Deep Learning as a Parton Shower

We make the connection between certain deep learning architectures and the renormalisation group explicit in the context of QCD by using a deep learning network to construct a toy parton shower model. The model aims to describe proton-proton collisions at the Large Hadron Collider. A convolutional autoencoder learns a set of kernels that efficiently encode the behaviour of fully showered QCD collision events. The network is structured recursively so as to ensure self-similarity, and the number of trained network parameters is low. Randomness is introduced via a novel custom masking layer, which also preserves existing parton splittings by using layer-skipping connections. By applying a shower merging procedure, the network can be evaluated on unshowered events produced by a matrix element calculation. The trained network behaves as a parton shower that qualitatively reproduces jet-based observables.

Note:

26 pages, 13 figures

Phenomenological Models
Jets
parton: showers
parton: splitting
p p: scattering
neural network
quantum chromodynamics
renormalization group
CERN LHC Coll

References(36)

Figures(21)

[1]

Wavelet Analysis: Event De-noising, Shower Evolution and Jet Substructure Without Jets

J.W. Monk
(
- Bohr Inst.
)

e-Print:
- 1405.5008

[2]

An exact mapping between the Variational Renormalization Group and Deep Learning

e-Print:
- 1410.3831

[3]

Deep learning and the renormalization group

Cédric Bény

e-Print:
- 1301.3124

[4]

Criticality & Deep Learning II: Momentum Renormalisation Group

D. Oprisa
,
P. Toth

e-Print:
- 1705.11023

[5]

JUNIPR: a Framework for Unsupervised Machine Learning in Particle Physics

- Eur.Phys.J.C 79 (2019) 2, 102
•
e-Print:
- 1804.09720
•
DOI:
- 10.1140/epjc/s10052-019-6607-9

[6]

Learning Particle Physics by Example: Location-Aware Generative Adversarial Networks for Physics Synthesis

Luke de Oliveira
(
- LBNL, Berkeley
)
,
Michela Paganini
(
- LBNL, Berkeley and
- Yale U.
)
,
Benjamin Nachman
(
- LBNL, Berkeley
)

- Comput.Softw.Big Sci. 1 (2017) 1, 4
•
e-Print:
- 1701.05927
•
DOI:
- 10.1007/s41781-017-0004-6

[7]

Keras

F. Chollet

[8]

TensorFlow: Large-Scale Machine Learning on Heterogeneous Distributed Systems

et al.

e-Print:
- 1603.04467

[9]

APEMEN. Autoencoding Parton Emitting Model Encoded in Networks

J. Monk

[10]

Event generation with SHERPA 1.1

et al.

- JHEP 02 (2009) 007
•
e-Print:
- 0811.4622
•
DOI:
- 10.1088/1126-6708/2009/02/007

[11]

A Parton shower algorithm based on Catani-Seymour dipole factorisation

Steffen Schumann
(
- Dresden, Tech. U.
)
,
Frank Krauss
(
- Durham U., IPPP
)

- JHEP 03 (2008) 038
•
e-Print:
- 0709.1027
•
DOI:
- 10.1088/1126-6708/2008/03/038

[12]

AMEGIC++ 1.0: A Matrix element generator in C++

F. Krauss
(
- Cambridge U.
)
,
R. Kuhn
(
- Dresden, Max Planck Inst. and
- Dresden, Tech. U.
)
,
G. Soff
(
- Dresden, Tech. U.
)

- JHEP 02 (2002) 044
•
e-Print:
- hep-ph/0109036
•
DOI:
- 10.1088/1126-6708/2002/02/044

[13]

Comix, a new matrix element generator

Tanju Gleisberg
(
- SLAC
)
,
Stefan Hoeche
(
- Durham U., IPPP
)

- JHEP 12 (2008) 039
•
e-Print:
- 0808.3674
•
DOI:
- 10.1088/1126-6708/2008/12/039

[14]

QCD matrix elements and truncated showers

- JHEP 05 (2009) 053
•
e-Print:
- 0903.1219
•
DOI:
- 10.1088/1126-6708/2009/05/053

[15]

Parton distributions for the LHC Run II

NNPDF

Collaboration

•

Richard D. Ball
(
- U. Edinburgh, Higgs Ctr. Theor. Phys. and
- CERN
)

et al.

- JHEP 04 (2015) 040
•
e-Print:
- 1410.8849
•
DOI:
- 10.1007/JHEP04(2015)040

[16]

The anti- $k_t$ jet clustering algorithm

- JHEP 04 (2008) 063
•
e-Print:
- 0802.1189
•
DOI:
- 10.1088/1126-6708/2008/04/063

[17]

FastJet User Manual

Matteo Cacciari
(
- Paris, LPTHE and
- Diderot U., Paris
)
,
Gavin P. Salam
(
- CERN and
- Princeton U. and
- Paris, LPTHE
)
,
Gregory Soyez
(
- Saclay, SPhT
)

- Eur.Phys.J.C 72 (2012) 1896
•
e-Print:
- 1111.6097
•
DOI:
- 10.1140/epjc/s10052-012-1896-2

[18]

Incorporating Nesterov Momentum into Adam, in proceedings of the International Conference on Learning Representations(ICLR), Caribe Hilton, San Juan, Puerto Rico, 2-4 May

T. Dozat

[19]

An overview of gradient descent optimization algorithms

Sebastian Ruder

e-Print:
- 1609.04747

[20]

Understanding the difficulty of training deep feedforward neural networks, in proceedings of Thirteenth International Conference on Artificial Intelligence and Statistics (AISTATS), Chia Laguna Resort, Sardinia, Italy, 13-15 May, volume 9, Y.W. Teh and M. Titterington (eds.) / PMLR , pp. 249-256