P29: A Deep Learning Tool for Fast Simulation
SessionPoster Reception
Event Type
ACM Student Research Competition
Poster
Reception
TimeTuesday, November 14th5:15pm -
7pm
LocationFour Seasons Ballroom
DescriptionWe present the first application of Volumetric
Generative Adversarial Network (VGAN) to High Energy
Physics simulation. We generate three dimensional images
of particles depositing energy in calorimeters. This is
the first time such an approach is taken in HEP where
most of data is three dimensional in nature but it is
customary to convert it into two dimensional slices. The
volumetric approach leads to a larger number of
parameters, but two dimensional slicing loses the
volumetric dependencies inherent in the dataset. The
present work proves the success of handling those
dependencies through VGANs. Energy showers are
faithfully reproduced in all dimensions and show a
reasonable agreement with standard techniques. We also
demonstrate the ability to condition training on several
parameters such as particle type and energy. This work
aims at proving Deep Learning techniques represent a
valid fast alternative to standard MonteCarlo approaches
and is part of the GEANTV project.
