Argonne’s Computational Science Division (CPS) develops and employs a wide spectrum of methodologies combining theory and large-scale computational resources to address scientific challenges in chemical and material systems — in particular, problems that cannot be solved otherwise. Our work spans a multitude of models from force field-based to highly correlated quantum mechanics methods. Examples include semi-empirical, Density Functional Theory, coupled-cluster, and Quantum Monte Carlo methods, which have applications in structural and dynamical property determination for systems ranging from atomic clusters to semi-conductor materials.
We develop and enhance scalable methods and, in coordination with external collaborators, employ multi-physics approaches in multiple domains; for example, combining atomistic and continuum approaches to tackle problems that neither method could solve alone. We provide expertise in tailoring complex automated workflows to maximize computational resources for data-driven and high-throughput projects, deploy data analytic tools for exploration and visualization of data sets, and utilize machine learning and deep learning approaches to efficiently produce predictive models.
We build on our successful collaborations with chemistry and materials science experts at academic, government, and industrial institutions. Through participation in projects like the DOE Exascale Computing Project, we explore the mapping of paper-and-pencil algorithms onto current and upcoming advanced computational architectures and systems to accelerate scientific discoveries, expanding the boundaries of what’s possible. Many in CPS lead or closely collaborate with widely used, community-developed scientific applications and contribute novel algorithmic and methodological advances.
We also train future computational scientists by mentoring students and postdocs in research collaborations and contributing to and leading tutorials and workshops.