More Like this

1. The joint automated repository for various integrated simulations (JARVIS) for data-driven materials design

   https://doi.org/10.1038/s41524-020-00440-1  

   Choudhary, Kamal; Garrity, Kevin F.; Reid, Andrew C.; DeCost, Brian; Biacchi, Adam J.; Hight Walker, Angela R.; Trautt, Zachary; Hattrick-Simpers, Jason; Kusne, A. Gilad; Centrone, Andrea; et al (December 2020, npj Computational Materials)

   null (Ed.)

   Abstract The Joint Automated Repository for Various Integrated Simulations (JARVIS) is an integrated infrastructure to accelerate materials discovery and design using density functional theory (DFT), classical force-fields (FF), and machine learning (ML) techniques. JARVIS is motivated by the Materials Genome Initiative (MGI) principles of developing open-access databases and tools to reduce the cost and development time of materials discovery, optimization, and deployment. The major features of JARVIS are: JARVIS-DFT, JARVIS-FF, JARVIS-ML, and JARVIS-tools. To date, JARVIS consists of ≈40,000 materials and ≈1 million calculated properties in JARVIS-DFT, ≈500 materials and ≈110 force-fields in JARVIS-FF, and ≈25 ML models for material-property predictions in JARVIS-ML, all of which are continuously expanding. JARVIS-tools provides scripts and workflows for running and analyzing various simulations. We compare our computational data to experiments or high-fidelity computational methods wherever applicable to evaluate error/uncertainty in predictions. In addition to the existing workflows, the infrastructure can support a wide variety of other technologically important applications as part of the data-driven materials design paradigm. The JARVIS datasets and tools are publicly available at the website: https://jarvis.nist.gov . 

   more » « less
2. Machine learning of 27Al NMR electric field gradient tensors for crystalline structures from DFT

   https://doi.org/10.1038/s41598-025-10017-x  

   Sun, He; Dwaraknath, Shyam; Ling, Handong; Persson, Kristin A; Hayes, Sophia E (July 2025, Scientific Reports)

   NMR crystallography has emerged as a promising technique for the determination and refinement of atomic coordinates in crystal structures. The crystal structure of compounds containing quadrupolar nuclei, such as 27Al, can be improved by directly comparing solid-state NMR measurements to DFT computations of the electric field gradient (EFG) tensor. The non-negligible computational cost of these first-principles calculations limits the applicability of this method to all but the most well-defined structures. We developed a fast, low-cost machine learning model to predict EFG parameters based on local structural motifs and elemental parameters. We computed 8081 EFG tensors from 1681 27Al crystalline solids using DFT and benchmarked them against 105 experimentally measured 27Al sites. Surprisingly, simple local geometric features dominate the predictive performance of the resulting random-forest model, yielding an R2 value of 0.98 and an RMSE of 0.61 MHz for CQ, the quadrupolar coupling constant. This model accuracy should enable pre-refining future structural assignments before finally validating with first-principles calculations. Such a catalogue of 27Al NMR tensors can serve as a tool for researchers assigning complex NMR spectra influenced by the nuclear electric quadrupole interaction. 

   more » « less
3. Ab Initio Molecular Dynamics Study of Quadrupolar Spin Relaxation in an Ionic Liquid

   https://doi.org/10.1002/jcc.70311  

   Vidal, Luciano N; Ducati, Lucas C; Autschbach, Jochen (January 2026, Journal of Computational Chemistry)

   ABSTRACT This work implements a methodology for studying quadrupolar nuclear spin relaxation in ionic liquids. The dynamic behavior of the ions in the liquid is described by ab initio molecular dynamics (aiMD) with forces obtained from density functional theory (DFT) calculations with periodic boundary conditions and a non‐hybrid functional. The electric field gradient (EFG) driving the quadrupolar relaxation was calculated with free boundary conditions, using clusters that contained the ion of interest surrounded by two coordination shells treated quantum mechanically and augmented with a solvation model. Tests showed that EFG calculations using only the first coordination shell, containing five nearest neighbors, also provide a suitable model, because the relaxation rates differ by no more than 4% from the results from the two‐shell solvation. The results of this study show that the relaxation of the deuterated ethylammonium nitrate () occurs within the extreme narrowing regime for a spectrometer magnetic field T and is therefore characterized by the ensemble variance of the EFG and the correlation time associated with the EFG autocorrelation function. The quadrupolar relaxation of demanded molecular dynamics production times longer than 330 ps and averaging over multiple ions, as well as independent trajectories to get suitably converged relaxation rates. The calculated relaxation rate is Hz, about 60% above the rate reported experimentally. However, the approach utilized in the present study has an accuracy similar to, or better than, what has been previously reported for systems involving non‐ionic solvents that required simulations of 100 ps duration or less. 

   more » « less
4. Exploiting Extensive-Form Structure in Empirical Game-Theoretic Analysis

   Konicki, Christine; Chakraborty, Mithun; Wellman, Michael P. (December 2022, Lecture notes in computer science)

   Hansen, Kristoffer Arnsfelt; Liu, Tracy Xiao; Malekian, Azarakhsh (Ed.)

   Empirical game-theoretic analysis (EGTA) is a general framework for reasoning about complex games using agent-based simulation. Data from simulating select strategy profiles is employed to estimate a cogent and tractable game model approximating the underlying game. To date, EGTA methodology has focused on game models in normal form; though the simulations play out in sequential observations and decisions over time, the game model abstracts away this temporal structure. Richer models of extensive-form games (EFGs) provide a means to capture temporal patterns in action and information, using tree representations. We propose tree-exploiting EGTA (TE-EGTA), an approach to incorporate EFG models into EGTA. TE-EGTA constructs game models that express observations and temporal organization of activity, albeit at a coarser grain than the underlying agent-based simulation model. The idea is to exploit key structure while maintaining tractability. We establish theoretically and experimentally that exploiting even a little temporal structure can vastly reduce estimation error in strategy-profile payoffs compared to the normal-form model. Further, we explore the implications of EFG models for iterative approaches to EGTA, where strategy spaces are extended incrementally. Our experiments on several game instances demonstrate that TE-EGTA can also improve performance in the iterative setting, as measured by the quality of equilibrium approximation as the strategy spaces are expanded. 

   more » « less
5. Bayesian Symmetry-Respecting Machine Learning Framework for Predicting Electronic Structures

   Ghosh, S (November 2026, IMECE 2026 - International Mechanical Engineering Congress & Exposition)

   Kohn-Sham Density Functional Theory (DFT) is the workhorse method for electronic structure calculations of materials in a wide variety of fields. However, its computational cost scales cubically with the system size making its use infeasible for systems beyond a few thousand atoms, which is a bottleneck for materials research. ML models for electron densities have been proposed to address this, but they often lack one or more of the following: uncertainty quantification (UQ) capability, respect for essential symmetries, incorporation of physics, generalizability, accuracy, efficiency, and scalability. This proposed research aims to address all these issues within a single ML framework. To achieve these goals, we focus on gaining fundamental insights into the atomic configurations, element compositions, and corresponding electronic structure by creating a machine learning (ML) model to predict electron density. The proposed ML model can predict electron densities for any atomic configurations and element compositions, thereby eliminating the need for further DFT calculations for the class of materials it is trained. The broad applicability of the ML model is demonstrated for various materials systems. The proposed ML model can be extended to a universal model for a group of elements and hence will be useful for materials design. This research will be tightly coupled with several educational and outreach activities to train students of various levels on ML and ML-based materials design. A key novelty of the proposed research is the following: a neural network model is developed that is simultaneously Bayesian and symmetry-equivariant for the first time. It shows greater generalization and UQ capability through Bayesian approaches while ensuring the rotational and translational equivariance. Therefore, it allows for combining two different aspects in a single model: generalization to a wide range of atomic-configurations or material-compositions, and inductive biases for problems with rotational and translational symmetry. DFT is the dominant method for first-principles electronic structure calculations for various materials in several disciplines, especially for materials research. United States leads in large-scale DFT calculations. Hence, the advancements in ML models for electron density prediction will enable the United States to remain at the forefront of materials design and discovery, which will have far-reaching benefits across multiple industries, including semiconductors, biotechnology, solar energy, hydrogen energy, energy storage, advanced manufacturing, quantum computing, and next-generation sensors and actuators. This fundamental research is expected to find broad applicability across various disciplines, particularly those that need both Bayesian and symmetry-equivariant models, such as biomedical imaging and continuum physics problems. The codes, data generated, and lecture videos will be released via public repositories 

   more » « less