Bulletin of the American Physical Society
APS March Meeting 2023
Volume 68, Number 3
Las Vegas, Nevada (March 5-10)
Virtual (March 20-22); Time Zone: Pacific Time
Session N20: Matter at Extreme Conditions: Simulation and Novel MethodsFocus
|
Hide Abstracts |
Sponsoring Units: GSCCM Chair: J Matthew Lane, Sandia National Laboratories Room: Room 212 |
Wednesday, March 8, 2023 11:30AM - 11:42AM |
N20.00001: Crystal structure prediction starting from a liquid using machine-learning potentials Aniruddha M Dive, James Chapman, Stanimir Bonev Recent experimental measurements have discovered a number of temperature-driven phase transitions at high pressure, underlying the importance of including lattice dynamics in theoretical predictions of phase stability. However, direct structure prediction at finite temperature remains a challenge. In this talk we will discuss the feasibility of finite-temperature structure prediction by quenching liquids in large-scale simulations with machine learning (ML) potentials. The approach is based on solid-liquid structural similarities for efficiently training ML potentials and predicting crystal structures at target densities. To showcase the methodology, we have considered dense Li, where we have been able to simulate its rich phase diagram over a large pressure range. |
Wednesday, March 8, 2023 11:42AM - 11:54AM |
N20.00002: Automatic Differentiation in Dynamic Topology Optimization Kevin Korner Recent advances in scientific programming, particularly with regards to topology optimization and machine learning, have necessitated computational methods that generate gradients for optimization. One method to do so is to utilize a tool called automatic differentiation, a mechanism to algorithmically calculate derivatives of functions and combine them to generate gradients of compositions of functions. Code bases such as Jax and PyTorch (which particularly focused on machine learning applications) have demonstrated the ability to scale automatic differentiation to large problems. This allows for rapid gradient calculations, leading to reduced development time as well as significantly higher complexity in the equations used to study a physical phenomenon. While these methods have been studied in the context of machine learning, these approaches have only been applied to mechanics in a handful of cases. This presents an opportunity to study a large variety of optimization problems, such as topology, material parameters, or initial-value problems using this automatic differentiation infrastructure. By taking advantage of the sequential structure of time dependent problems, we develop rapid algorithms for gradient calculations that can be utilized in a large variety of different contexts. We demonstrate the efficacy of these algorithms by studying various dynamics problems. One example is in designing the initial velocity profile of an elastic wave to generate a specific final state. This can be utilized to amplify or reduce particular wave characteristics in non-trivial ways. |
Wednesday, March 8, 2023 11:54AM - 12:06PM |
N20.00003: Machine learning-based quantum accurate interatomic potentials for warm dense matter Sandeep Kumar, Hossein Tahmasbi, Mani Lokamani, Kushal Ramakrishna, Attila Cangi Modeling warm dense matter is relevant for various applications including the interior of gas giants and exoplanets, inertial confinement fusion, and ablation of metals. Ongoing and upcoming experimental campaigns in photon sources around the globe rely on numerical simulations that are accurate on the level of electronic structures. In that regard, density functional theory molecular dynamics (DFT-MD) simulations [1] have been widely used to compute thermodynamical properties of warm dense matter. However, two challenges impede further progress: (1) DFT-MD becomes computationally infeasible with increasing temperature (2) finite-size effects render many computational observables inaccurate because DFT-MD is limited to a few hundred atoms on current HPC platforms. Recently, molecular dynamics simulations using machine learning-based interatomic potentials (ML-IAP) could overcome these computational limitations. Here, we propose a method to construct ML-IAPs from DFT data based on SNAP descriptors [2]. We investigate the transferability of ML-IAPs over a large range of temperatures (1,000 to 100,000 K) which currently is a topic of active research. |
Wednesday, March 8, 2023 12:06PM - 12:18PM |
N20.00004: Development of Artificial intelligence-based Interatomic Potentials High-Entropy Diborides for Modelling the Physical and Thermal Properties Nur Aziz Octoviawan, Gregory E Hilmas, William G Fahrenholtz, Ridwan Sakidja The interatomic potentials designed for binary/high entropy diborides, and ultra-high temperature composites (UHTC) have been developed through the implementation of deep neural network (DNN) algorithms. These algorithms employed two different approaches and corresponding codes; 1) strictly local & invariant scalar-based descriptors as implemented in the DEEPMD code and 2) equivariant tensor-based descriptors as included in the ALLEGRO code. The samples for training and validation sets of the forces, energy, and virial data were obtained from the ab-initio molecular dynamics (AIMD) simulations and Density Functional Theory (DFT) calculations including the simulation data from the ultra-high temperature region (> 2000K). We then compared the accuracy of the Deep Learning potentials to predict not only the ground-state properties such as the elastic constants and the phonon dispersion curves but also the ultra-high temperature properties including the lattice parameters and melting behaviors. |
Wednesday, March 8, 2023 12:18PM - 12:30PM |
N20.00005: Ab initio high pressure and temperature phase diagram of Au Johann Bouchet, Pauline Richard, Francois Bottin, Grégory Geneste, Alois Castellano, Romuad Béjaud High pressure behaviour of gold is of fundamental interest since it is used as a calibration standard in experiments. However, recent ab initio [1-2] and experimental studies [3-5] show diverging solid phase stability domains. To simulate materials under extreme conditions that experiments cannot reach and that the existing numerical potentials fail to reproduce, Ab Initio Molecular Dynamics (AIMD) is the most appropriate method. Unfortunately, it is highly time consuming and therefore too prohibitive to build a complete phase diagram. Alternative methods exist, as the quasi-harmonic approximation[1] but its validity at high temperature is difficult to assess. Here to clarify conflicting results between experiments and simulations we use the Machine Learning Assisted Canonical Sampling (MLACS) [6], which accelerates AIMD by two orders of magnitude while maintaining an ab initio accuracy. We used MLACS and thermodynamic integration to extract Gibbs free energies of fcc, bcc and hcp phases, and built the phase diagram of Au between 0 and 1 TPa and from 0 K to 10 000 K. Our phase diagram is in agreement with the most recent experiments[4]. |
Wednesday, March 8, 2023 12:30PM - 12:42PM |
N20.00006: Stable solid molecular hydrogen above 900K from a machine-learned potential trained with diffusion Quantum Monte Carlo Scott Jensen, Hongwei Niu, Yubo Yang, Markus Holzmann, CARLO PIERLEONI, David M Ceperley Predicting the phase diagram of molecular hydrogen remains an important challenge for computational condensed matter physics. Accurate quantum Monte Carlo (QMC) methods are restricted to small system sizes where finite-size effects make it difficult to study the solid-solid and melting phase transitions. Recent advances allowed us to train a machine learned (ML) interatomic potential with many QMC calculations. We have produced a substantial publicly accessible QMC database for training interatomic potentials and used it to train a deep-neural network ML potential using QMC forces. We used this new potential in large-scale path integral molecular dynamics simulations to study molecular hydrogen. We find a phase diagram with HCP and C2/c-24 phases and two new structures with Fmmm-4 molecular centers. The Fmmm-4 structures show a molecular orientational order transition from an ordered low-temperature structure to an isotropic high-temperature phase which melts to a molecular liquid with a maximum melting temperature of 1450K at 150 GPa. This finding will likely lead to new experimental studies of the melting curve for molecular hydrogen. |
Wednesday, March 8, 2023 12:42PM - 1:18PM |
N20.00007: Variant Selections under Shock-induced Phase Transformation and deformation Twinning in BCC Metal Microstructures. Invited Speaker: Avinash Dongare The deformation behavior of BCC metal microstructures at high pressures has contributions from dislocation slip, deformation twinning, and phase transformation. While deformation twinning is a common mode of deformation in BCC metals at high pressures, Fe-based microstructures also demonstrate a BCC → HCP phase transformation when deformed above a threshold pressure of ~13 GPa. Recent studies have demonstrated the BCC → HCP → BCC phase transformation can result in a distribution of twins in the bcc microstructure. This distribution of twins is attributed to the selection of HCP phase variants during compression and their stability and reverse transformations during unloading. Understanding and predicting the role of microstructure and stress-states on these variant selections during deformation and their stability and reverse transformation behavior during unloading is important. The current efforts to investigate variant selections in BCC metals are largely limited to real-time in situ x-ray diffraction (XRD) and for single-crystal (sc) systems. In addition, the interpretations of the plasticity contributions from diffractograms are non-trivial, especially when multiple modes of deformation may be operating. Molecular dynamics (MD) simulations can successfully capture various deformation modes in metals and complement experiments using simulated diffractograms at various stages of evolution. This talk will discuss the use of virtual XRD and a newly developed virtual texture analysis (VirTex) algorithm to characterize the variant selections and their plasticity contributions during plastic deformation. The simulations investigate the role of BCC microstructure and loading stress-states on variant selections during phase transformation and twinning in BCC Fe microstructures as predicted using MD simulations. In addition, the simulations investigate the stability and the reverse transformation behavior of the variants during unloading. The simulations are able to unravel the role of variant selection and transformation that renders a distribution of twin boundaries in unloaded microstructures. |
Wednesday, March 8, 2023 1:18PM - 1:30PM |
N20.00008: Vortical flow and the modulationof jetting processes William Schill We present a combined theoretical/computational framework to model jetting processes following |
Wednesday, March 8, 2023 1:30PM - 1:42PM |
N20.00009: Finite-temperature lattice dynamics of FeV at high pressure from first principles Jorge A Munoz, Homero Reyes Pulido, BIMAL K C, Russell J Hemley, Ravhi Kumar The phonon dispersions of FeV in the B2 crystal structure computed using the finite displacement method and density functional theory (DFT) data show that the system becomes mechanically unstable when the lattice parameter is smaller than 2.78 Å. The instability coincides with a pressure-induced electronic topological transition and is accompanied by a charge transfer from the s to the (directional) d orbitals that occurs preferentially at the Fe sites. A method was developed to efficiently sample the phase space of molecular dynamics (MD) simulations and was used to extract harmonic force constants from finite-temperature DFT MD runs. The phonon dispersions obtained from such force constants for the FeV system at 2.77 Å show that temperature stabilizes the system at high enough temperature even in the absence of magnetism, and that the system is mechanically stable even at 10K when magnetism is considered, hinting at the importance of phonon anharmonicity in the phase stability of the system. X-ray diffraction patterns of FeV in a diamond-anvil cell at 300K and up to 80 GPa did not show any phase transitions. |
Wednesday, March 8, 2023 1:42PM - 1:54PM |
N20.00010: Superconductivity of cerium hydride at high pressure from DFT+DFPT+DMFT Yao Wei, Siyu Chen, Evgeny Plekhanov, Samuel Poncé, Bartomeu Monserrat, Cedric Weber Lanthanide hydrides are attracting considerable attention in condensed matter physics due to their high-temperature superconductivity under extreme pressure. The majority of the Lanthanides have valence f-orbital electrons, which suggests that strongly correlated interactions will play an important role in determining their physical properties. In this work, we propose a novel method to calculate the density of states, band structure, and phonon dispersion combining density functional theory with dynamical mean-field theory (DFT+DMFT). These results are then used in the context of density functional perturbation theory combined with maximally localized Wannier functions to evaluate the superconducting transition temperature of Lanthanide hydrides, taking CeH9 as an example. We find significant differences in the superconducting transition temperature when electronic correlations are included in the calculations. Our approach should be valuable for the study of phonon and electron-phonon related properties in strongly correlated materials. |
Wednesday, March 8, 2023 1:54PM - 2:06PM |
N20.00011: GPU-Accelerated Lattice Boltzmann Method for Shock Physics Computations Peter D Yeh Lattice Boltzmann Methods (LBM) have seen considerable success in computational fluid dynamics, especially for incompressible flows. LBM's simple and highly parallelizable algorithm is well-suited for newer multi-threaded architectures such as GPUs. Despite theoretical limitations that present difficulties for LBM's feasibility in the compressible flow regime, recent advances have slowly pushed past this boundary and allowed for similar parallelizability and algorithmic efficiency for moderate supersonic flows that admit shock waves. In this work, we implement a GPU-accelerated LBM method that can resolve shocks at moderate Mach numbers. Through demonstration of example problems, we compare the computational performance of LBM to traditional shock physics solvers (that use Lagrangian deformations plus a re-map to the original grid). We find improvements in time-to-solution for the LBM methods, but we also explore further areas of research that could increase LBM's robustness for shock physics computations. |
Follow Us |
Engage
Become an APS Member |
My APS
Renew Membership |
Information for |
About APSThe American Physical Society (APS) is a non-profit membership organization working to advance the knowledge of physics. |
© 2024 American Physical Society
| All rights reserved | Terms of Use
| Contact Us
Headquarters
1 Physics Ellipse, College Park, MD 20740-3844
(301) 209-3200
Editorial Office
100 Motor Pkwy, Suite 110, Hauppauge, NY 11788
(631) 591-4000
Office of Public Affairs
529 14th St NW, Suite 1050, Washington, D.C. 20045-2001
(202) 662-8700