Adrian Feiguin

Sponsor: Department of Energy, Basic Energy Sciences

Artificial intelligence and data enabled predictive modeling of collective phenomena in strongly correlated quantum materials

Our ultimate goal is to accelerate discovery in quantum materials at DOE-supported user facilities. We will meet this goal through three specific aims. Aim 1 — generating and confirming novel low-energy effective many-body models for quantum materials — will provide robust methods for extracting low-energy effective models describing the collective behavior of quantum materials. Aim 2 — accelerating model solutions for advanced non-perturbative computational methods — is creating new state-of-the-art computational approaches for solving these models. Our new algorithms have already opened up new parameter regimes that were previously out of reach by enabling the first DQMC simulations of physically realistic phonon energies. In this new cycle, we will further accelerate these algorithms and develop new AI-powered approaches to solve the models generated in Aim 1 and beyond. Finally, aim 3 — Creating end-to-end experiment and theory workflows — is laying the foundation for integrating Aims 1 and 2 into new scientific workflows for scattering experiments. Building on several proof-of-concept studies, we will focus on developing robust methods for automating the extraction of effective models from RIXS experiments in this cycle. Together, these three aims will accelerate discovery in quantum materials and significantly improve our ability to understand the microscopic origins of their novel behavior and ultimately control their properties.