Bulletin of the American Physical Society
2020 Annual Meeting of the APS Four Corners Section (Virtual)
Volume 65, Number 16
Friday–Saturday, October 23–24, 2020; Albuquerque, NM (Virtual)
Session M02: Computational Physics IILive
|
Hide Abstracts |
Chair: Josh Berger, Colorado State University |
Saturday, October 24, 2020 12:30PM - 12:42PM Live |
M02.00001: Two Dimensional Turbulence with a Driven Relativistic Lattice Boltzmann Model Mark Watson We investigate a relativistic adaptation of the Lattice Boltzmann Method that reproduces the equations of motion for a turbulent relativistic hydrodynamic system. The classical Lattice Boltzmann method and it's extension to relativistic fluid dynamics is described. The numeric formulation is evaluated using a zero-averaged stirring force introduced into the numerics to induce turbulence, and the turbulent flow characteristics produced are compared to properties of a classical turbulent hydrodynamic flow. [Preview Abstract] |
Saturday, October 24, 2020 12:42PM - 12:54PM Live |
M02.00002: Predicting Microwave Cavity Resonances through Machine Learning Nathan Schwartz Microwave cavities are special resonators consisting of a closed metal structure that confines electromagnetic fields. The purpose of our research is to create and implement various neural networks that can accurately predict electromagnetic field data including resonance modes and frequencies for any given cavity configuration containing two dielectric resonators. To this end, we have generated 50,000 configurations of training data and 10,000 configurations of validation data with their respective solutions. This data was then used to train our neural networks. When data which was not part of the neural network training was used to test the results, the neural network was able predict frequencies, modes, and coefficients to a high degree of accuracy for most regions of our sampled configurations. We are currently developing and implementing functions that will assist us in sampling specific high-error configurations. By sampling more of these configurations, we hope to better train the neural networks in these regions, and thereby reduce the average error in the resulting predictions. [Preview Abstract] |
Saturday, October 24, 2020 12:54PM - 1:06PM Live |
M02.00003: Why is uncertainty quantification of sloppy models challenging? Yonatan Kurniawan, Cody Petrie, Kinamo Williams, Mark Transtrum Interatomic models (IMs) are used in materials modeling to predict material's properties of interest. The development of a single IM can take anywhere from several months to years and relies on expert intuition, and yet these potentials are usually only valid for a particular application of interest. Extending existing IMs to new applications is an active area of research. Quantifying the uncertainty of an IM can tell us how much we can trust the predictions it makes. I compare Bayesian (Markov Chain Monte Carlo) and Frequentist (profile likelihood) methods to quantify uncertainty of IM's parameters. I demonstrate these methods on Lennard-Jones and Morse potentials in predicting the energy and forces of the bases atoms of a triclinic body-centered crystal structure from the OpenKIM database. Results indicate that these models are "sloppy" in some of their parameters, i.e., likelihood surfaces have long, narrow canyons and broad, flat plateaus. I discuss difficulties and challenges from applying these uncertainty quantification methods to sloppy models. [Preview Abstract] |
Saturday, October 24, 2020 1:06PM - 1:18PM Live |
M02.00004: Force Field Optimization for Molten Salts Talmage Porter, Dennis Della Corte, Todd Millecam With the inclusion of Molten Salt Reactors (MSRs) in the selection of Generation IV Nuclear Reactors (Gen IV - a selection of nuclear reactor designs to be researched for commercial applications), molten salts have come into focus. The primary use of molten salts is as coolant for MSRs. In-lab testing of various salt mixtures has proven to be difficult with issues including safety hazards and time constraints. Due to these complications, simulation techniques, like molecular dynamics (MD), are a viable option for testing. Atomic level simulations are a well-researched field primarily due to its heavy use in biophysical and material science research, but the necessary forcefields for molten salts are still lacking in quality. Here, we will discuss current shortcomings in available molten salt forcefields and approaches that will lead to better parametrization. We will also present how improved MD can be used to derive relevant thermodynamical properties, necessary for MSR design regulatory approval. [Preview Abstract] |
Saturday, October 24, 2020 1:18PM - 1:30PM Live |
M02.00005: Order to Disorder in Quasiperiodic Systems David Morison, Ben Murphy, Elena Cherkaev, Ken Golden Four decades since the discovery of quasicrystals, the material properties arising from quasiperiodic microgeometry remain an active topic of theoretical intrigue and engineering utili ty. Here we introduce a class of Moiré type quasiperiodic media with novel macroscopic behavior. As the microgeometry of a Moiré system is tuned, the transport properties switch from those of ordered to randomly disordered materials in a fashion which parallels the Anderson localization transition, even though there are no scattering or interference effects at play. This transition is evident within an integral representation that applies broadly to the effective electrical, thermal, elastic and optical properties of two-phase composite media. This representation distills the relationship between microgeometry and bulk transport into the spectral characteristics of an operator, which is analogous to the Hamiltonian in quantum transport phenomena. Periodic media display sharp resonances, band gaps, and spatially extended eigenstates separated by "mobility edges" of localized states. As we tune system parameters to aperiodicity and disorder, eigenstates become more uniformly extended, and level repulsion increases as the spectral properties transition toward obeying universal Wigner-Dyson statistics. [Preview Abstract] |
Saturday, October 24, 2020 1:30PM - 1:42PM Live |
M02.00006: Bayesian Approach to Uncertainty Quantification of Interatomic Models in OpenKIM Database Kinamo Williams, Yonatan Kurniawan, Cody Petrie, Mark Transtrum Interatomic models (IMs) are used in molecular modeling to predict material properties of interest. The development of a single IM can take anywhere from several months to years and relies on expert intuition, and yet these potentials are usually only valid for a particular application of interest. Extending existing IMs to new applications is an active area of research. Quantifying the uncertainty of an IM can tell us how much we can trust the predictions it makes. I will discuss two methods for analyzing uncertainty: Fisher Information Matrix, FIM, and Markov Chain Monte Carlo, MCMC. Using Monte Carlo methods, I sample from the posterior distribution of the parameters when trained on data. I demonstrate this method on Lennard-Jones and Morse potentials fit to triclinic crystal configurations from the OpenKIM database. In particular, IMs are often sloppy, i.e., have likelihood surfaces with long, narrow canyons and broad, flat plateaus. I will be comparing the benefits and drawbacks of the two methods. [Preview Abstract] |
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