A simple method to enlarge a basin of attraction using a memristive function

Alexandre R. Nieto, Rubén Capeáns, Miguel A.F. Sanjuán · 2025

Read A simple method to enlarge a basin of attraction using a memristive function by Alexandre R. Nieto, Rubén Capeáns, Miguel A.F. Sanjuán (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2501-12220.txt.

Introduction

Alexandre Nieto, Rubén Capeáns, and Miguel Sanjuán offered in 2025 a simple trick for keeping chaotic systems on their desired attractor: swap a fixed parameter for a memristive function. A memristor's behavior depends on its history, and here a smooth sigmoid-shaped version does the work inside the classic Hénon map.

The idea leans on neighboring attractors with wider basins. The memristive term lures wandering orbits toward the wide basin and hands them over to the target attractor. Simulations showed a substantial widening of the target's basin and sturdier stability. No explicit parameter steering was needed during the run. The authors pitch the method as a lightweight control for discrete chaotic systems. One component, swapped in, does the herding.

Catalog record

CREATORAlexandre R. Nieto, Rubén Capeáns, Miguel A.F. Sanjuán
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE15 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2501-12220

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

A simple method to enlarge a basin of attraction using a memristive function Alexandre R. Nieto, Rubén Capeáns, Miguel A.F. Sanjuán Abstract This study presents an innovative approach to chaotic attractor stabilization introducing a memristor in discrete dynamical systems. Using the Hénon map as a test case, we replace a system parameter with a memristive function governed by a sigmoid activation function. The method relies on leveraging attractors with larger basins of attraction to attract the orbits and guide them towards the desired chaotic attractor. The effectiveness of the method is confirmed through numerical simulations, showing substantial enhancement in attractor stability without requiring explicit parameter control. Keywords Numerical simulations; memristor; chaotic attractor; Hénon map; basin of attraction 2020 Mathematics Subject Classifications 37E30; 37M05; 65P40 I Introduction In 1971 , Prof. Chua theoretically predicted the existence of the memristor Chua71 —a resistor whose resistance depends on the amount of electric charge that has flowed through it. Almost 40 years later, in 2008 , Hewlett-Packard Labs demonstrated a nanoscale titanium dioxide memristor Strukov08 . In addition to their numerous technological applications in the field of nanoelectronics Pickett09 ; Brown23 , memristors have been extensively studied in the context of chaos theory. Beyond the many studies exploring its effects on systems of nonlinear ODEs, the impact of memristors on discrete chaotic systems has recently attracted considerable attention Wang22 ; Rong22 ; Peng20 . Some of these recent works have used paradigmatic maps, such as the Hénon map and the Lozi map, to demonstrate that the incorporation of a memristor can completely alter the system’s dynamics. A continuous model of a memristor can be defined as \displaystyle\begin{aligned} &v(t)=M(q(t))i(t),\\ &\dot{q}=i(t),\end{aligned} (1) where i(t) is the current, q(t) the charge, v(t) the voltage, and M(q(t)) the memristance. By discretizing the derivatives, a discrete memristor model can be expressed as \displaystyle\begin{aligned} &v_{n}=M(q_{n})i_{n},\\ &q_{n+1}=q_{n}+hi_{n},\end{aligned} (2) where h is the time step and the memristive function M(q_{n}) is often modeled as a sigmoid function. Building on this simple discrete memristor model, we present a method to enlarge the basin of attraction of a chaotic attractor by replacing a system parameter with a memristive function. For all numerical simulations, we have used the Hénon map as the model, demonstrating the method by enlarging the basin of attraction of the Hénon attractor. The structure of this paper is as follows. In Sec. II we explain the method. Numerical simulations using the Hénon map, which demonstrate how the basins of attraction are effectively enlarged, are presented in Sec. III . Finally, in Sec. IV , we summarize the key findings. II The method For this research we have used the Hénon map as a model, so we will use the same system to explain the method. The Hénon map is a paradigmatic two-dimensional map given by \displaystyle{\cal{H}}(a,b,c):\begin{cases}x_{n+1}=1+ax^{2}+cy_{n},\\ y_{n+1}=bx_{n},\\ \end{cases} (3) where a,b,c\in\mathbb{R} are parameters. In the canonical Hénon map c=1 , but we have included a free parameter for the purposes of this research. For parameter values a=1.4 , b=0.3 , and c=1 , the only attractive set of \cal{H} is a chaotic attractor \cal{C} . Thus, if we define an arbitrary phase space region \Omega\in[-10,10]\times[-50,50] , all initial conditions outside the basin of attraction of \cal{C} generate unbounded orbits, meaning that they leave \Omega . We show the chaotic attractor and its basin of attraction in Fig. 1 . The area of this basin in \Omega is \mu[{\cal{B}}({\cal{C}})]=21.8 . Numerically, the area of the basin is calculated as the fraction of initial conditions belonging to the basin, multiplied by the total area of \Omega . The objective of our method is to enlarge the basin of attraction shown in in Fig. 1 . If we achieve this, a wider range of initial conditions will generate orbits that converge to \cal{C} , so the stability of the chaotic attractor would be enhanced. Figure 1: In black, chaotic attractor \cal{C} of the Hénon map given by Eq. ( 3 ) with parameters a=1.4 , b=0.3 , and c=1 . In red, basin of attraction of \cal{C} . White color represents initial conditions that generate unbounded orbits that diverge to \pm\infty . For this purpose, we replace one of the parameters of \cal{H} by a memristive function M(z_{n}) , where z is the inner state of the memristor (charge in Eq. ( 2 )). To maintain simplicity, we modify only one parameter at a time, producing three variants of the memristive Hénon map given by \displaystyle{\cal{H}}_{a}(M,b,c):\begin{cases}x_{n+1}=1+M(z_{n})x^{2}+cy_{n},\\ y_{n+1}=bx_{n},\\ z_{n+1}=z_{n}+x_{n}.\end{cases} (4) \displaystyle{\cal{H}}_{b}(a,M,c):\begin{cases}x_{n+1}=1+ax^{2}+cy_{n},\\ y_{n+1}=M(z_{n})x_{n},\\ z_{n+1}=z_{n}+x_{n}.\end{cases} (5) \displaystyle{\cal{H}}_{c}(a,b,M):\begin{cases}x_{n+1}=1+ax^{2}+M(z_{n})y_{n},\\ y_{n+1}=bx_{n},\\ z_{n+1}=z_{n}+y_{n}.\end{cases} (6) In Eqs. ( 4 - 6 ), we define the memristive function as M(z_{n})=p_{u}+(p_{t}-p_{u})\varphi(z_{n}), (7) where p\in\{a,b,c\} , and the subscripts t and u stand for ‘target’ and ‘undesired’, respectively, indicating that the method aims for M(z_{n}) to converge to the target parameter p_{t} rather than the undesired parameter p_{u} . The activation function \varphi(z_{n}) is the sigmoid \varphi(z_{n})=\frac{1}{1+e^{-z_{n}/\tau}}, (8) where \tau determines the steepness of the curve. Since \varphi(z_{n})\in(0,1) , the structure of the memristive function given by Eq. (7) guarantees M(z_{n})\in(p_{u},p_{t}) . When z_{n} diverges to +\infty , M(z_{n}) converges to the target parameter p_{t} ( a_{t}=1.4 , b_{t}=0.3 , or c_{t}=1 in our case), leading to orbit stabilization in the target attractor \mathcal{A}(p_{t})=\mathcal{C} . Conversely, if z_{n} diverges to -\infty , M(z_{n}) converges to an undesired parameter p_{u} , resulting in orbit stabilization in the attractor \mathcal{A}(p_{u}) . The final convergence to either attractor depends on the accumulated value of x or y along the orbit. Therefore, the asymptotic behavior of the memristive map {\cal{H}}_{p} can be understood by simply observing the appropriate coordinate in a bifurcation diagram that represents \cal{A} as a function of a , b , or c . These diagrams are shown in Fig. 2 . The left y -axis represents the asymptotic value of the coordinate, while the right y -axis represents the area of the basin of attraction of the attractor {\cal{A}}(p) for each parameter value. The vertical green line indicates the position of the target attractor \cal{C} . Figure 2: Bifurcation diagrams illustrating the asymptotic behavior of the Hénon map as a function of the system parameters. (a) Bifurcation diagram with respect to parameter a , (b) with respect to parameter b , and (c) with respect to parameter c . The green vertical line in each plot indicates the position of the target chaotic attractor \mathcal{C} . Below each bifurcation diagram, the area of the basin of attraction, \mu[{\cal{B}}({\cal{A}})] , is shown as a function of the corresponding parameter p\in\{a,b,c\} . This lower curve represents how the size of the basin of attraction changes with the parameter value, showing characteristic peaks that indicate regions where the attractors have larger basins of attraction. For all three parameters, there exist attractors \mathcal{A}(p) with basins of attraction larger than that of \mathcal{C} . Our method exploits these larger basins through the following mechanism: 1. The initial value of the memristive function M(z) is the midpoint of the parameter range M(0)=\frac{p_{t}+p_{u}}{2}\equiv p_{0}. 2. If p_{m} is the parameter for which the basin of attraction of {\cal{A}}(p) is the largest (see absolute maxima in Fig. 2 ), our best choice for p_{u} is the one that satisfies p_{0}=p_{m} . Since \mu[\mathcal{B}(\mathcal{A}(p_{0}))]>\mu[\mathcal{B}(\mathcal{C})]

Read the complete text