ATLAS+CMS EFT fitting exercise meeting

Informal CMS/ATLAS discussions
Zoom Meeting ID
Nicolas Berger
Useful links
Join via phone
Zoom URL

Treatment of theory uncertainties

  • Some theory uncertainties on the input processes are already baked into the covariance matrices (all uncertainties for Higgs, uncertainties on acceptance only for SM fiducial cross-sections)
  • EFT parameterization uncertainties are not considered for the moment
  • Computing scale variations on EFT predictions should be possible in the MG setups used to derive the parameterizations.
  • Technically EFT uncertainties are easy to add (the machinery for adding them through nuisance parameters is already available), but this would require a recipe on how to treat uncertainty correlations (between coefficients, measurement regions, etc.).

YAML format

  • Similar formats from CMS and ATLAS, will try to converge on a common version
  • Preference for using dictionaries whenever possible as this is self-documenting and easier to extend if we want to add more information.


  • Need to define a set of tools to use to run the full chain, ideally with proper version numbers, etc.
  • In CMS the EFT2Obs package contains all the necessary tools
  • ATLAS looking at solutions based on a docker image containing the necessary tools. Not in common use by the theory community, but could at least extract the tool information from the docker config.
  • Preference for moving away from ROOT whenever possible
  • Preference for not relying on CVMFS (the shared filesystem used by ATLAS/CMS to access common tool repositories for LCG grid computing) as this is not widely usable by theorists

Configurations for EFT generation

  • Need to ensure the full generation config is specified through MG cards
  • Will try to harmonize ATLAS and CMS options as much as possible
  • Will get in touch with Area 2 which is probably the better home for this effort.

Truncation tests

  • Tests on pT or angular variables (even less correlated with the q^2) should provide interesting information
  • Could also consider interpretations in a particular UV model, comparing the straight-to-UV case (measurement -> UV) with the two step interpretation through SMEFT (measurement -> SMEFT -> UV). For the latter, some truncations could be investigated and compared with the straight-to-UV case.

Action Items


  1. Document all the links to the code / inputs on the Area 4 twiki page
  2. Converge on a yaml format for parameterisations and measurements
  3. Sets of yaml files from each experiment will then be exchanged
  4. Perform a quick cross-check of the fit to each parameter from the CMS+ATLAS combination
  5. In parallel CMS will improve the Higgs parameterisations (e.g. to include all STXS bins and the decay side)


There are minutes attached to this event. Show them.
    • 11:00 AM 11:05 AM
      Introduction 5m
      Speakers: Andrew Gilbert (Northwestern University (US)), Nicolas Berger (Centre National de la Recherche Scientifique (FR))
    • 11:05 AM 11:25 AM
      CMS presentation 20m
      Speaker: Andrew Gilbert (Northwestern University (US))
    • 11:25 AM 11:45 AM
      ATLAS presentation 20m
      Speaker: Oleksii Lukianchuk (Université Paris-Saclay (FR))
    • 11:45 AM 12:05 PM
      Discussion 20m