More Like this

1. Machine Learning Local Potentials for Accelerated Electron-Phonon Interactions Calculations within the Projector Augmented-Wave Framework

   Xia, Yi (September 2026, Materials today physics)

   Electron–phonon interactions govern carrier dynamics, transport, and many optical and quan- tum phenomena in solids, but finite-displacement calculations within the projector augmented-wave (PAW) framework require up to 6N self-consistent supercell calculations for an N-atom system, making them costly for large, low-symmetry, and disordered materials. We introduce MLLocP, a machine learning strategy that learns the self-consistent local potential on real-space grids and supplies its displacement derivatives to the evaluation of all-electron PAW electron–phonon matrix elements. Analysis of the PAW decomposition identifies the local-potential contribution as a natural target for machine learning, while the remaining PAW quantities are retained within the established VASP/PHELEL framework. Using targeted grid-point sampling and feature-diversity selection for efficient model training, we validate MLLocP for elemental Cu, polar BAs, and chemically disordered Cu3Au. The learned potentials and their resulting displacement derivatives closely reproduce direct density functional theory calculations, yielding transport coefficients within approximately 3.5% for BAs mobility, 8% for Cu conductivity, and 11% for Cu3Au conductivity. For the 32-atom Cu3Au special quasirandom structure, using 40 sampled configurations reduces the number of self-consistent calculations by about fivefold relative to the 192 displaced structures required directly; using 20 or 10 configurations increases the acceleration to approximately 10- and 20-fold, with conductivity errors of 14.3% and 18.6%, respectively. MLLocP thus provides a scalable route to PAW electron–phonon calculations in complex materials while preserving the underlying all-electron formalism. 

   more » « less
2. Review of the second charged-particle transport coefficient code comparison workshop

   https://doi.org/10.1063/5.0198155  

   Stanek, Lucas J; Kononov, Alina; Hansen, Stephanie B; Haines, Brian M; Hu, S X; Knapp, Patrick F; Murillo, Michael S; Stanton, Liam G; Whitley, Heather D; Baalrud, Scott D; et al (May 2024, Physics of Plasmas)

   We report the results of the second charged-particle transport coefficient code comparison workshop, which was held in Livermore, California on 24–27 July 2023. This workshop gathered theoretical, computational, and experimental scientists to assess the state of computational and experimental techniques for understanding charged-particle transport coefficients relevant to high-energy-density plasma science. Data for electronic and ionic transport coefficients, namely, the direct current electrical conductivity, electron thermal conductivity, ion shear viscosity, and ion thermal conductivity were computed and compared for multiple plasma conditions. Additional comparisons were carried out for electron–ion properties such as the electron–ion equilibration time and alpha particle stopping power. Overall, 39 participants submitted calculated results from 18 independent approaches, spanning methods from parameterized semi-empirical models to time-dependent density functional theory. In the cases studied here, we find significant differences—several orders of magnitude—between approaches, particularly at lower temperatures, and smaller differences—roughly a factor of five—among first-principles models. We investigate the origins of these differences through comparisons of underlying predictions of ionic and electronic structure. The results of this workshop help to identify plasma conditions where computationally inexpensive approaches are accurate, where computationally expensive models are required, and where experimental measurements will have high impact. 

   more » « less
3. Deep neural network based quantum simulations and quasichemical theory for accurate modeling of molten salt thermodynamics

   https://doi.org/10.1039/d2sc02227c  

   Shi, Yu; Lam, Stephen T.; Beck, Thomas L. (July 2022, Chemical Science)

   With dual goals of efficient and accurate modeling of solvation thermodynamics in molten salt liquids, we employ ab initio molecular dynamics (AIMD) simulations, deep neural network interatomic potentials (NNIP), and quasichemical theory (QCT) to calculate the excess chemical potentials for the solute ions Na + and Cl − in the molten NaCl liquid. NNIP-based molecular dynamics simulations accelerate the calculations by 3 orders of magnitude and reduce the uncertainty to 1 kcal mol −1 . Using the Density Functional Theory (DFT) level of theory, the predicted excess chemical potential for the solute ion pair is −178.5 ± 1.1 kcal mol −1 . A quantum correction of 13.7 ± 1.9 kcal mol −1 is estimated via higher-level quantum chemistry calculations, leading to a final predicted ion pair excess chemical potential of −164.8 ± 2.2 kcal mol −1 . The result is in good agreement with a value of −163.5 kcal mol −1 obtained from thermo-chemical tables. This study validates the application of QCT and NNIP simulations to the molten salt liquids, allowing for significant insights into the solvation thermodynamics crucial for numerous molten salt applications. 

   more » « less
4. First order approximation of polyatomic ion and neutral diffusion in neutral gases under arbitrary fields

   https://doi.org/10.1063/5.0337543  

   Mubas-Sirah, Farah; Djoutsop, Wiljones; Larriba-Andaluz, Carlos (July 2026, The Journal of Chemical Physics)

   This work presents a trajectory-based method for predicting diffusion coefficients of ions and neutral molecules in dilute gases from the thermal limit to strongly driven conditions. Classical trajectory simulations are coupled with two-temperature kinetic theory to evaluate the collision integrals required by the generalized Einstein relations. Polyatomic ion geometries and charge distributions obtained from density functional theory are incorporated directly into 4–6–12 interaction potentials, allowing a consistent treatment of molecular structure, long-range interactions, and energy transfer during collisions. The method is implemented in the Mass Diffusivity Software (MaDiS) and validated against experimental data for both neutral and ionic systems across a wide range of conditions. For neutral molecules in the zero-field limit, the calculated diffusion coefficients reproduce experimental trends across multiple gases and temperatures with an average deviation of ∼5%, despite the use of non-optimized Lennard-Jones parameters. The simulations correctly capture the temperature dependence of diffusivity and remain consistent across a wide range of molecular sizes. Under applied electric fields, the method captures the departure from equilibrium and the resulting anisotropic transport, providing direct predictions of longitudinal and transverse diffusion coefficients. For monoatomic ions, longitudinal diffusion coefficients are reproduced with good agreement over substantial ranges of reduced electric field, typically up to 300–400 Td depending on the ion–gas system, consistent with benchmark datasets. Comparable behavior is obtained for polyatomic ions, demonstrating that the combination of trajectory-derived collision dynamics and the two-temperature formalism remains applicable when realistic molecular structure is included. 

   more » « less
5. Disentangling the effects of non-adiabatic interactions upon ion self-diffusion within warm dense hydrogen

   https://doi.org/10.1098/rsta.2023.0034  

   Angermeier, William A.; Scheiner, Brett S.; Shaffer, Nathaniel R.; White, Thomas G. (August 2023, Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences)

   Warm dense matter is a material state in the region of parameter space connecting condensed matter to classical plasma physics. In this intermediate regime, we investigate the significance of non-adiabatic electron-ion interactions upon ion dynamics. To disentangle non-adiabatic from adiabatic electron-ion interactions, we compare the ion self-diffusion coefficient from the non-adiabatic electron force field computational model with an adiabatic, classical molecular dynamics simulation. A classical pair potential developed through a force-matching algorithm ensures the only difference between the models is due to the electronic inertia. We implement this new method to characterize non-adiabatic effects on the self-diffusion of warm dense hydrogen over a wide range of temperatures and densities. Ultimately we show that the impact of non-adiabatic effects is negligible for equilibrium ion dynamics in warm dense hydrogen. This article is part of the theme issue ‘Dynamic and transient processes in warm dense matter’. 

   more » « less