# Suppressing correlations in massively parallel simulations of lattice models

## TL;DR

This paper compares domain decomposition schemes for lattice Monte Carlo simulations, identifying a method that minimizes correlations and enables efficient GPU-based parallelization, achieving significant speedups over CPU and sequential implementations.

## Contribution

The study introduces an effective domain decomposition scheme for GPU lattice simulations that reduces correlations and demonstrates high computational efficiency.

## Key findings

- Identified a domain decomposition scheme that yields correlation-free simulations on GPU.
- Achieved approximately 30x speedup over parallel CPU implementation.
- Achieved at least 180x speedup over sequential reference implementation.

## Abstract

For lattice Monte Carlo simulations parallelization is crucial to make studies of large systems and long simulation time feasible, while sequential simulations remain the gold-standard for correlation-free dynamics. Here, various domain decomposition schemes are compared, concluding with one which delivers virtually correlation-free simulations on GPU Extensive simulations of the octahedron model for $2+1$ dimensional Karda--Parisi--Zhang surface growth, which is very sensitive to correlation in the site-selection dynamics, were performed to show self-consistency of the parallel runs and agreement with the sequential algorithm. We present a GPU implementation providing a speedup of about $30\times$ over a parallel CPU implementation on a single socket and at least $180\times$ with respect to the sequential reference.