AI-MI Seminar Series: Learning, Understanding, and Predicting Quantum Phases in Two-Dimensional Materials
Join us for the next AI-MI Seminar Series talk as we explore how neural networks and Monte Carlo methods are cracking open one of the hardest unsolved problems in physical science.
This talk takes on the many-electron problem — where the sheer scale of quantum interactions has long outpaced our computational tools. The speaker presents a physics-inspired approach combining neural networks with Monte Carlo simulation to achieve breakthrough accuracy in modeling two-dimensional electron systems, uncovering exotic quantum phases that previous methods couldn’t reach.
Watch live at youtube.com/@AIMaterialsInstitute
Topic: The many-electron problem remains one of the great challenges in physical sciences. An incredible variety of phases arise from the competition between interaction, quantum zero-point motion, and the external environment provided by nuclei, including topology. Understanding and harnessing quantum phases can lead to profound advances, as has been evident through many of the technological breakthroughs in the past decades. Our tools for doing this are still very limited, however, because of the exponentially large dimensions in the many-electron problem, compounded by the fact that the quantum phases are often the outcome of a delicate balancing act between multiple competing or co-existing tendencies. The use of neural networks in combination with Monte Carlo methods has opened new opportunities. I will discuss our effort to apply neural networks in a physics-inspired manner for accurate and predictive computations for a class of Hamiltonians based on the two-dimensional electron gas, which are integral to the fast-growing field of two-dimensional materials. In a short period, these approaches have surpassed state-of-the-art computations in this area, and have made discoveries of new exotic quantum phases.
Speaker: Shiwei Zhang is a Senior Research Scientist/Group Leader at the Center for Computational Quantum Physics (CCQ), Flatiron Institute, Simons Foundation. He is recognized as a world leader in computational quantum physics, known for broad contributions in computational algorithm innovations and developments, especially in Monte Carlo methods, and their applications. Methods he pioneered have been applied in diverse areas including in condensed matter physics, quantum chemistry, ultra-cold atoms, and nuclear physics. He has led a number of international collaborative teams on major research projects in computational quantum physics. Zhang received his Ph.D in Physics from Cornell University. After three years of postdoctoral appointments, first at Los Alamos National Lab and then at Ohio State under an NSF postdoctoral fellowship, he joined the faculty of William & Mary, where he remained for over 20 years, holding the position of Chancellor Professor of Physics before he moved to Flatiron in 2018. He was recipient of multiple awards including an NSF CAREER award and the Cottrell Scholar Award. He is a Fellow of the American Physical Society.