The entropy production in stochastic dynamical systems is linked to the structure of their causal representation in terms of Bayesian networks. Such a connection was formalized for bipartite (or multipartite) systems with an integral fluctuation theorem in [Phys. Rev. Lett. 111, 180603 (2013)]. Here we introduce the information thermodynamics for time series, that are non-bipartite in general, and we show that the link between irreversibility and information can only result from an incomplete causal representation. In particular, we consider a backward transfer entropy lower bound to the conditional time series irreversibility that is induced by the absence of feedback in signal-response models. We study such a relation in a linear signal-response model providing analytical solutions, and in a nonlinear biological model of receptor-ligand systems where the time series irreversibility measures the signaling efficiency.
Bayesian chemotaxis is an information-based target search problem inspired by biological chemotaxis. It is defined by a decision strategy coupled to the dynamic estimation of target position from detections of signaling molecules. We extend the case of a point-like agent previously introduced (Vergassola et al., Nature (2007)), which establishes concentration sensing as the dominant contribution to information processing, to the case of a circle-shaped agent of small finite size. We identify gradient sensing and a Laplacian correction to concentration sensing as the two leading-order expansion terms in the expected entropy variation. Numerically, we find that the impact of gradient sensing is most relevant because it provides direct directional information to break symmetry in likelihood distributions, which are generally circle shaped by concentration sensing.
An integral relation is derived from the Fokker-Planck equation which connects the steady-state probability currents with the dynamics of relaxation on short timescales in the limit of small perturbation fields. As a consequence of this integral relation, a general lower bound on the steady-state entropy production is obtained. Two particular ensembles of perturbation fields are then considered, respectively constant gradients and density displacements, and correspondingly two different averaging-based thermodynamic bounds are derived from the integral relation. These provide feasible methods to estimate the steady-state entropy production from relaxation experiments.
We theoretically address minimal search strategies of self‐propelled particles towards hidden targets in three‐dimensional space. The particles can sense if targets are close, e. g., by detecting released signaling molecules, but they cannot deduce directional cues. We investigate composite search strategies, where particles switch between extensive outer search and intensive inner search; inner search is started when proximity of a target is detected and ends when a certain inner search time has elapsed. In the simplest strategy, active particles move ballistically during outer search, and transiently reduce their directional persistence during inner search. In a second, adaptive strategy, particles exploit a dynamic scattering effect by reducing directional persistence only outside a well‐defined target zone. These two search strategies require only minimal information processing capabilities, yet increase target encounter rates substantially. The optimal inner search time scales as power‐law with exponent −2/3 with target density, reflecting a trade‐off between exploration and exploitation.