Computational Laboratory for Energy And Nanoscience

University Homepage | Faculty of Science | Faculty of Graduate Studies | Registrar's Office |
Physics Homepage | Graduate Studies in Physics | Department contact information
Map it | City of Oshawa | Regional News | Local Weather | Government of Canada
subglobal4 link | subglobal4 link | subglobal4 link | subglobal4 link | subglobal4 link | subglobal4 link | subglobal4 link
subglobal5 link | subglobal5 link | subglobal5 link | subglobal5 link | subglobal5 link | subglobal5 link | subglobal5 link
subglobal6 link | subglobal6 link | subglobal6 link | subglobal6 link | subglobal6 link | subglobal6 link | subglobal6 link
subglobal7 link | subglobal7 link | subglobal7 link | subglobal7 link | subglobal7 link | subglobal7 link | subglobal7 link
subglobal8 link | subglobal8 link | subglobal8 link | subglobal8 link | subglobal8 link | subglobal8 link | subglobal8 link

Manuscript Summary - Phase space sampling and operator confidence with generative adversarial networks

Generative adverseraial networks are a relatively new form of deep neural networks which are based on a game between two competing sub-networks (the generator and the discriminator). The role of the generator is to trick the discriminator into thinking that it has produced "real" data. By real, we mean data that looks similar to something the discriminator has seen before. In this paper, we teach a generator to produce configurations from the Ising-model which we sampled using a number of standard techniques. The resulting generator is able to rapidly and efficiently sample phase space. Moreover, the discriminator, once trained, can predict the energy of a new configuration with a high degree of accurary (similar to our previously published work). Importantly however, it also reports a level of confidence in the prediction which correlates well with error. The discriminator is smart enough to know when it is right, and when it may be wrong. This is important when using deep neural networks for simulation.

UOIT