DS10 - Défi de tous les savoirs

Optimal use of jets at LHC and future colliders – OptimalJets


The Large Hadron Collider operating at CERN explores the frontiers of our knowledge o fparticle physics. «Jets« which emerge from the high-energy quarks and gluons produced by the collisions are a major tool for probong fundamental interactions. The OptimalJets project will develop our knowledge of the substructure of jets, a field increasingly relied upon at for the LHC physics program.

Develop the foundations for the optimal use of jets in the coming decade and beyond

Boosted jets have attracted a lot of attention over the last few years and are already used at the LHC. However, the current state of the art is mostly empirical, in that jet substructure techniques are studied using simulated Monte Carlo events. This presents several fundamental limitations. The originality of our proposal is to take the jet substructure field to the next level by anchoring new, robust, tools in a first-principle understanding based on the theory of strong interactions.<br /><br />The objectives of the projects are generically as follows: (1) understand from first-principle how existing jet-substructure tools work, (2) use this knowledge to develop new, more efficient, tools, (3) provide precise, robust, predictions for jet substructure observables, (4)provide high-quality, open-source, software implementation of our new tools. The new methods will be both more powerful (i.e. with better signal-to-background ratio), and more robust (i.e. with controlled theoretical uncertainties). <br /><br />This project has ramifications beyond the field of boosted jets, including quark/gluon discrimination, pileup mitigation, precise measurement of the parameters of the Standard Model, heavy-ion collisions and machine learning.

The key element we will use to reach our objectives is that instead of relying on simulated events we will use first-principles in the theory of strong interactions. This approach is key to get a firm grasp on the dominant physics effects which drive the performance of jet substructure tools. Once we have this firm understanding we can venture into providing new tools which make the best use of this fundamental understanding:

In practice the project will focus on two systems: two-prong decays (relevant for electroweak bosons like the W/Z or Higgs bosons) and three-prong decays (typically relevant for boosted top quarks).

In both case we will organise our studies in three steps. We will first gain some insight by understanding the behaviour of existing tools from first principles. We will then develop new tools with improved performance. Finally we will provide, whenever possible, precise calculations in QCD with controlled uncertainties.

Whenever possible, we will also investigate the consequences of our work beyond their applications to boosted jets.

As of today, we have already obtained numerous results in the context of the OptimalJets project.

Regarding two-pringed studies, we have at the very lease reached our objectives: we have studied basic observables [1], proposed two new methods [3,4] and provided precision calculations [5,7,12].

Regarding boosted tops, we have recently finalised our first study [11] and we are actively working on an interesting extension.

We have also obtained a large number of results in opther aspects of jet physics and jet substructure:
- we have contributed t oa systematic study of the discrimination between quark and gluon jets [6],
- we have studied teh small-R limit of jest [2],
- we have introduced a new «grooming« technique with a large potential range of application, called Recursive SoftDrop [9],
- we have introduced a very powerful method to characterise the internal dynamics of jets [10]. This is under study in the LHC experimental groups and we actively work on several extensions.
- we work on the development of jets in heavy-ion collisions [8].
- finally, we have contributed to a sereis of lecture notes on jet substructure [13]

We should also note that we have organised the BOOST meeting in July 2018 in Paris.

For the remaining time of the project, we will mostly focus on finalising on-going projects.

This includes a to-quark study where one imposes both the presence of 3 hard prongs ad a constraint on the internal radiation

We are also studying several results related to our new and promising «Lund Plane« approach developed in [9].

Moreover, we are currently finalising a study of a potential measurement of the strong coupling constant at LEP using SoftDrop. We foresee an extension of our results to NNLL accuracy.

In a different context, we are studying basic porperties of th einclusive jet cross-section.

Finally, we study jet substructure observables in heavy-ion collisions, with some results on «zg« to appear soon..

[1] Jet shapes for boosted jet two-prong decays from first-principles, Mrinal Dasgupta, Lais Schunk, Gregory Soyez, JHEP 1604 (2016) 166 [arXiv:1512.00516]
[2] Inclusive jet spectrum for small-radius jets, Mrinal Dasgupta,. Frédéric A. Dreyer, Gavin P. Salam, Gregory Soyez, JHEP 1606 (2016) 057 [arXiv:1602.01110]
[3] Improved jet substructure methods: Y-splitter and variants with grooming, Mrinal Dasgupta, Alexander Powling, Lais Schunk, Gregory Soyez, JHEP 1612 (2016) 079 [arXiv:1609.07149]
[4] Dichroic subjettiness ratios to distinguish colour flows in boosted boson tagging, Gavin P. Salam, Lais Schunk, Gregory Soyez, JHEP 1703 (2017) 022 [arXiv:1612.03917]
[5] A study of jet mass distributions with grooming, Simone Marzani, Lais Schunk, Gregory Soyez, JHEP 1707 (2017) 132 [arXiv:1704.02210]
[6] Systematics of quark/gluon tagging, Philippe Gras et al., JHEP 1707 (2017) 091 [arXiv:1704.03878]
[7] The jet mass distribution after Soft Drop, Simone Marzani, Lais Schunk, Gregory Soyez, Eur.Phys.J. C78 (2018) no.2, 96 [arXiv:1712.05105]
[8] Vacuum-like jet fragmentation in a dense QCD medium, P. Caucal, E. Iancu, A.H. Mueller, G. Soyez, Phys.Rev.Lett. 120 (2018) 232001 [arXiv:1801.09703]
[9] Recursive Soft Drop, Frédéric A. Dreyer, Lina Necib, Gregory Soyez, Jesse Thaler, JHEP 1806 (2018) 093 [arXiv:1804.03657]
[10] The Lund Jet Plane, Frédéric A. Dreyer, Gavin P. Salam, Grégory Soyez, JHEP 1812 (2018) 064 [arXiv:1807.04758]
[11] Top tagging : an analytical perspective, Mrinal Dasgupta, Marco Guzzi, Jacob Rawling, Gregory Soyez, JHEP 1809 (2018) 170 [arXiv:1807.04767]
[12] Computing N-subjettiness for boosted jets, Davide Napoletano, Gregory Soyez, JHEP 1812 (2018) 031 [ arXiv:1809.04602]
[13] Looking inside jets: an introduction to jet substructure and boosted-object phenomenology, Simone Marzani, Gregory Soyez, Michael Spannowsky to appear in Springer Lecture Notes, arXiv:1901.10342

The Large Hadron Collider (LHC) at CERN explores the frontier of particle physics by colliding protons. Quarks and gluons are abundantly produced in high-energy collisions and are a major handle to probe fundamental interactions and discover new physics. Practically, quarks and gluons are observed in the form of collimated showers of particles, called jets. We have established the current framework to reconstruct jets at the LHC.

With the higher energies reached at the LHC, massive objects such as W/Z or Higgs bosons or the top quark can be produced boosted, that is with an energy much larger than their mass. When they decay via strong interactions, their decay products will be collimated, and they will be seen as a single jet. This is a shift from the standard paradigm where a "jet" is now a proxy for more than just a quark or a gluon. To maximise the LHC potential one must be able to isolate the rare boosted jets coming from the decay of massive objects from the much more frequent quarks and gluons. This is achieved by exploiting the substructure of the jets, i.e. the properties of their constituents.

Jet substructure has been a growingly active field of research during Run I of the LHC, with many methods proposed, mainly by the theory part of the community, and many validations and measurements performed on the experimental side. Today, we therefore have evidence that jet substructure techniques work and help in identifying boosted jets.

Currently, the methods are usually proposed and tested numerically, based on event simulated using Monte-Carlo event generators. This empirical approach suffers from severe limitations: it does not explain why a method works better than another; it makes it hard and time-consuming to explore vast parameter spaces and extrapolate across a wide energy range; it does not allow for robust estimations of the theoretical uncertainties; it does not explain subtle differences between Monte-Carlo simulations and actual experimental data and, above all, it shines no light on how to improve existing techniques.

The objective of this project is to alleviate the above limitations by providing a first-principle understanding of jet substructure based on the theory of strong interactions. First, this means understanding existing techniques analytically. Our major goals are then to use this understanding to (i) develop new, optimised, jet substructure methods and (ii) provide precise calculations to estimate the theoretical uncertainties and gauge the robustness of each method. The analytic approach that we will use has been proven very successful in preliminary studies. This also means that our project is not only feasible but also very promising.

Besides the main, crucial, application to boosted-objects identification, this project can also have more generic implications on optimisation of final-state information at the LHC such as pileup mitigation or jet vetoes.

Furthermore, we shall also provide high-quality open-source software implementations of our tools in order to make them easily accessible by the relevant scientific community.

Altogether, our project will bring the field of jet substructure to a new level. It will set the framework of this rapidly growing field for the next decade and beyond. This has an impact on a broad range of analyses at the LHC and will improve its discovery potential. The experience of the members of this team in jet physics puts us in a unique position to achieve our goals.

Project coordination

Gregory Soyez (Institut de Physique Théorique (IPhT))

The author of this summary is the project coordinator, who is responsible for the content of this summary. The ANR declines any responsibility as for its contents.


LPTHE Laboratoire de Physique Théorique des Hautes Energies (LPTHE)
IPhT Institut de Physique Théorique (IPhT)

Help of the ANR 208,000 euros
Beginning and duration of the scientific project: September 2016 - 36 Months

Useful links

Explorez notre base de projets financés



ANR makes available its datasets on funded projects, click here to find more.

Sign up for the latest news:
Subscribe to our newsletter