Make mixing dataset public

Big Picture

HH4b important for self-coupling (systematics drive sensitivity / particulary on Multi-jet background)

Multi-jet backgroudn data driven QCD MC unacceptable

  • Techniques for Data-driven background important Usually some for ABCD
  • Interesting ML application Transfer learning
  • HH4b important long-term channel. Dominated by systematics on Multijet QCD unacceptable
  • Mixed data 15x stats… signal free.
    • Allow bias of exaplotion and varince of method to be
  • Could be a nice use case for a data challege
  • Used in CMS publications
  • VAlidate ATLAS method (Cross check systgemstisc)

Need

  • 3b data and mixed data

To do

  • [ ] Show signal injection

IDes

  • Some good ideas: Mixing paper ==> public dataset ==> dataset challenge ==> implement summary-vars / FvT / OT ? Probably also need some SvB function that people can call
  • CMS publication rules:

https://cms-docdb.cern.ch/cgi-bin/DocDB/RetrieveFile?docid=14372&filename=Limited_Authorship_Rules_and_Guidelines_v1.9.pdf&version=7

Talk in data preservation and open acceess meeting

Open data chat Thomas Mc Cauley <thomas.mccauley@cern.ch>

  • Slides: what to release and why
  • [>>] June 1st
  • Will make slides for June 1st… then there is a CB 2nd week of June
  • Would like Run2: Mixed data and 3b data (trigger XXXX) Could derive independently a SvB