Dresden 2017 – wissenschaftliches Programm

Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe

DY: Fachverband Dynamik und Statistische Physik

DY 26: Posters - Statistical Physics, Stochastic Thermodynamics

DY 26.5: Poster

Dienstag, 21. März 2017, 18:15–21:00, P3

Machine learning quantum phases of matter beyond the fermion sign problem — •Peter Broecker¹, Simon Trebst¹, Juan Carrasquilla², and Roger Melko^2,3 — ¹Institute for Theoretical Physics, University of Cologne, 50937 Cologne, Germany — ²Perimeter Institute for Theoretical Physics, Waterloo, Ontario N2L 2Y5, Canada — ³Department of Physics and Astronomy, University of Waterloo, Ontario, N2L 3G1, Canada

Many-fermion systems exhibit some of the most intriguing physical phenomena in condensed matter physics such as the formation of non-Fermi liquids, superconductivity, or Mott insulators with fractionalized excitations. Exact analytical methods for the study of such fascinating phenomena are scarce and numerical tools are called for. However, one of the most powerful methods at our disposal - quantum Monte Carlo sampling - is often hampered by the fermion sign problem, which manifests itself most strikingly in the exponential growth of statistical errors and thus an exponential slowdown of the sampling. Here, we present the use of artificial neural networks as a powerful approach that is able to learn and distinguish quantum phases of matter despite the sign problem.