SciPost logo

SciPost Submission Page

Event Generation and Density Estimation with Surjective Normalizing Flows

by Rob Verheyen

This Submission thread is now published as

Submission summary

As Contributors: Rob Verheyen
Arxiv Link: https://arxiv.org/abs/2205.01697v2 (pdf)
Code repository: https://github.com/rbvh/surflows
Date accepted: 2022-07-05
Date submitted: 2022-06-17 08:23
Submitted by: Verheyen, Rob
Submitted to: SciPost Physics
Academic field: Physics
Specialties:
  • High-Energy Physics - Phenomenology
Approaches: Computational, Phenomenological

Abstract

Normalizing flows are a class of generative models that enable exact likelihood evaluation. While these models have already found various applications in particle physics, normalizing flows are not flexible enough to model many of the peripheral features of collision events. Using the framework of Nielsen et al. (2020), we introduce several surjective and stochastic transform layers to a baseline normalizing flow to improve modelling of permutation symmetry, varying dimensionality and discrete features, which are all commonly encountered in particle physics events. We assess their efficacy in the context of the generation of a matrix element-level process, and in the context of anomaly detection in detector-level LHC events.

Published as SciPost Phys. 13, 047 (2022)



List of changes

All points raised by the referees have been addressed.


Reports on this Submission

Anonymous Report 2 on 2022-7-1 (Invited Report)

Report

The author has addressed all the points that I raised to my satisfaction. The explanations in the manuscript improved and are more clear now. I can recommend this paper for publication.

  • validity: high
  • significance: top
  • originality: high
  • clarity: high
  • formatting: excellent
  • grammar: excellent

Anonymous Report 1 on 2022-6-23 (Invited Report)

Report

The author has incorporated and answered all my requests satisfactorily. Therefore, I recommend the paper for publication

  • validity: top
  • significance: high
  • originality: good
  • clarity: high
  • formatting: excellent
  • grammar: excellent

Login to report or comment