Machine Learning Interatomic Potentials

Also known as MLIP

Machine learning interatomic potentials approximate quantum-mechanical energies and forces from atomic structures, making molecular dynamics and materials simulations less computationally expensive. Common approaches include local descriptors, equivariant graph neural networks and transformers, with newer models adding long-range electrostatics, dispersion, and pretrained representations. Researchers use them to study molecules, reactions, crystals, interfaces, and materials at finite temperature, while active learning and uncertainty estimates help select useful training data and flag risky predictions. A central challenge is balancing speed with reliable transfer to unfamiliar structures and conditions: low benchmark errors do not necessarily guarantee accurate dynamics, thermodynamics, or out-of-distribution predictions.

Latest papers

91 papers13 papers in the last four weeks, up 86% on the four weeks before. 0.1% of all new papers.

Tensor-reduced atomic density representations
Oct 2, 2022James P. Darby, Dávid P. Kovács, Ilyes Batatia +4Molecular Representation Learning Invariant Representation Learning
Warwick Centre for Predictive Modelling, School of Engineering, University of Warwick, Coventry, CV4 7AL, UK · Engineering Laboratory, University of Cambridge, Cambridge, CB2 1PZ UK · ENS Paris-Saclay, Université Paris-Saclay, 91190 Gif-sur-Yvette, France +3

Oct 2, 2022J/K move · Enter open · S save

Tensor-reduced atomic density representations

James P. Darby, Dávid P. Kovács, Ilyes Batatia, Miguel A. Caro+3

Warwick Centre for Predictive Modelling, School of Engineering, University of Warwick, Coventry, CV4 7AL, UK · Engineering Laboratory, University of Cambridge, Cambridge, CB2 1PZ UK · ENS Paris-Saclay, Université Paris-Saclay, 91190 Gif-sur-Yvette, France+3

Molecular Representation Learning Invariant Representation Learning Machine Learning Interatomic Potentials Computational Materials Science Tensor Decomposition

Open paper AskPDF arXiv

Density based representations of atomic environments that are invariant under Euclidean symmetries have become a widely used tool in the machine learning of interatomic potentials, broader data-driven atomistic modelling and the visualisation and analysis of materials datasets.The standard mechanism used to incorporate chemical element information is to create separate densities for each element and form tensor products between them. This leads to a steep scaling in the size of the representation as the number of elements increases. Graph neural networks, which do not explicitly use density representations, escape this scaling by mapping the chemical element information into a fixed dimensional space in a learnable way. We recast this approach as tensor factorisation by exploiting the tensor structure of standard neighbour density based descriptors. In doing so, we form compact tensor-reduced representations whose size does not depend on the number of chemical elements, but remain systematically convergeable and are therefore applicable to a wide range of data analysis and regression tasks.