Subspace state-rescaling control for non-linear dynamical systems
This paper introduces a subspace-selective state-rescaling control law for suppressing pathological dynamics in complex neural systems, without relying on high-voltage additive forcing. Conventional neurostimulation superimposes external stimulation currents that override native network states, leading to severe energy inefficiency and side-effect propagation. The proposed law instead modulates the system’s native vector field, or discrete state map, through a diagonal gain matrix K whose entries are chosen model by model: gains associated with targeted pathological channels are set below unity, while the remaining entries are fixed at unity so that the corresponding state directions are left unscaled. This subspace-selective attenuation enables individual Floquet multipliers to be relocated inside the unit circle without globally slowing the entire system, a capability that scalar time reparameterization cannot provide. We establish formal Lyapunov and Floquet stability proofs together with explicit bifurcation criteria (Flip and Neimark–Sacker) for the resulting closed-loop dynamics, and we validate the control law across discrete maps (Logistic, Hénon, Delay Logistic, Rulkov) and continuous biophysical models (FitzHugh–Nagumo, Hindmarsh–Rose, Wendling Neural Mass, Epileptor). The resulting multi-scale validation establishes a foundation for low-energy, subspace-selective neuromorphic closed-loop execution.
Authors
- Sabir Jacquir (ORCID: https://orcid.org/0000-0002-6296-7888)
- Mounira Kesmia
Institutions
- University Frères Mentouri Constantine 1 (DZ)
- Centre National de la Recherche Scientifique (FR)
- Université Paris-Saclay (FR)
- Institut des Neurosciences Paris-Saclay (FR)
- Centre Universitaire de Mila (DZ)
Publication Details
- Journal
- Chaos Solitons & Fractals
- Published
- 2026-11-18
- DOI
- https://doi.org/10.1016/j.chaos.2026.119519
- Primary Topic
- Chaos control and synchronization
- Type
- article
- Field-Weighted Citation Impact
- 0.00