Symposium MT01-Integrating Machine Learning and Simulations for Materials Modeling

A new generation of computational approaches for materials modeling has emerged from innovative applications of machine learning to numerical simulations of materials. In this nascent and vibrant field, conventional methods of computational materials science are blended with data-science tools to produce physically-consistent models and conceptual knowledge. Moreover, this combined approach has also led to progress on long-standing technical challenges for numerical methods in materials modeling, such as alleviating length and time-scale limitations, improved accuracy, reducing computational costs, development inverse-design capabilities, model interpretability and transferability, visualization of materials representations in high-dimensional spaces, and handling of scarce and heterogeneous data sets. This symposium will explore emerging trends in combining data-driven frameworks with physics-based materials simulations across scales, from ab initio electronic structure calculations to large-scale atomistic simulations, mesoscale models, and continuum approaches. Applications will be considered for systems of relevance in materials science broadly construed: from structural materials to soft matter, functional materials, and quantum materials. Our goal is to deepen our understanding of novel methodological capabilities and highlight challenging issues that need to be tackled in order to enable widespread application and adoption of these approaches in academia and industry.

Topics will include:

Uncovering structural and dynamical complexity in large-scale simulations
Machine learning potentials for crystal defects and other heterogeneities
Physics-based materials modeling with machine learning
Reduced-order machine learning models for atomistic simulations
Multi-fidelity models, data-fusion, and transfer learning approaches
End-to-end differentiable frameworks, inverse problems, and deep generative models
Rare-events sampling and automated identification of collective variables
Topology optimization and development of tailored microstructure using machine learning

Invited Speakers:

Joshua Agar (Drexel University, USA)
Amartya Banerjee (University of California, Los Angeles, USA)
Christopher Bartel (University of Minnesota, USA)
Mathieu Bauchy (University of California, Los Angeles, USA)
Brad Boyce (Sandia National Laboratories, USA)
Penghui Cao (University of California, Irvine, USA)
Gábor Csányi (University of Cambridge, United Kingdom)
Felipe da Jornada (Stanford University, USA)
Brian DeCost (National Institute of Standards and Technology, USA)
Julia Dshemuchadse (Cornell University, USA)
Diego Gomez-Gualdron (Colorado School of Mines, USA)
Sara Kadkhodaei (University of Illinois at Chicago, USA)
Boris Kozinsky (Harvard University, USA)
Mihai-Cosmin Marinica (Commissariat à l’énergie atomique et aux énergies alternatives, France)
Benji Maruyama (Air Force Research Laboratory, USA)
Katsuyuki Matsunaga (Nagoya University, Japan)
Megan McCarthy (Sandia National Laboratories, USA)
Dane Morgan (University of Wisconsin–Madison, USA)
Danny Perez (Los Alamos National Laboratory, USA)
James Rondinelli (Northwestern University, USA)
Christopher Schuh (Massachusetts Institute of Technology, USA)
Daniel Schwalbe-Koda (Lawrence Livermore National Laboratory, USA)
Ryan Sills (Rutgers University, USA)
Taylor Sparks (The University of Utah, USA)
Alejandro Strachan (Purdue University, USA)
Wenhao Sun (University of Michigan, USA)
Aidan Thompson (Sandia National Laboratories, USA)
Eric Vanden-Eijnden (New York University, USA)
Wennie Wang (The University of Texas at Austin, USA)