Ceramics and Glasses Simulations and Machine Learning: Machine Learning and Artificial Intelligence
Sponsored by: ACerS Glass & Optical Materials Division
Program Organizers: Mathieu Bauchy, University of California, Los Angeles; Peter Kroll, University of Texas at Arlington; N. M. Anoop Krishnan, Indian Institute of Technology Delhi
Monday 8:00 AM
November 2, 2020
Room: Virtual Meeting Room 18
Location: MS&T Virtual
Session Chair: Mathieu Bauchy, UCLA; Peter Kroll, UT Arlington; Anoop Krishnan, IIT Delhi
8:00 AM
Introductory Comments: Ceramics and Glasses Simulations and Machine Learning: Mathieu Bauchy1; 1University of California Los Angeles
Introductory Comments
8:05 AM Keynote
Application of Natural Language Processing to Zeolites and Cementitious Materials: Elsa Olivetti1; 1Massachusetts Institute of Technology
Advances in applying natural language processing (NLP) to scientific text have been successfully applied to well-studied material systems with large amounts of data. However, we need ways to leverage literature data in materials domains without thousands of papers. Applying NLP pipelines to these types of materials science systems can be challenging due to the general schema and the noisiness of automatically extracted data. In this presentation, we demonstrate how to leverage domain knowledge to build upon existing data extraction techniques and improve extraction accuracy using examples in the zeolite and alternative cement fields. This presentation will describe an effort to integrate artificial intelligence with material science to support the development of low environmental impact concrete mixtures. Generative modeling approaches can be used to learn from this and other data to optimize the design of concrete mixtures.
8:45 AM Keynote
Data, Materials and Disorder: Stefano Curtarolo1; 1Duke University
Critical understanding of large amount of data exposes the unavoidability of disorder and leads to new descriptors for discovering entropic materials. The formalism, based on the energy distribution spectrum of randomized calculations, captures the accessibility of equally-sampled states near the ground state and quantifies configurational disorder capable of stabilizing high-entropy homogeneous phases. The combination of these descriptors and Machine Learning uncover scientific surprises. Research sponsored by DOD-ONR.
9:25 AM Keynote
JAX, M.D.: End-to-End Differentiable, Hardware Accelerated, Molecular Dynamics in Pure Python: Samuel Schoenholz1; Ekin Cubuk1; 1Google Brain
Molecular Dynamics (MD) software is used across a vast range of subjects from physics and materials science to biochemistry and drug discovery. Most MD software involves significant use of handwritten derivatives and code reuse across C++, FORTRAN, and CUDA. In this work we bring the substantial advances in software that have taken place in machine learning to MD with JAX, M.D. (JAX MD). JAX MD is an end-to-end differentiable MD package written entirely in Python that can be just-in-time compiled to CPU, GPU, or TPU. JAX MD allows researchers to iterate extremely quickly and lets researchers easily incorporate machine learning models into their workflows. In addition to making workloads easier, JAX MD allows researchers to take derivatives through whole-simulations to design Physical systems with desirable properties. We discuss the architecture of JAX MD through several vignettes with an eye towards glass physics. Code available at www.github.com/google/jax-md.
10:05 AM Invited
De Novo Discovery of Nanoporous Structures with Tailored Sorption Isotherm by Machine Learning: Yuhan Liu1; Mathieu Bauchy1; 1University of California, Los Angeles
Nanoporous materials (e.g., zeolite, activated carbon, metal-organic framework, polymeric membranes, etc.) have various technological applications, including gas separation, gas storage, catalytic transformations, etc. The functionalities of nanoporous materials strongly depend on their pore size and shape distribution—which present virtually limitless degrees of freedom. Here, based on high-throughput lattice density functional theory (LDFT) simulations and a convolutional neural network (CNN) predictor, we present a model allowing us to predict the water sorption isotherm of nanoporous configurations. The training of an inverse CNN generator then enables the inverse design of optimal porous microstructures featuring tailored/unusual sorption isotherms.