SciPost Submission Page
Measuring QCD Splittings with Invertible Networks
by Sebastian Bieringer, Anja Butter, Theo Heimel, Stefan Höche, Ullrich Köthe, Tilman Plehn, Stefan T. Radev
This is not the latest submitted version.
This Submission thread is now published as
|Authors (as registered SciPost users):
|Sebastian Bieringer · Stefan Höche · Tilman Plehn
QCD splittings are among the most fundamental theory concepts at the LHC. We show how they can be studied systematically with the help of invertible neural networks. These networks work with sub-jet information to extract fundamental parameters from jet samples. Our approach expands the LEP measurements of QCD Casimirs to a systematic test of QCD properties based on low-level jet observables. Starting with an toy example we study the effect of the full shower, hadronization, and detector effects in detail.
Submission & Refereeing History
You are currently on this page
Reports on this Submission
- Cite as: Anonymous, Report on arXiv:2012.09873v1, delivered 2021-02-17, doi: 10.21468/SciPost.Report.2578
The authors present a machine learning approach to study QCD splittings with conditional and invertible neural networks. I consider their work a useful extension of the growing machine learning literature in high-energy physics. However, there are several points that the authors should address before I can recommend their work for publication.
- Fundamental parameters in QCD (alpha_s or the quadratic SU(3) Casimir invariants CA/CF, as mentioned in the introduction) are typically extracted from cross sections which can be calculated perturbatively to NNLO+NNLL or better. Using a parton shower, where we do not have the same level of perturbative accuracy, would make such precision extractions difficult. Therefore, it seems that high-level but well-controlled observables are a better choice instead of trying to make use of all the low-level information that we get from collider experiments. The discussion in the introduction about precision extractions of CA/CF from LEP data and the relation to the proposed framework of the authors should be further clarified.
- From the introduction, it appears that it is generally difficult to understand the main purpose of the paper. Is it precision extractions, as mentioned above, or the tuning of parameters of parton showers (here, the prefactors introduced in the parametrized splitting functions)? Moreover, it would be helpful for the reader if the authors can comment in more detail on the relationship of their work to the available literature.
- Equation 14 and the sentence before equation 14, “all sub-leading jets are ignored”: Are the authors referring to reconstructed jets or partons from the shower? If these are reconstructed jets, the algorithm and jet radius should be specified. The same question appears in other parts of the paper.
- In the outlook, the authors write that the decomposition of the splitting function in equation 11 corresponds to ``logarithmically enhanced, finite and rest’’ terms. This should be stated more clearly earlier in the paper. How exactly is the functional form of the parametrized splitting functions (in terms of y, z) obtained, especially the term that vanishes in QCD (~C_ij)? I suppose that the decomposition of the logarithmically enhanced terms and the finite terms is not unique and partial fractioning can be used to reshuffle the terms?
- Lastly, I have a question about the k_T sorting which is mentioned several times throughout the paper. Are the authors referring to the ordering from the shower (which is not accessible in experiments like the ``truth sorting’’) or with the help of a clustering/declustering procedure of a jet algorithm (which is accessible experimentally but not directly related to the actual shower history)? Do the considered partons correspond to particles inside a reconstructed jet or to the entire event?