Massively Parallel Genetic Optimization through Asynchronous Propagation of Populations

TL;DR

Propulate is a parallel genetic optimization algorithm that improves efficiency by asynchronously propagating populations, enabling faster hyperparameter searches on HPC systems without losing solution quality.

Contribution

It introduces a novel asynchronous propagation method for genetic algorithms, significantly increasing optimization speed on HPC resources compared to traditional synchronized approaches.

Findings

Propulate is up to 1000 times faster than traditional methods.

It maintains comparable solution accuracy to established tools.

The approach is effective for hyperparameter optimization in high-performance computing environments.

Abstract

We present Propulate, an evolutionary optimization algorithm and software package for global optimization and in particular hyperparameter search. For efficient use of HPC resources, Propulate omits the synchronization after each generation as done in conventional genetic algorithms. Instead, it steers the search with the complete population present at time of breeding new individuals. We provide an MPI-based implementation of our algorithm, which features variants of selection, mutation, crossover, and migration and is easy to extend with custom functionality. We compare Propulate to the established optimization tool Optuna. We find that Propulate is up to three orders of magnitude faster without sacrificing solution accuracy, demonstrating the efficiency and efficacy of our lazy synchronization approach. Code and documentation are available at https://github.com/Helmholtz-AI-Energy/propulate