattern numerically obtained,

16
it can be noticed that increasing values of α contribute to “accelerate” the bifurcation
pattern with respect to the anisotropy parameter, a, where the relative a-distance between
successive bifurcations decreases as it is favored by the introduction of quantum distortions.
From such results, one can realize that, for the prey-predator-like dynamics, the chaotic
behavior is increasingly induced by quantum distortions.

V. CONCLUSIONS

The Wigner flow framework for addressing the role of quantum fluctuations on the hyper-
bolic equilibrium configurations of prey-predator-like systems, as well as for understanding
their relations with the triggers for chaotic patterns, has been evaluated. Firstly, in the
broad range context covered by the Wigner framework for non-linear Hamiltonians like
H = K(k) + V(x), a consistent map of hyperbolic stability conditions for prey-predator-like
systems was provided. Throughout the results here obtained, quantum fluctuations over the
prey-predator-like classical background has been shown to affect the pattern of hyperbolic
equilibrium of their quantum analogs. Conversely, the combination of the hyperbolic stabil-
ity quantifiers with the Wigner features is shown to be relevant in distinguishing quantum
fluctuations from non-linear effects when they cannot be systematically investigated through
the Schrödinger framework.
From a theoretical perspective, the results obtained in Sec. III could be read as the
macroscopic emergence of quantum distortions over the classical pattern. The quantum
analog non-commutative property related to phase-space coordinates, [x, k] 6= 0, is reflected
by the non-extinction hypothesis parameterized by minimal phase-space elementary unity,
δx δk ∼ 1. As a matter of fact, in quantum mechanics, position and momentum operators
have their non-commutative nature expressed by the Moyal star-product definition which,
by the way, recovers the WW formalism. If x and k measurements affect each other, the
quantum analog of the uncertainty principle is expressed by δx δk & 1 in correspondence with
[x, k] = i. Paradigmatically, if the LV maps the prey-predator species oscillation, it is real-
ized as a suppression of any deterministic species evolution parameterized either by y(τ ) and
z(τ ) or by y(z) ↔ z(y). However, the Wigner framework circumvents the deterministic solu-
tion by exhibiting a semiclassical (averaged-out) solution from which the quantum imprints
can be detected via the hyperbolic equilibrium pattern as well as onto Poincaré maps and bi-

17
FIG. 6: (Color online) Bifurcation maps for: a) a = 0.97; b) 0.99; c) 1.01; and d) 1.03; as function of the
Gaussian localization parameter α. . Red points are for the evolution of the equilibrium points for the x
coordinate, and blue points are for the evolution of the equilibrium points for the k coordinate. Increasing
values of α contribute to a complete disorder behavior which destroys classical prey-predator bifurcation
patterns. Results are for ∆t 7→  and t 7→ n with  = 5 × 10−4 , and n = 0, 1, 2, . . . .

18
furcation maps (for time discretization). From a non-deterministic perspective, competitive
species, in analogy with interacting quantum states [31], have their existence predictability
– which should be related to a measurement operation – associated to a quantum statisti-
cal ensemble description. In fact, that is the case of the prey-predator-like dynamics here
considered, where a Gaussian (single- or multi-particle) phase-space probability distribution
is considered. Such an analysis is consistent with the framework where several competitive
microbiological and biochemical systems [32, 33] face environmental effects [34] and com-
plex and self-organizing hierarchical mechanisms [35] included in their dynamics by means
of quantum-based treatments supported by fundamental single- and multi-particle quantum
mechanics.
Finally, when the above analysis is extrapolated to non-hyperbolic bifurcation regimes,
the values probed or approached asymptotically by prey-predator anisotropy and Gaussian
localization parameters, a and α, allowed for the complete visualization of the bifurcation
theory and the correspondence between classical and effective quantum regimes.
To conclude, our results show that the LV system, as a particular example of a broad
class of non-linear Hamiltonian systems, once admitted as a map of prey-predator dynamics
sensitive to quantum mechanical corrections and equilibrium theory analysis, exhibit mea-
surable patterns which, in the context of the WW phase-space framework, can be detected
either microscopically by the identification of quantum topological phases or macroscopi-
cally by time evolved averaged-out statistical imprints, aspects that, of course, may deserve
a fine-tuning phenomenological analysis.

Acknowledgments – The work of AEB is supported by the Brazilian Agencies FAPESP

(Grant No. 20/01976-5 and Grant No. 23/00392-8) and CNPq (Grant No. 301485/2022-4).

Appendix I – The WW quasi-probability distribution

The WW formalism is thought of as the bridge between operator methods and path
integral techniques [39–41] encoded by a Weyl transform operation defined by
Z +∞ Z +∞
W
O (q, p) = 2 ds exp [2 i p s/~] hq − s|Ô|q + si = 2 dr exp [−2 i q r/~] hp − r|Ô|p + ri,
−∞ −∞
(25)

19
where an arbitrary quantum operator is identified by Ô. For the case where Ô is identified
as a density matrix operator, ρ̂ = |ψihψ|, the Weyl transformed operator, OW (q, p), results
in the so-called Wigner quasi-probability distribution function,
Z +∞
−1 −1
h ρ̂ → W (q, p) = (π~) ds exp [2 i p s/~] ψ(q − s) ψ ∗ (q + s). (26)
−∞

This is interpreted as the Fourier transform of the off-diagonal elements of the associated
density matrix, ρ̂, where h = 2π~ is the Planck constant. As a mandatory constraint, it pre-
sumes a consistent probability distribution interpretation constrained by the normalization
condition over ρ̂, that is T r{q,p} [ρ̂] = 1.
Eq. (26) was proposed by Wigner’s seminal work [24] when accounting for quantum
corrections to TD equilibrium states. Without affecting the predictive power of QM, and
preserving the symmetries of the Heisenberg-Weyl group of translations, the Wigner phase-
space quasi-distribution function is therefore associated with a density operator ρ̂, in the
form of an overlap integral, Eq. (26). The reason for the quasi-distribution nomenclature is
due to the point that the Wigner function’s most elementary property is concerned with its
marginal distributions, which return position and momentum distributions upon integrations
over the momentum and position coordinates, respectively,
Z +∞ Z +∞
2 2
|ψ(q)| = dp W (q, p) ↔ |ϕ(p)| = dq W (q, p). (27)
−∞ −∞

In fact, strictly connected with the Hilbert space features of the Schrödinger QM, the Fourier
transform of the above addressed wave functions,
Z +∞
−1/2
ϕ(p) = (2π~) dq exp [i p q/~] ψ(q), (28)
−∞

is the property that suppresses the coexistence of positive-definite position and/or momen-
tum probability distributions.
Hence, the connection of the Wigner function to the matrix operator QM, through
Eqs. (25) and (26)), allows for computing the expected values of quantum observables de-
scribed by generic operators, Â, evaluated through an overlap integral over the infinite
volume described by the phase-space coordinates, q and p, as [24, 26]
Z +∞ Z +∞
hOi = dp dq W (q, p) AW (q, p). (29)
−∞ −∞
h i
This corresponds to the trace of the product between ρ̂ and Ô, T r{q,p} ρ̂Ô . In addition,
the statistical aspects evinced from the definition of W (q, p) admit extensions from pure

20
states to statistical mixtures, through which the replacement of OW (q, p) by W (q, p) into
Eq. (29) leads to the quantum purity computed through an analogous of the trace operation,
T r{q,p} [ρ̂2 ], i.e.
Z +∞ Z +∞
2
T r{q,p} [ρ̂ ] = 2π~ dp dq W (q, p)2 , (30)
−∞ −∞

which satisfies the pure state constraint, T r{q,p} [ρ̂2 ] = T r{q,p} [ρ̂] = 1.
Finally, flow properties of the Wigner function, W (q, p) → W (q, p; t), can be, for in-
stance, connected to the Hamiltonian dynamics. In this case, a vector flux [23, 27, 29],
J(q, p; t), decomposed into the phase-space coordinate directions, q̂ and p̂, as J = Jq q̂ +Jp p̂,
reproduces a flow field connected to the Wigner function dynamics through the continuity
equation [23, 25–27, 29],
∂t W + ∂q Jq + ∂p Jp = 0, (31)

where a shortened notation for partial derivatives, ∂a ≡ ∂/∂a, has been used.
To connect the framework with prey-predator dynamics, all the above quantities have
1/2
been more properly written in terms of dimensionless variables, x = (m ω ~−1 ) q and
k = (m ω ~)−1/2 p. In this case, one should have the dimensionless Hamiltonian H =
 
−1/2
(~ω)−1 H(q, p) = K(k) + V(x), with V(x) = (~ω)−1 V (q) = (~ω)−1 V (m ω ~−1 ) x
 
and K(k) = (~ω)−1 K (p) = (~ω)−1 K (m ω ~)1/2 k , where m is a mass scale parameter
and ω is an arbitrary angular frequency. Therefore the Wigner function can be cast into
the dimensionless form of W(x, k; ωt) ≡ ~ W (q, p; t), with ~ absorbed by dp dq → ~ dx dk
integrations,
Z +∞
−1
W(x, k; τ ) = π dw exp (2 i k w) ψ(x − w; τ ) ψ ∗ (x + w; τ ), (32)
−∞

1/2
with w = (m ω ~−1 ) s and τ = ωt, as it appears in Eq. (4) with τ suppressed from notation.

[1] A. J. Lotka, Elements of physical biology (Williams & Wilkins Co., Baltimore 1925).
[2] V. Volterra, Variazioni e fluttuazioni del numero d’individui in specie animali conviventi,
Mem. R. Accad. Naz. Lincei. (Ser. VI) 2, 31-113 (1926).
[3] A. E. Bernardini and O. Bertolami, Phys. Rev. E 106, 024202 (2022).
[4] M. Parker and A. Kamenev, Phys. Rev. E80, 021129 (2009).
[5] N. R. Smith and B. Meerson, Phys. Rev E93, 032109 (2016).
[6] D. P Paula et al., Molecular Ecology Resources 15, 880 (2015).

21
 [7] T. Fujii and T. Rondelez, ACS Nano 7, 27 (2013).
[8] I. R. Epstein and J. A. Pojman, An Introduction to Nonlinear Chemical Dynamics (Oxford
University Press, New York 1998).
[9] J. Ackermann, B. Wlotzka and J. S. McCaskill, Bull. Math. Biol. 60, 329 (1998).
[10] B. Wlotzka and J. S. McCaskill, Chem. Biol. 4, 25 (1997).
[11] M. Kumar, B. Ji, K. Zengler et al., Nat. Microbiol. 4, 1253 (2019).
[12] S. Butler and J. P. O’Dwyer, Nat Commun 9, 2970 (2018).
[13] J. Grasman and O. A. van Herwaarden, Asymptotic methods for the Fokker-Planck equation
and the exit problem in applications (Springer, Berlin 1999).
[14] L. J. S. Allen, An introduction to stochastic processes with applications to biology- 2nd Ed.
(New York, NY: Chapman & Hall-CRC 2010).
[15] T. Reichenbach, M. Mobilia and E. Frey, Phys. Rev. E 74, 051907 (2006)
[16] G. Szabo and T. Czaran, Phys. Rev. E63, 061904 (2001).
[17] T. Tahara et al., Sci Rep 8, 7029 (2018).
[18] Yi-An Ma and Hong Qian, Proceedings of the Royal Society A471, 20150456 (2015).
[19] D. M. Grobman, Homeomorphisms of systems of differential equations Doklady Akademii
Nauk SSSR, 128 880 (1959); P. Hartman, Proceedings of the American Mathematical Society
11 (4), 610 (1960).
[20] J. Guckenheimer and P. Holmes, Nonlinear Oscillations, Dynamical systems and Bifurcations
of Vector Fields (Springer-Verlag, New York 1983).
[21] Yu. A. Kuznetsov, Elements of Applied Bifurcation Theory (Springer 2004).
[22] A. E. Bernardini and O. Bertolami, Phys. Rev. A 105, 032207 (2022).
[23] O. Steuernagel, D. Kakofengitis and G. Ritter, Phys. Rev. Lett. 110, 030401 (2013).
[24] E. Wigner, Phys. Rev. 40 749 (1932).
[25] L. E. Ballentine, Quantum Mechanics: a Modern Development, pp. 633 (World Scientific,
Singapore 1998).
[26] W. B. Case, Am. J. Phys. 76, 937 (2008).
[27] A. E. Bernardini, Phys. Rev. A 98, 052128 (2018).
[28] W. H. Zurek, Phys. Today 44, 36 (1991).
[29] A. E. Bernardini and O. Bertolami, EPL 120, 20002 (2017); A. E. Bernardini and O. Berto-
lami, Journal of Physics: Conf. Series 1275, 012032 (2019).
[30] A. E. Bernardini and O. Bertolami, Quantum prey-predator dynamics: a Gaussian ensemble
analysis, arXiv:2209.02450 [quant-ph].
[31] R. Real, A. Márcia Barbosa and J. W. Bull, Syst. Biol. 66, 453 (2017).
[32] S. H. V. Leme de Mattos, J. R. C. Piqueira, J. Vasconcelos-Neto and F. M. Orsatti, Ecological
Modelling 200, 183 (2007).
[33] E. Del Giudice, R. M. Pulselli and E. Tiezzi, Ecological Modelling 220, 1874 (2009).
[34] A. D. Kirwan Jr., Entropy 10, 58 (2008).

22
[35] S. E. Jorgensen and E. Tiezzi, Ecological Modelling 220, 1855 (2009).
[36] S. D. Bartlett, T. Rudolph and R. W. Spekkens, Phys. Rev. A 86, 012103 (2012).
[37] I. S. Gradshteyn and I. Ryzhik, Tables of Integrals, Series and Products (Academic Press,
New York 1994).
[38] A. Vanselow, S. Wieczorek and U. Feudel, Journal of Theoretical Biology 479, 64 (2019).
[39] A. A. Abrikosov, L. P. Gorkov and I. E. Dzyaloshinskii, Quantum Field Theoretical Methods
in Statistical Physics, 2nd Ed. (Pergamon, Oxford 1965).
[40] L. S. Schulman, Techniques and Applications of Path Integral (Jonh Wiley & Sons Inc. 1981).
[41] G. Parisi, Statistical Field Theory (Benjamin/Cummings 1988).

23