Understanding quantum many-body physics with neural network wavefunctions
Christopher Roth
Flatiron Institute
Solids, which arise when atoms become arranged in a crystalline form, have incredibly diverse properties. Typically materials are well characterized by a simple picture where electrons move around in the potential given by the periodic arrangement of nuclei. Occasionally, the electrons notice each other, which can lead to a whole bunch of exotic physics including superconductivity, where electrons somehow conspire to move freely with no resistance. In order to simulate systems where electron-electron interactions are relevant, a massive amount of data is required, as each electron interacts with every other, leading to combinatorial complexity. I'll discuss how deep neural networks can be used to compress this data in order to make sense of exotic materials that are still poorly understood. Specifically, I'll show how deep neural networks can be used to learn the ground state of quantum many-body systems.
ORGANIZER
Vadim Oganesyan (CSI/GC-CUNY)