Coursera Flash Sale
40% Off Coursera Plus for 3 Months!
Grab it
Learn to benchmark universal machine learning force fields using CHIPS-FF, an open-source platform designed for evaluating machine learning force fields in computational materials science. Explore the comprehensive benchmarking infrastructure that integrates the Atomic Simulation Environment (ASE) with JARVIS-Tools to facilitate high-throughput simulations for complex material properties including elastic constants, phonon spectra, defect formation energies, surface energies, and interfacial and amorphous phase properties. Discover how CHIPS-FF evaluates the accuracy and computational cost of several graph-based universal MLFFs including ALIGNN-FF, CHGNet, MatGL, MACE, SevenNet, ORB, MatterSim and OMat24 specifically for semiconductor applications. Master the workflow and scaling capabilities of the platform while examining test sets of 104 materials commonly used in semiconductor devices. Understand the motivation behind multiscale modeling and the need for universal pretrained MLFFs in bridging quantum mechanical simulations with large-scale device modeling. Gain insights into DFT datasets used for MLFF training, recent benchmarking efforts, and the JARVIS-Leaderboard system for large-scale benchmarks. Access hands-on experience with the CHIPS-FF tool through the included tutorial section and learn how to contribute to community-driven benchmarking efforts in computational materials science.
