Four talks related to OpenKIM at SES 2020


The 57th Annual Technical Meeting of Society of Engineering Science (SES) was scheduled to be held in Minneapolis, MN, on October 20–22, 2020. Due to the COVID-19 pandemic, this meeting as canceled, and instead SES held a Virtual Technical Meeting on September 29–October 1, 2020.

Four talks related to the KIM project were presented at the SES Virtual Technical Meeting:

  • "Quantifying potential dependence of grain boundary properties using OpenKIM", Brandon Runnels, Daniel Karls, Ryan Elliott, Ellad Tadmor, Live Presentation, Wednesday, 30 September, 3:30-3:45. Abstract

  • "Predicting the kink-pair formation enthalpy for screw dislocations in tungsten via line-on-substrate and line-tension models that include core energy contributions", David Cereceda, Yichen Qian, Nicolas Bertin, Daniel Karls, Yaser Afshar, Sicong He, Vasily Bulatov, Jaime Marian, Ellad Tadmor, Live Presentation, Wednesday, 30 September, 3:45-4:00. Abstract

  • "Uncertainty quantification of classical interatomic potentials in OpenKIM database", Yonatan Kurniawan, Cody Petrie, Mark Transtrum, Kinamo Jahali Williams, Pre-Recorded Video Session. Abstract

  • "Information geometry to explore the parameter space of interatomic models", Cody Petrie, Yonatan Kurniawan, Kinamo Jahali Williams, Mark Transtrum, Pre-Recorded Video Session. Abstract

The abstracts for the the talks are given below.

Quantifying potential dependence of grain boundary properties using OpenKIM

Brandon Runnels1, Daniel Karls2, Ryan Elliott2, Ellad Tadmor2.

1University of Colorado, Colorado Springs, 2University of Minnesota.

Grain boundaries (GBs) are of critical importance in modeling micromechanical phenomena such as plasticity, twinning and phase transformation, and solidification, all of which directly affect macroscopic material properties. Atomistic simulations based on empirical interatomic potentials (IPs) provide a method for quantifying GB properties (specifically, excess energy) as a function of orientation. The current literature is replete with GB energy data computed using different IPs for a variety of configurations and materials systems, making a thorough survey of available data difficult. Furthermore, the precise dependence of GB data on the all-important IP has yet to be quantified, and current data is in perpetual danger of deprication as IPs are constantly improved. In this work, the OpenKIM framework is used to implement automatic generation of symmetric tilt grain boundaries (STGBs) for any rational angle, along with a grid search procedure for estimating the absolute (global) minimum energy. GB energy data is presented that was generated automatically for several metals with more than 40 IPs. The integration with OpenKIM ensures that the tests will be immediately re-run each time the potentials are updated, but also that each set of grain boundary data is inextricably linked to an immutable record of its potential. The results presented here exhibit the characteristic cusp structure for nearly all the potentials; however, the large-angle GB energy varies drastically depending on the potential. This finding has far-reaching implications for future atomistic polycrystal simulations, and the framework provides an automated mechanism for quantifying this potential dependence.

Predicting the kink-pair formation enthalpy for screw dislocations in tungsten via line-on-substrate and line-tension models that include core energy contributions

David Cereceda1, Yichen Qian1, Nicolas Bertin2, Daniel Karls3, Yaser Afshar3, Sicong He4, Vasily Bulatov2, Jaime Marian4, Ellad Tadmor3.

1Villanova University, 2Lawrence Livermore National Laboratory, 3University of Minnesota, 4UCLA.

The plastic behavior of body-centered cubic (bcc) metals like tungsten is governed by the motion of 1⁄2 <111> screw dislocations at low to medium homologous temperatures via the nucleation and propagation of kink pairs in close-packed planes. This process can be studied at the atomistic scale via molecular dynamics (MD) or density-function theory (DFT) calculations. However, these techniques would require an elevated and sometimes unaffordable computational cost to study the presence of kinks within long dislocation lines.

In addition to the well-characterized elastic contribution, the energy of a dislocation contains an inelastic, or ‘core’, term that reflects the loss of validity of elasticity theory at dislocation segments. While the elastic part is known to be symmetric about its maximum value for the edge orientation (minimum for screw), in bcc metals, the core energy displays an asymmetry than can be characterized using atomistic calculations. In kink-pair configurations on screw dislocations, this asymmetry leads to a difference in energy between ‘right’ and ‘left’ kinks that is not captured in elastic models.

In this work, we calculate dislocation segment self-energies as a function of dislocation character in bcc tungsten and kink-pair enthalpies as a function of stress. To avoid finite-size artifacts in atomistic simulations, we develop continuum models of kink-pair configurations based on full elasticity and line tension approaches, parameterized with a substrate Peierls potential and dislocation self-energies obtained from atomistic calculations. The novelty of this approach also lies on including the proposed methodology to calculate dislocation core energies within the OpenKIM framework.

Uncertainty quantification of classical interatomic potentials in OpenKIM database

Yonatan Kurniawan1, Cody Petrie1, Mark Transtrum1, Kinamo Jahali Williams1.

1Brigham Young University.

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 compare Bayesian (Markov Chain Monte Carlo) and Frequentist (profile likelihood) methods to quantify uncertainty of IM parameters. I demonstrate these methods on Lennard-Jones and Morse potentials fit to triclinic crystal configurations 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 disscuss the relative strenghts and weaknesses of each approach.

Information geometry to explore the parameter space of interatomic models

Cody Petrie1, Yonatan Kurniawan1, Mark Transtrum1, Kinamo Jahali Williams1.

1Brigham Young University.

Atomistic simulations often use fitted Interatomic Models (IMs) due to their ability to quickly compute the energy and forces on collections of atoms. Estimating the uncertainty in the fitted parameters is important for assessing the reliability of a model's predictions. Multiparameter models, including many IMs, are often sloppy, i.e., exhibit an extreme insensitivity to coordinated changes in some of their parameter values. Consequently, fitted parameters often have large uncertainties and quantifying this uncertainty can be challenging. We use an information geometry approach to systematically explore the parameter space of families of IMs. In this approach, a multiparameter model is interpreted as a manifold of potential predictions with parameters as coordinates. We numerically calculate geodesics on the model manifold and identify boundaries of the manifold that are associated with coordinated, extreme values of the parameters. We show how these boundaries are related to sloppiness and uncertainty in fitted parameter values and discuss implications for model selection and uncertainty quantification in atomistic simulations.