## Description

Back to topContemporary computational materials science relies on an ecosystem of models that span an extremely broad range of characteristic time and length scales. These range from quantum mechanics-based methods at the smallest length and timescales to macroscale finite element approaches at the largest length scales. This includes for instance models to predict the evolution of defects in materials, such the kinetic Monte Carlo method or cluster dynamics; or models of plasticity that employ either dislocation dynamics at the mesoscopic scale or crystal plasticity, at the macroscopic scale. The evolution of defects and microstructures can in parallel be studied with experimental characterizations and imaging devices, e.g., techniques dedicated to monitoring the evolution of microstructures (such as grain coarsening with X-ray tomography).

A longstanding challenge in the field is to develop systematic techniques to leverage all available data sources to develop accurate materials models. However, due to the wide range of different computational model formulations and scales (phase field, discrete defect models, reaction-diffusion equations), of numerical approaches (spectral methods, finite elements, particle solvers), and of experimental data streams, mathematical challenges related to the design and efficient execution of data-driven meso and macro-scale models abound.

This workshop will focus on the challenge of informing meso and macro-scale models from data, either obtained from lower-scale computations or directly from experiments. Topics of interest include the use of data-driven methods to learn effective models from measured data (e.g., using sparse system identification methods, or backpropagation through PDE solvers), the development of rigorous data-driven scale-bridging techniques, or the development of optimal design of experiments methods to identify small sets of experiments or calculations that would best constrain the models at the lowest cost. We also welcome contributions related to high-throughput data generation approaches applicable to meso and macro-scale materials modeling and uncertainty quantification methods for data-driven models.

## Organizers

Back to top## Speakers

Back to top## Schedule

Back to top**Speaker: **Amanda Howard (Pacific Northwest National Laboratory (PNNL))

**Speaker: **Weiqi Chu (UMass Amherst)

**Speaker: **Feliks Nüske (Max Planck Institute DCTS Magdeburg)

**Speaker: **Sichen Yang (Johns Hopkins University)

**Speaker: **Vivek Oommen (Brown University)

**Speaker: **Krishna Garikipati (University of Southern California (USC))

**Speaker: **Yue Yu (Lehigh University)

**Speaker: **Soumendu Bagchi (Oak Ridge National Laboratory)

**Speaker: **Yixiang Deng (Ragon Institute of Mass General, MIT, and Harvard)

**Speaker: **Thomas Swinburne (CNRS)

**Speaker: **Molei Tao (Georgia Institute of Technology)

**Speaker: **Xiaochuan Tian (University of California, San Diego)

**Speaker: **Nicolas Bertin (Lawrence Livermore National Laboratory)

**Speaker: **Jason Hattrick-Simpers (University of Toronto)

**Speaker: **Raymundo Arroyave (Texas A&M University)

**Speaker: **Dallas Trinkle (University of Illinois at Urbana-Champaign)

**Speaker: **Anjana Talapatra (Los Alamos National Laboratory (LANL))

**Speaker: **Sergei Kalinin (University of Tennessee, Knoxville, and Pacific Northwest National Laboratory)

**Speaker: **Abigail Hunter (Los Alamos National Laboratory (LANL))

**Speaker: **Thomas Hudson (University of Warwick)

**Speaker: **Fei Lu (Johns Hopkins University)

**Speaker: **Quanjun Lang (Duke University)