Reformulating non-equilibrium steady-states and generalised Hopfield discrimination

Authors:
  • UCUgur Cetiner
    Ugur Cetiner
  • JGJeremy Gunawardena
    Jeremy Gunawardena

Abstract
Markov processes are widely used to model stochastic systems in physics and biology. Their steady-state (s.s.) probabilities are given in terms of their transition rates by the Matrix-Tree theorem (MTT). The MTT uses spanning trees in a graph-theoretic representation of the system and reveals that, away from thermodynamic equilibrium, s.s. probabilities become globally dependent on all transition rates and the resulting expressions grow super-exponentially in the graph size. The overwhelming complexity and lack of thermodynamic insight have impeded analysis, despite substantial progress elsewhere in non-equilibrium physics. Assuming Arrhenius rates with vertex energies and edge barrier energies, we show that the s.s. probability of vertex $i$ is proportional to the average of $\exp(-S(P))$, where $S(P)$ is the entropy generated along a minimal path, $P$, from $i$ to a reference vertex. The average is taken over a Boltzmann-like probability distribution on spanning trees, whose ``energies'' are their total edge barrier energies. This reformulation offers a thermodynamic interpretation that smoothly generalises equilibrium statistical mechanics and it reorganises the expression complexity: the number of distinct minimal-path entropies depends on the number of edges where energy is expended, not on graph size. We demonstrate the power of this reformulation by extending Hopfield's analysis of discrimination by kinetic proofreading to any graph in which energy is expended at only one edge, for which we derive a general formula for the s.s. error ratio.