Research

Multiscale Modeling and Computation Core

Overall Scientific Challenge

The overall scientific challenge is the development of multiscale theories and computational tools capable of handling several orders of magnitude in length scales (Ångstrom to mm) with required fidelity to integrate with and be informed by experimental studies.

Core Co-leaders: Chris Cox (Math, Clemson) and Qi Wang (Math, USC)

Contributing Faculty: Wolfgang Dahmen (Math, USC), Bijoy Dey (Chem, Claflin), Sophya Garashchuk (Chem, USC), Rachel Getman (ChemE, Clemson), Andreas Heyden (ChemE, USC), Jianhun Hu (CSE, USC), Olga Kuksenok (MSE, Clemson), Vitaly Rassolov (Chem, USC), Sapna Sarupria (ChemE, Clemson), Ulf Schiller (MSE, Clemson) and Victor Zordan (SoC, Clemson)

Affiliated Faculty: Paul Anderson (CS, CofC), Yiming Ji (CS/Math, USCB), Xuwei Liang (CS/Math, USCB)

The overall challenge in this area is the development of multiscale theories and computational tools capable of handling several orders of magnitude in length scales (Ångstrom to mm) with required fidelity to fully integrate with and be informed by experimental studies. The national vision of materials design has prioritized the use of modeling and computation, integrated with physical experimentation and cyberinfrastructure. Since the properties of a material are determined by its structure at different length and time scales, our goals are the development of advanced multiscale theoretical foundations, fast algorithms to handle high throughput computations, high resolution/fidelity imaging and visualization, andbig data analytics including uncertainty quantification. These models and tools must not only bridge the physics across multiple scales, but also be fully integrated with experimental observations and materials genome databases via an iterative design loop (Figure 2). A lasting legacy will be the availability of the developed computational tools and the database (e.g., Materials Data Bank, MDB, discussed below) to the broad materials research community.

While the value of multiscale modeling and computation in materials science is receiving increasing recognition, most current efforts are focused on theories with well-separated and limited scale ranges. We will integrate proven theories at local scales to develop a theoretical and computational framework enabling the exploration of material design, synthesis, structure and properties at an unprecedented range of scales, applicable to the model systems explored in the thrusts and a broader set of materials. The research goals are organized into two broad areas: first, we will develop multiscale theories and computational tools to guide and assist the three thrusts in the screening of materials’ composition, assembly and processing. New model reduction, novel learning methodologies and multi-level image analysis strategies will be incorporated to accelerate the modeling-experiment iterative development cycle. In addition, we will develop new courses and curricula for training students in multiscale theories and simulations. Especially, a new MS degree program in computational science will be launched at USCB. Second, we will implement our multiscale models and image analysis algorithms in an open source computational infrastructure designed to meet infrastructure needs for the material systems in this proposal. This framework will combine data analytics, experimental observations, visualization, and machine learning using the interactive, iterative design loop illustrated in Figure 2. The models and computational tools that we develop will be shared with the broader materials science community.