Noise-induced bistability of information-driven active agents in turbulence
Phys. Rev. E 114, 045403 – Published 16 October, 2026
DOI: https://doi.org/10.1103/kxdj-mlr5
Abstract
Information-driven active matter relies on continuous feedback loops subject to intrinsic processing latency. We investigate how this temporal nonlocality interacts with the spatial fluctuations of homogeneous isotropic turbulence for agents operating in the dilute, noncommunicating limit. This interplay induces a spontaneous symmetry breaking: as the control latency increases, the velocity statistics of the active agents continuously reshape from a unimodal Gaussian distribution to a bimodal attractor. This transition occurs in a regime where the underlying deterministic dynamics are physically bounded to remain stable. Theoretical analysis, supported by direct numerical simulations, demonstrates that this phenomenon is a noise-induced phase transition driven by the multiplicative turbulent strain rate, which renormalizes the effective stochastic potential to overcome viscous dissipation. These results indicate that rather than minimizing latency to avoid instability, agents operating near the phase boundary can utilize finite delays to extract energy from microscopic fluid noise and sustain macroscopic motility.
