we have the pleasure to invite you to the Particle Theory Seminar on MONDAY,
October 17 at 12:15, which will be given by
David Albandea (University of Valencia)
“Learning trivializing maps“
Abstract: The recent introduction of machine learning techniques, especially normalizing flows, for the sampling of lattice gauge theories has shed some hope on improving the sampling efficiency of the traditional HMC algorithm. However, naive usage of normalizing flows has been shown to lead to bad scaling with the volume. In this talk I will introduce normalizing flows commenting on its computational cost scaling towards the continuum. I will then propose using local normalizing flows at a scale given by the correlation length: even if naively these transformations have a very small acceptance, when combined with HMC lead to algorithms with high acceptance and reduced autocorrelation times compared with HMC.
The seminar will be in a hybrid mode, local people are invited to meet in the seminar room D-2-02.