Applications and Advances in Machine Learning Force Fields

Abstract

Force fields (FFs) form the basis of molecular simulations and have significant implications in diverse fields such as materials science, chemistry, physics, and biology. A suitable FF is required to accurately describe system properties. However, an off-the-shelf FF may not be suitable for certain specialized systems, and researchers often need to tailor the FF that fits specific requirements. Before applying machine learning (ML) techniques to construct FFs, the mainstream FFs were primarily based on first-principles force fields (FPFF) and empirical FFs. However, the drawbacks of FPFF and empirical FFs are high cost and low accuracy, respectively, so there is a growing interest in using ML as an effective and precise tool for reconciling this trade-off in developing FFs. In this review, we introduce the fundamental principles of ML and FFs in the context of machine learning force fields (MLFF). We also discuss the advantages and applications of MLFF compared to traditional FFs, as well as the MLFF toolkits widely employed in numerous applications.

Pathways to a Shiny Future: Building the Foundation for Computational Physical Chemistry and Biophysics in 2050

“…The model and method development is also a vital part of computational physical chemistry, which can greatly benefit from AI concepts. The (semi)­automatized or/and AI-driven model/force field generation is already under extensive development − but yet to be tested beyond the current “comfort zone” of a limited range of systems. While methodological studies frequently incorporate experimental data, more often than not, they repurpose already available data from well-understood systems.…”

Roadmapping Report: Focused Research Organizations in Atomistic Simulation

“…From Quantum Chemistry in Drug Discovery by Corin Wagen 4 In the age of AI, it's natural to ask if machine learning can improve these methods. Much work has gone into developing fast force fields that approach the accuracy of DFT or higher levels of theory [5][6][7][8][9][10][11][12][13] . Many researchers believe that such a force field is inevitable and will supplant techniques such as DFT, turning to older methods to refine initial results.…”

“…On the other hand, dedicated software to enable citizen scientists to conduct basic simulations may create new opportunities for crowdsourced research and provide a wealth of user interaction data. 5. Language models show promise as research assistants in atomistic simulation, with the potential to conduct autonomous theoretical research.…”