mlsanityData to 14 Oct 2026

A practical guide to machine learning interatomic potentials -- Status and future

arXiv 2503.09814 ↗First posted 12 Mar 2025cond-mat.mtrl-sci cs.LGcited by 26

Not on the map. Why?

Abstract (arXiv)

The rapid development and large body of literature on machine learning interatomic potentials (MLIPs) can make it difficult to know how to proceed for researchers who are not experts but wish to use these tools. The spirit of this review is to help such researchers by serving as a practical, accessible guide to the state-of-the-art in MLIPs. This review paper covers a broad range of topics related to MLIPs, including (i) central aspects of how and why MLIPs are enablers of many exciting advancements in molecular modeling, (ii) the main underpinnings of different types of MLIPs, including their basic structure and formalism, (iii) the potentially transformative impact of universal MLIPs for both organic and inorganic systems, including an overview of the most recent advances, capabilities, downsides, and potential applications of this nascent class of MLIPs, (iv) a practical guide for estimating and understanding the execution speed of MLIPs, including guidance for users based on hardware availability, type of MLIP used, and prospective simulation size and time, (v) a manual for what MLIP a user should choose for a given application by considering hardware resources, speed requirements, energy and force accuracy requirements, as well as guidance for choosing pre-trained potentials or fitting a new potential from scratch, (vi) discussion around MLIP infrastructure, including sources of training data, pre-trained potentials, and hardware resources for training, (vii) summary of some key limitations of present MLIPs and current approaches to mitigate such limitations, including methods of including long-range interactions, handling magnetic systems, and treatment of excited states, and finally (viii) we finish with some more speculative thoughts on what the future holds for the development and application of MLIPs over the next 3-10+ years.

Key facts. Cited by 26 papers in the map (data through 14 Oct 2026). Builds on 38 of its 173 references that are in the map. First posted on arXiv 12 Mar 2025.

Builds on

38 of 173 references.

References · 38

  1. 2016Semi-Supervised Classification with Graph Convolutional Networkscited by 11,361
  2. 2021E(3)-Equivariant Graph Neural Networks for Data-Efficient and Accurate Interatomic Potentialscited by 708
  3. 2022MACE: Higher Order Equivariant Message Passing Neural Networks for Fast and Accurate Force Fieldscited by 474
  4. 2021Crystal Diffusion Variational Autoencoder for Periodic Material Generationcited by 270
  5. 2022Learning Local Equivariant Representations for Large-Scale Atomistic Dynamicscited by 256
  6. 2020Machine Learning Force Fieldscited by 207
  7. 2023EquiformerV2: Improved Equivariant Transformer for Scaling to Higher-Degree Representationscited by 204
  8. 2022e3nn: Euclidean Neural Networkscited by 174
  9. 2019Deep neural network solution of the electronic Schr"odinger equationcited by 127
  10. 2019Machine learning for molecular simulationcited by 110
  11. 2022The Design Space of E(3)-Equivariant Atom-Centered Interatomic Potentialscited by 101
  12. 2023MatterGen: a generative model for inorganic materials designcited by 98
26 more
  1. 2022Forces are not Enough: Benchmark and Critical Evaluation for Machine Learning Force Fields with Molecular Simulationscited by 95
  2. 2024Fine-Tuned Language Models Generate Stable Inorganic Materials as Textcited by 93
  3. 2018Less is more: sampling chemical space with active learningcited by 82
  4. 2017DeePMD-kit: A deep learning package for many-body potential energy representation and molecular dynamicscited by 79
  5. 2024Orb: A Fast, Scalable Neural Network Potentialcited by 66
  6. 2019DScribe: Library of Descriptors for Machine Learning in Materials Sciencecited by 63
  7. 2023Scalable Diffusion for Materials Generationcited by 54
  8. 2023From Molecules to Materials: Pre-training Large Generalizable Models for Atomic Property Predictioncited by 39
  9. 2023Matbench Discovery -- A framework to evaluate machine learning crystal stability predictionscited by 35
  10. 2023Scaling the leading accuracy of deep equivariant models to biomolecular simulations of realistic sizecited by 33
  11. 2020Machine learning for electronically excited states of moleculescited by 30
  12. 2022Hyperactive Learning (HAL) for Data-Driven Interatomic Potentialscited by 17
  13. 2019TeaNet: universal neural network interatomic potential inspired by iterative electronic relaxationscited by 16
  14. 202086 PFLOPS Deep Potential Molecular Dynamics simulation of 100 million atoms with ab initio accuracycited by 15
  15. 2020Combining SchNet and SHARC: The SchNarc machine learning approach for excited-state dynamicscited by 10
  16. 2022Tensor-reduced atomic density representationscited by 10
  17. 2017WACSF - Weighted Atom-Centered Symmetry Functions as Descriptors in Machine Learning Potentialscited by 9
  18. 2022Spin-Dependent Graph Neural Network Potential for Magnetic Materialscited by 9
  19. 2018Machine learning enables long time scale molecular photodynamics simulationscited by 8
  20. 2020Accelerating Finite-temperature Kohn-Sham Density Functional Theory with Deep Neural Networkscited by 7
  21. 2024Overcoming systematic softening in universal machine learning interatomic potentials by fine-tuningcited by 6
  22. 2021Complex Spin Hamiltonian Represented by Artificial Neural Networkcited by 4
  23. 2024Efficient Symmetry-Aware Materials Generation via Hierarchical Generative Flow Networkscited by 3
  24. 2020Machine learning and excited-state molecular dynamicscited by 2
  25. 2024Breaking the Molecular Dynamics Timescale Barrier Using a Wafer-Scale Systemcited by 2
  26. 2024FastCHGNet: Training one Universal Interatomic Potential to 1.5 Hours with 32 GPUscited by 2

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2025: 14; 2026: 12.

26 papers.

Sort:

By area:

  1. 2025FlashMD: long-stride, universal prediction of molecular dynamicscited by 12
  2. 2025Scalable Boltzmann Generators for equilibrium sampling of large-scale materialscited by 11
  3. 2026MatRIS: Toward Reliable and Efficient Pretrained Machine Learning Interatomic Potentialscited by 7
  4. 2026Harnessing AtomisticSkills for Agentic Atomistic Researchcited by 6
  5. 2025Molecular Machine Learning in Chemical Process Designcited by 5
  6. 2025Learning Inter-Atomic Potentials without Explicit Equivariancecited by 5
  7. 2025Generative Models for Crystalline Materialscited by 4
  8. 2026Pretrained Model Representations as Acquisition Signals for Active Learning of MLIPscited by 3
  9. 2026Smooth Dynamic Cutoffs for Machine Learning Interatomic Potentialscited by 2
  10. 2026Force-Aware Neural Tangent Kernels for Scalable and Robust Active Learning of MLIPscited by 2
  11. 2025ADAPT: Lightweight, Long-Range Machine Learning Force Fields Without Graphscited by 1
  12. 2026Proof-Carrying Materials: Falsifiable Safety Certificates for Machine-Learned Interatomic Potentialscited by 1
14 more
  1. 2025Accurate, transferable, and verifiable machine-learned interatomic potentials for layered materialscited by 0
  2. 2025Knowledge Distillation Framework for Accelerating High-Accuracy Neural Network-Based Molecular Dynamics Simulationscited by 0
  3. 2025Model Accuracy and Data Heterogeneity Shape Uncertainty Quantification in Machine Learning Interatomic Potentialscited by 0
  4. 2025Motional representation; the ability to predict odor characters using molecular vibrationscited by 0
  5. 2025Why Physics Still Matters: Improving Machine Learning Prediction of Material Properties with Phonon-Informed Datasetscited by 0
  6. 2025On The Finetuning of MLIPs Through the Lens of Iterated Maps With BPTTcited by 0
  7. 2025From Regression to Classification: Exploring the Benefits of Categorical Representations of Energy in MLIPscited by 0
  8. 2025Comparing the latent features of universal machine-learning interatomic potentialscited by 0
  9. 2026Autotuning T-PaiNN: Enabling Data-Efficient GNN Interatomic Potential Development via Classical-to-Quantum Transfer Learningcited by 0
  10. 2026Making Room for AI: Multi-GPU Molecular Dynamics with Deep Potentials in GROMACScited by 0
  11. 2026GFFMERGE: Efficient Merging of Graph Neural Force Fields and Beyondcited by 0
  12. 2026Scaling Density Functional Theory with Gaussian Splattingcited by 0
  13. 2026MOF-VERIFY: A Failure-Aware Agentic Harness for MOF Hypothesis Verificationcited by 0
  14. 2026Automatically Building and Updating a Knowledge Graph of MLIP Modelscited by 0