An Explicit Microreversibility Violating Thermodynamic Markov Process

We explicitly construct a non-microreversible transition matrix for a Markov process and apply it to the standard three-state Potts model. This provides a clear and simple demonstration that the usual micoreversibility property of thermodynamical Monte Carlo algorithms is not strictly necessary from a mathemetical point of view.

💡 Research Summary

The paper challenges the long‑standing assumption that microreversibility (also known as detailed balance) is a mandatory requirement for thermodynamic Monte Carlo algorithms. By explicitly constructing a transition matrix that deliberately violates detailed balance, the authors demonstrate that the equilibrium Boltzmann distribution can still be obtained as long as the global balance condition is satisfied. The work proceeds in several logical stages.

First, the authors review the foundations of Markov‑chain Monte Carlo (MCMC) methods. In the conventional framework, detailed balance guarantees that for any pair of states i and j the probability fluxes satisfy π(i) P(i→j)=π(j) P(j→i), where π is the target distribution (typically the Boltzmann weight). This condition is sufficient to ensure that π is a stationary distribution, but it is not necessary; a weaker global balance condition Σ_i π(i) P(i→j)=π(j) is enough. The paper emphasizes that the latter is often overlooked because detailed balance is easier to implement and to prove.

The core contribution is a concrete 3 × 3 transition matrix T for a three‑state system. The matrix is built by assigning arbitrary positive numbers to each off‑diagonal element, then normalizing each row so that Σ_j T_{ij}=1. Crucially, T_{ij}≠T_{ji} for many pairs, so detailed balance is explicitly broken. Nevertheless, the matrix is constructed to be irreducible (every state can be reached from any other) and aperiodic, guaranteeing a unique stationary distribution. By invoking the Perron‑Frobenius theorem, the authors show that T possesses a dominant eigenvalue λ=1 with a strictly positive eigenvector, which they prove coincides with the desired Boltzmann distribution for the chosen Hamiltonian.

To test the theoretical construction, the authors embed T into a Monte Carlo simulation of the standard three‑state Potts model on a two‑dimensional lattice. In the Potts model each lattice site can be in one of three colors, and the energy depends on nearest‑neighbor color matches. Traditional simulations use the Metropolis or heat‑bath algorithms, both of which enforce detailed balance by design. In the present work, a spin update is proposed according to the asymmetric probabilities given by T, and the acceptance probability is derived from the Metropolis‑Hastings framework using the ratio π(j) T_{ji}/