Unified graph neural network force-field for the periodic table: solid state applications. (2nd February 2023)

• Record Type: Journal Article
• Authors: Choudhary, Kamal; DeCost, Brian; Major, Lily; Butler, Keith; Thiyagalingam, Jeyan; Tavazza, Francesca
• Is Part Of: Digital discovery. Volume 2:Number 2(2023)
• Page Start: 346
• Page End: 355
• Publication Date: 2023-02-02
• Subjects: Chemistry -- Data processing -- Periodicals; Medical sciences -- Data processing -- Periodicals; Machine learning -- Periodicals; 542.85
• DOI: 10.1039/d2dd00096b
• Languages: English
• ISSNs: 2635-098X

Abstract

Abstract : Classical force fields (FFs) based on machine learning (ML) methods show great potential for large scale simulations of solids. Abstract : Classical force fields (FFs) based on machine learning (ML) methods show great potential for large scale simulations of solids. MLFFs have hitherto largely been designed and fitted for specific systems and are not usually transferable to chemistries beyond the specific training set. We develop a unified atomisitic line graph neural network-based FF (ALIGNN-FF) that can model both structurally and chemically diverse solids with any combination of 89 elements from the periodic table. To train the ALIGNN-FF model, we use the JARVIS-DFT dataset which contains around 75 000 materials and 4 million energy-force entries, out of which 307 113 are used in the training. We demonstrate the applicability of this method for fast optimization of atomic structures in the crystallography open database and by predicting accurate crystal structures using a genetic algorithm for alloys.