Event generation with normalizing flows

  title={Event generation with normalizing flows},
  author={Christina Gao and Stefan Hoeche and Joshua Isaacson and Claudius Krause and Holger Schulz},
  journal={Physical Review D},
We present a novel integrator based on normalizing flows which can be used to improve the unweighting efficiency of Monte-Carlo event generators for collider physics simulations. In contrast to machine learning approaches based on surrogate models, our method generates the correct result even if the underlying neural networks are not optimally trained. We exemplify the new strategy using the example of Drell-Yan type processes at the LHC, both at leading and partially at next-to-leading order… 

Figures and Tables from this paper

Unbiased elimination of negative weights in Monte Carlo samples

We propose a novel method for the elimination of negative Monte Carlo event weights. The method is process-agnostic, independent of any analysis, and preserves all physical observables. We

Exhaustive Neural Importance Sampling applied to Monte Carlo event generation

Exhaustive Neural Importance Sampling (ENIS), a method based on normalizing flows to find a suitable proposal density for rejection sampling automatically and efficiently, is presented and how this technique solves common issues of the rejection algorithm is discussed.

Generative Networks for LHC Events

LHC physics crucially relies on our ability to simulate events efficiently from first principles. Modern machine learning, specifically generative networks, will help us tackle simulation challenges

Optimising simulations for diphoton production at hadron colliders using amplitude neural networks

This work focuses on the case of loop-induced diphoton production through gluon fusion, and develops a realistic simulation method that can be applied to hadron collider observables.

Loop Amplitudes from Precision Networks

It is shown that simple, Bayesian networks can learn such amplitudes and model their uncertainties reliably and a boosted training of the Bayesian network further improves the uncertainty estimate and the network precision in critical phase space regions.

Exploring phase space with Neural Importance Sampling

An importance sampling technique capable of overcoming typical deficiencies of existing approaches by incorporating neural networks is proposed, which guarantees full phase space coverage and the exact reproduction of the desired target distribution, in this case given by the squared transition matrix element.

Accelerating Monte Carlo event generation -- rejection sampling using neural network event-weight estimates

The generation of unit-weight events for complex scattering processes presents a severe challenge to modern Monte Carlo event generators. Even when using sophisticated phase-space sampling techniques

How to GAN Event Unweighting

This work targets a known bottleneck of standard simulations and shows how their unweighting procedure can be improved by generative networks, which can lead to a very significant gain in simulation speed.

Neural resampler for Monte Carlo reweighting with preserved uncertainties

Monte Carlo event generators are an essential tool for data analysis in collider physics. To include subleading quantum corrections, these generators often need to produce negative weight events,

A survey of machine learning-based physics event generation

The state of the art of machine learning efforts at building physics event generators in high-energy nuclear and particle physics is surveyed, and various approaches of incorporating physics into the ML model designs to overcome challenges are discussed.



How to GAN LHC events

Event generation for the LHC can be supplemented by generative adversarial networks, which generate physical events and avoid highly inefficient event unweighting. For top pair production we show how

Thread-scalable evaluation of multi-jet observables

The feasibility of a GPU-based event generator with an emphasis on the constraints imposed by the hardware is studied and some studies of Monte Carlo convergence and accuracy are presented.

Systematic event generator tuning for the LHC

In this article we describe Professor, a new program for tuning model parameters of Monte Carlo event generators to experimental data by parameterising the per-bin generator response to parameter

Simulation of vector boson plus many jet final states at the high luminosity LHC

We present a novel event generation framework for the efficient simulation of vector boson plus multi-jet backgrounds at the high-luminosity LHC and at possible future hadron colliders. MPI

MINLO: multi-scale improved NLO

A bstractIn the present work we consider the assignment of the factorization and renormalization scales in hadron collider processes with associated jet production, at next-to-leading order (NLO) in

A Generative-Adversarial Network Approach for the Simulation of QCD Dijet Events at the LHC

A Generative-Adversarial Network based on convolutional neural networks is used to simulate the production of pairs of jets at the LHC and a number of kinematic distributions both at Monte Carlo truth level and after the detector simulation can be reproduced by the generator network.

Event generation with Sherpa 2.2

This work summarises essential features and improvements of the Sherpa 2.2 release series, which is heavily used for event generation in the analysis and interpretation of LHC Run 1 and Run 2 data.

Event generation with SHERPA 1.1

In this paper the current release of the Monte Carlo event generator Sherpa, version 1.1, is presented. Sherpa is a general-purpose tool for the simulation of particle collisions at high-energy