Skip to content
Snippets Groups Projects

CAT paper

    • Open Issue created by Louis Moureaux @lmoureau

    Try answering the following questions with up to 3 pages.

    • What is the scope of the tool?
      • Measurements of processes with asymptotic statistics
      • Differential observables with full description of correlations
    • What is the origin of the tool?
      • Jet differential measurements
    • What are the technically or conceptually innovative features of this tool?
      • Physics from the shell
      • Friends
      • Reproducibility
      • Modularity
      • Optimise debugging time rather than running time
      • Storage of systematic in objects
      • Prefix commands
      • campaigns = specific studies in the course of an analysis
    • Can you describe the typical workflow from the input data tier to the final results?
      • Assuming that all corrections have been implemented and that only the analysis has to be run:
        • Run n-tupliser (most likely via CRAB)
        • Write a scenario in shell and run it locally with a HTCondor DAG
        • Unfold / fit the data
        • Save the metadata somewhere
      • Assuming that one is starting a new analysis (e.g. existing correction but new data, or vice-versa)
        • Identify existing and missing steps
        • Debug existing steps
        • Implement missing steps
    • How do you handle metadata (e.g. cross sections, systematic variations, recipes)?
      • We store the information in the TList* retrievable with TTree::GetUserInfo()
      • Update it step by step in the course of the analysis.
      • Metadata are convertible to a config file in standard format that can be used to reproduce the same analysis.
    • Can you describe, possibly using example code fences, how the user expresses an analysis?
      • Pick a jet campaign (we can refer to CMS-NOTE-2024/1)
      • Make a minimal Drell-Yan analysis (this would illustrate that the unfolding is generic)
    • Do you have any interfaces with other tools that are worth noting?
      • TUnfold (even if it's part of ROOT)
      • libgit / Git
      • Boost (worth mentioning?)
      • correctionlib
      • HTCondor
      • combine
        \to
        may not yet be the case, but we should check how easy it is to adapt it?
    • How/where is the code maintained and documented?
      • CAT GitLab for CMS-related stuff
      • Independent GitLab repo for general functionalities (Darwin)
      • Doxygen
      • campaigns
    • Can you point to existing analyses performed with this tool?
      • SMP-20-011
      • SMP-21-006
      • SMP-21-008 (partly)
      • SMP-22-005
      • plus other ongoing analyses
    • Can you outline the future prospects?
      • Include more objects (at least electrons and MET -- tauons only on very long term)
      • Metadata merging
      • Python bindings
      • Metadata in PDF
    0 of 10 checklist items completed · Edited

    Designs

      Child items ...

      2. Activity

      Please register or sign in to reply