Necessary and sufficient symmetries in Event-Chain Monte Carlo with generalized flows and Application to hard dimers

Tristan Guyon,Arnaud Guillin,Manon Michel

2023-07-05

Abstract:Event-Chain Monte Carlo methods generate continuous-time and non-reversible Markov processes which often display important accelerations compared to their reversible counterparts. However their generalization to any system may appear less straightforward. In this work, we build on the recent analytical characterization of such methods as generating Piecewise Deterministic Markov Processes (PDMP) to clearly decipher the necessary symmetries the PDMP must obey from the sufficient ones which may prove to be too restrictive in a general setting. Thus, we derive a necessary rotational invariance of the probability flows and the minimum event rate, which identifies with the corresponding infinitesimal rejection rate. Such conditions always yield a correct ECMC scheme. We then generalize such results to the case of more general deterministic flows than the translational ones. In particular, we define two classes of interest of general flows, the ideal and uniform-ideal ones, which respectively suppresses or reduces the event rates. From there, we implement a complete non-reversible sampling of a systems of hard dimers, thanks to the introduction of rotational flows, which are uniform-ideal and shows a speed-up of up to ~3 compared to the state-of-the-art ECMC/Metropolis hybrid scheme.

Statistical Mechanics

What problem does this paper attempt to address?

### Problems the paper attempts to solve

This paper aims to solve how to introduce more general deterministic flows (especially rotational flows) in the Event - Chain Monte Carlo (ECMC) method to improve the ability to sample anisotropic particle systems (such as dimers) efficiently. Specifically, the paper focuses on the following aspects:

1. **Distinction between necessary and sufficient conditions**:
- First, by regarding the ECMC method as a Piecewise - Deterministic Markov Process (PDMP), the paper clearly distinguishes the necessary symmetries required to generate a PDMP from the possibly overly strict sufficient symmetries. This helps to understand the basic requirements of the ECMC method, rather than just the existing simplified assumptions.

2. **Rotational invariance of probability flow**:
- The paper derives the necessary rotational invariance of probability flow and the minimum event rate, which ensure the correct ECMC scheme. In particular, the minimum event rate is equivalent to the corresponding infinitesimal Metropolis rejection rate.

3. **Generalization of general deterministic flows**:
- The paper further generalizes the above results, considering more general deterministic flows than translational flows. Two types of flows of particular interest are defined in the paper: ideal flows and uniform ideal flows. Ideal flows can completely suppress the minimum event rate, while uniform ideal flows only suppress the minimum event rate under a uniform stationary distribution.

4. **Application to anisotropic particle systems**:
- The paper achieves a complete non - reversible sampling of the hard - dimer system by introducing a rotational flow (which belongs to the uniform ideal flow). Experimental results show that at the densest density (ρ = 0.7), compared with the state - of - the - art ECMC/Metropolis hybrid scheme, the speed of this method is increased by about 3 times.

### Solutions to specific problems

1. **Analysis of symmetries**:
- Through the PDMP framework, the paper analyzes in detail the necessary and sufficient symmetries in the ECMC method. In particular, by analyzing the compensation relationship of the integral parts of the transport term and the event term, the necessary rotational invariance of probability flow and the minimum event rate are obtained.

2. **Definition of ideal and uniform ideal flows**:
- The paper defines ideal flows and uniform ideal flows and discusses their implementation. Ideal flows can completely suppress the minimum event rate, but may be more difficult to implement algorithmically. Uniform ideal flows are easier to implement, including translational and rotational flows.

3. **Numerical verification**:
- The paper verifies the effect of introducing rotational flow through numerical experiments. Experimental results show that the rotational flow significantly improves the sampling efficiency while maintaining good irreducibility, especially in the hard - dimer system.

### Summary

By deeply analyzing the symmetries and probability flow characteristics of the ECMC method, this paper successfully generalizes the ECMC method to more general deterministic flows, especially rotational flows. This generalization not only improves the sampling efficiency of anisotropic particle systems but also provides a theoretical basis for further research and optimization of the ECMC method.