Learning Everywhere: A Taxonomy for the Integration of Machine Learning and Simulations

TL;DR

This paper introduces a comprehensive taxonomy categorizing how machine learning techniques are integrated with simulations, covering various patterns, algorithmic areas, and action strategies to enhance simulation capabilities.

Contribution

It provides the first detailed taxonomy and activity catalog for integrating machine learning with simulations across multiple patterns and algorithmic domains.

Findings

Identifies eight key patterns for ML-simulation integration.

Classifies three main algorithmic areas: particle dynamics, agent-based models, PDEs.

Proposes three action areas for improving and surrogating simulations.

Abstract

We present a taxonomy of research on Machine Learning (ML) applied to enhance simulations together with a catalog of some activities. We cover eight patterns for the link of ML to the simulations or systems plus three algorithmic areas: particle dynamics, agent-based models and partial differential equations. The patterns are further divided into three action areas: Improving simulation with Configurations and Integration of Data, Learn Structure, Theory and Model for Simulation, and Learn to make Surrogates.