SKM 2023 – wissenschaftliches Programm
Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe
DY: Fachverband Dynamik und Statistische Physik
DY 2: Focus Session: Physics Meets ML I – Machine Learning for Complex Quantum Systems (joint session TT/DY)
DY 2.8: Vortrag
Montag, 27. März 2023, 12:30–12:45, HSZ 03
Efficient optimization of deep neural quantum states toward machine precision — •Ao Chen and Markus Heyl — Center for Electronic Correlations and Magnetism, University of Augsburg, 86135 Augsburg, Germany
Neural quantum states (NQSs) have emerged as a novel promising numerical method to solve the quantum many-body problem. However, it has remained a central challenge to train modern large-scale deep network architectures, which would be key to utilize the full power of NQSs and to make them competitive or superior to conventional numerical approaches. Here, we propose a minimum-step stochastic reconfiguration (MinSR) method that reduces the optimization complexity by orders of magnitude while keeping similar accuracy as compared to conventional stochastic reconfiguration. In this talk, I will show MinSR allows for an accurate training on unprecedentedly deep NQS with up to 64 layers and more than 105 parameters in the spin-1/2 Heisenberg J1-J2 models on the square lattice. With limited numerical resources, partly obtained on single workstations, we find that this approach yields better variational energies as compared to existing NQS results and we further observe that the accuracy of our ground state calculations approaches different levels of machine precision on modern GPU and TPU hardware.