Molecular Dynamics– tag –
-
arXiv Paper Examines Limitations and Potential of MACE and CHGNet Foundation MLIPs in d4/d6/d7 Perovskite Oxide MD Simulations
arXiv International Overview This paper provides a detailed examination of the successes and limitations of foundation machine-learning interatomic potentials (MLIPs) like MACE and CHGNet in molecular dynamics (MD) simulations of d4/d6/d... -
YouTube Channel yaavikmaterials: ML Interatomic Potentials Enable Large-Scale MD with DFT Accuracy, Resolving System Size Trade-offs
yaavikmaterials (YouTube) Unknown Overview A YouTube video by yaavikmaterials explains how machine-learning interatomic potentials (MLIPs) like MACE, NequIP, and CHGNet are resolving the system size and simulation time trade-offs in mole... -
GPUMD 4.0 from Chalmers University Unleashes Machine Learning Power for Advanced Materials Simulations
Computational Materials Group @ Chalmers Sweden Overview Chalmers University of Technology's Computational Materials Group has released GPUMD 4.0, a high-performance molecular dynamics (MD) software that integrates cutting-edge machine l... -
arXiv Paper Benchmarks 6 Universal ML Interatomic Potentials, Evaluating Structural Fidelity and Performance for Lunar Regolith MD Simulations
arXiv International Overview Six universal machine-learning interatomic potentials (MLIPs)—MACE-MH, MatterSim, SevenNet-0, UPET, UMA, and NequIP-OAM-L—were comprehensively benchmarked for molecular dynamics (MD) simulations of lunar rego... -
LAMMPS Integrates Diverse Machine Learning Potentials Like MACE and CHGNet, Enhancing Atomic Simulation Versatility
LAMMPS Molecular Dynamics Simulator Unknown Overview The LAMMPS molecular dynamics simulator has significantly enhanced its interoperability with leading machine learning potential (MLP) frameworks, including MACE, CHGNet, DeePMD-kit, Ne... -
Benchmark of 23 ML Interatomic Potentials Reveals Large Models Offer Minimal Accuracy Gains at Significant Speed Cost
arXiv News International Overview A comprehensive benchmark study of 23 open-source machine-learning interatomic potentials (MLIPs) has revealed a critical trade-off between accuracy and speed. The findings indicate that large, state-of-... -
arXiv Paper: Adaptive Multi-Teacher Routing Significantly Boosts Reliability and Generalization of Universal ML Interatomic Potentials
arXiv International Overview Researchers have proposed an Adaptive Multi-Teacher Routing (ATR) framework that dramatically improves the reliability and generalization of universal machine-learning interatomic potentials (uMLIPs) by filte... -
Challenge of Atomic Partial Charges in MLIPs Impedes High-Precision Simulation of Non-Covalent Interactions
ACS Publications (Journal of Chemical Theory and Computation) International Overview While Machine Learning Interatomic Potentials (MLIPs) hold great promise for achieving DFT-level accuracy at low computational cost in molecular dynamic... -
Bias Identified in Universal Machine-Learned Interatomic Potentials; Iterative Fine-Tuning Improves Accuracy
Journal of Chemical Theory and Computation (ACS Publications) USA Overview This study thoroughly investigated intrinsic biases in universal machine-learned interatomic potentials (uMLIPs), such as MACE, and their impact on fine-tuning qu... -
arXiv Paper Presents ML Model for High-Precision Prediction of Metallic Glass Critical Cooling Rates Using Elemental and Molecular Simulation Features
arXiv USA Overview This study presents a machine learning model for predicting critical cooling rates of metallic glasses using computationally derived properties, specifically elemental and molecular dynamics simulation-based features. ...
