Order-tunable chaotic dynamics, memory-state-dependent coexistence, and reservoir computing in a fractional memfractor-based oscillator
- 2 November 2026
- journal article
- Published by IOP Publishing in Physica Scripta
- Vol. 101 (44) , 445201
Abstract
In this paper, an autonomous oscillator utilizing a memfractor with fractional-order elements is presented and analyzed with regard to its order-dependent chaotic behavior, memory state-dependent coexistence, and reservoir computing properties. The design includes a Lorenz-like nonlinear element and a memfractor branch where the effective input is generated using the memory state along with a fractional-order accumulation state. The memfractor is first analyzed on an elemental basis through frequency-dependent pinched v–i hysteresis loops, which confirm the expected fading-memory signature. Thereafter, the generated oscillator is analyzed using equilibria, stability metrics, phase planes, bifurcations, finite-time Lyapunov analysis, and initial-condition scans. For the default parameter set, nine equilibrium points are detected, including two locally stable equilibria and seven unstable equilibria. Under the same parameters, the system exhibits coexistence between stable equilibrium convergence and bounded chaotic oscillations, mainly selected by the initial memfractor memory state. The memfractor coupling strength controls the transition toward chaos, while the fractional orders tune the intensity of memory-induced divergence. The numerical credibility of the reported regimes is supported by Grünwald–Letnikov and predict–evaluate–correct–evaluate / Adams–Bashforth–Moulton solver cross-validation, together with time-step, simulation-length, and memory-length sensitivity tests. Finally, the oscillator is evaluated as a memory-tunable reservoir on NARMA10 and NARMA20 tasks. In the reservoir-computing tests, the full fractional memfractor configuration gives the lowest representative NARMA10 and NARMA20 prediction errors. The ablation results further indicate that fractional dynamics and memfractor feedback contribute differently to temporal retention and nonlinear prediction. These results show that the fractional memfractor provides a compact mechanism for memory-controlled chaos, coexistence, and reservoir-performance tuning.Keywords
This publication has 36 references indexed in Scilit:
- Fractional order mem-elements: from modeling to applications : a reviewNonlinear Dynamics, 2026
- Exploiting Bistability and Viscoelasticity in Reservoir ComputingNonlinear Dynamics, 2025
- Models of nonlinear higher-order elements for autonomous fractional circuitsNonlinear Dynamics, 2025
- Extremely multi-stable grid-scroll memristive chaotic system with omni-directional extended attractors and application of weak signal detectionChaos, Solitons & Fractals, 2024
- Dynamic analysis and circuit design of tunable multi-vortex chaotic systems based on memristorsNonlinear Dynamics, 2024
- Coexisting hyperchaos and multistability in a discrete memristor-coupled bi-neuron modelNonlinear Dynamics, 2024
- Complex dynamics of a new multiscroll memristive neural networkNonlinear Dynamics, 2024
- Determining Lyapunov exponents of fractional-order systems: A general method based on memory principleChaos, Solitons & Fractals, 2023
- Bifurcation and chaos in time delayed fractional order chaotic memfractor oscillator and its sliding mode synchronization with uncertaintiesChaos, Solitons & Fractals, 2017
- The Grünwald–Letnikov Fractional-Order Derivative with Fixed Memory LengthMediterranean Journal of Mathematics, 2015