\n

DTSTAMP:20210416T032445Z
END:VEVENT
END:VCALENDAR\n

\n\n

\nGiven access to accurate solutions of the many-electron Schrödinger equation\, most of condensed matter physi cs\, chemistry and materials physics could be derived from first principle s. Exact wave functions of systems with more than a few electrons are out of reach because they are NP-hard to compute in general\, but approximatio ns can be found using polynomially scaling algorithms. The key challenge f or many of these algorithms is the choice of an approximate parameterized wave function\, which must trade accuracy for efficiency. Neural networks have shown impressive power as practical function approximators and promis e as a way of representing wave functions for spin systems\, but electroni c wave functions have to obey Fermi-Dirac statistics. This talk introduces a new deep learning architecture\, the Fermionic neural network\, which i s capable of approximating many-electron wavefunctions and greatly outperf orms conventional approximations. The use of FermiNet wave functions boost s the accuracy of the simple and appealing variational quantum Monte Carlo method until it rivals the very best conventional quantum chemical approa ches. It is also expected to scale much better with system size\, opening the possibility of understanding previously intractable many-electron syst ems.

\n