Minimax entropy production for arbitrary processes

Mahran Yousef; Ben Ansbacher; David H. Wolpert

Home/Archive/arXiv

arXiv · 2610.03478

Minimax entropy production for arbitrary processes

Mahran Yousef·Ben Ansbacher·David H. Wolpert

View original source ↗Read source document ↗Export metadata ↓

Abstract

For any physical process, varying the initial distribution over its states will change the resulting expected entropy production (EP). An initial distribution that minimizes the EP of a particular process is called the prior of that process. Varying the microphysical details of the process will change its prior.We consider the following question: if an engineer wants to design a system to implement a given map, what prior should they design into the system to minimize the EP for the worst-case actual distribution? We derive the closed-form solution to this problem when the map is deterministic. We also show that for stochastic maps, a worst-case actual distribution can always be chosen to be a point mass on a single state, and that the associated minimax prior resembles a Boltzmann-Gibbs equation, and can be found by a globally convergent numerical algorithm. We also show that the associated minimax mismatch cost is the maximum uncertainty about the input that remains after observing the output. Finally, we extend the analysis to the continuous-time regime, showing that the minimax instantaneous mismatch-cost diverges, while the corresponding finite-time minimax mismatch cost has a universal short-time scaling set by the largest escape rate of the dynamics.

Explore related subjects

cond-mat.stat-mech

Keep this discovery

Explore connections, maps & timelines

BibTeX RIS

Mahran Yousef, Ben Ansbacher, David H. Wolpert. 2026-10-02. Minimax entropy production for arbitrary processes. https://arxiv.org/abs/2610.03478

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related papers

arXiv2026

Glass and jamming transitions in a random organization model

We study a two-dimensional, off-lattice particle model introduced to describe absorbing phase transitions in driven non-Brownian suspensions. We numerically explore the $(ϕ,ε)$ phase diagram, where $ϕ$ is the packing fraction and $ε$ controls the amplitude of particle jumps. We use a binary mixture to suppress crystallization, which allows us to disentangle the model's distinct phase transitions between amorphous states. At large $ϕ$, we find that the approach to the absorbing transition is preceded by a non-equilibrium glass transition to a non-diffusive amorphous state. This dynamic arrest makes the location of the critical absorbing transition protocol-dependent. The $ε\to 0$ end-point of the transition line defines a jamming transition whose location is shown to vary continuously with the preparation protocol, and cannot serve as a unique definition of random close packing. Near jamming, we observe a complex landscape and marginal stability, reminiscent of Gardner phases found in thermal glasses. The critical exponents characterizing packings at the jamming transition numerically agree with alternative approaches based on energy minimization, and with analytic predictions from mean-field replica theory. We analyze hyperuniformity of large-scale density fluctuations in fluid and glass phases, where it emerges with qualitatively distinct signatures, and show that random organization dynamics does not determine the hyperuniformity observed in jammed packings, which is found to be non-universal. Our results show that random organization models share deep physical similarities with thermal soft-particle systems undergoing glass and jamming transitions, with little impact of the non-equilibrium nature of the microscopic dynamics on emerging physical properties.

cond-mat.stat-mech↗

arXiv2026

Structural crossover of complex networks: bridging degree correlation and fractality

We investigate the relationship between long-range degree correlations and fractality in scale-free random networks. By analyzing degree correlations in scale-free random networks at the percolation threshold, at which a giant component emerges, we derive the ball volume $\tildeν_k(l)$, defined as the average number of nodes within distance $l$ from a root node with degree $k$. The resulting expression predicts a degree-dependent structural crossover. For $l\ll k^{1/(\df-1)}$, the ball volume grows as $\tildeν_k(l)\sim kl$, whereas for $l\gg k^{1/(\df-1)}$, it crosses over to the global fractal scaling $\tildeν_k(l)\sim l^{\df}$. Here, $\df$ is the fractal dimension of critical scale-free random networks. We further show that the structural crossover can be naturally understood in terms of critical branching processes and is also observed in empirical fractal networks.

cond-mat.stat-mech↗

arXiv2026

Probing fluctuating hydrodynamics through current fluctuations in hard rod gas

We study the full counting statistics of the integrated current in a one-dimensional hard-rod gas in equilibrium, employing two recently proposed fluctuating hydrodynamic frameworks: (i) Ballistic Macroscopic Fluctuation Theory (BMFT) and (ii) diffusive fluctuating hydrodynamic theory based on correlated Brownian motion approximation of hard-rod quasiparticles (CBFHD). We investigate how the predictions of these two approaches compare with each other and with the underlying microscopic dynamics. We show within BMFT that the current distribution admits a large-deviation form under ballistic scaling, and we derive an explicit expression for the large-deviation function, valid for an arbitrary inhomogeneous initial state factorized in point-particle coordinates. The corresponding current cumulants are also obtained. We subsequently analyze the current statistics within the CBFHD framework, which explicitly incorporates stochastic noise and dissipative terms, and demonstrate that the current distribution likewise obeys a large-deviation principle in this approach. We find that, while BMFT predictions for the first few cumulants and the rate function are in excellent agreement with direct numerical simulations of the microscopic hard-rod dynamics, CBFHD prediction is not. We further verify the predictions of the CBFHD independently using a collection of correlated Brownian motions, which provides an approximate description of the dynamics of hard-rod quasiparticles. Our findings indicate that CBFHD does not accurately capture large-scale current fluctuations in the hard-rod gas, whereas BMFT yields predictions in quantitative agreement with the microscopic dynamics, at least in the context of full counting statistics.

cond-mat.stat-mech↗