Preparation of DAS short exercises

Europe/Berlin
SR 4a/b

SR 4a/b

Description

https://cern.zoom.us/j/64087077370?pwd=bX5hTzCoYEwMnTsUaT4MX0JPxwnn8l.1

General points:

  • Please focus on Run 3 if you can

  • Test test test! 

  • How much space do you need?

  • At least for long exercises: make sure the space is used well, like installing one thing in one place and another thing in another space. No 100 GB AFS space available for participants :)

 

MS: preparations for the students

 

Computing ressources:

  • Can we assure that everyone has a CERN account? Better to use special NAF accounts?

    • Aliya: during stat school, major issues with NAF ressources

    • Alex: last time, very poor support

    • Dominic: in pre-exercises, ask them to clean up home directory to get most of 10 GB space back

  • For each exercise, give an estimate of how much space needed → students might need to clear space on afs in between exercises

 

Combine (Kuan):

  • Aliya: ask students to install docker in pre-exercises

  • Good point: test the exercises beforehand!

 

ML (Joern, Matthias Komm)

  • Need priority access to GPUs, with SWAN notebooks

  • CERN can add temporary more, but needs to happen before the US wakes up at 3pm

  • Alexi wants to attend, try not to overlap with tau exercise

  • Code disentangled from CMSSW

 

Tracking (Jan, Marco, Brunella)

  • Tested exercises yesterday, lxplus is fine, runs at a minute per execution

  • Produce just 1 output file of a couple MB he thinks (< 1GB) and could use /tmp

 

e/gamma

  • Will change to Run 3 🙂 - general point as well

  • miniAOD

  • Use lxplus and swan

  • Ying (contract to Oct 31) and Alberto (but contract might be up, but will know by end of Sept) → best to find another facilitator

  • Not big output

 

Jets (Andreas Hinzmann)

  • Need 1-2 extra tutors: Patrick too busy on long exercise? Dennis? 

  • Need to interface with Patrick on related long exercise

  • Output just plots don’t expect GBs

 

BTV (Phillip, )

  • Decided not to do ML, lxplus/SWAN sufficient

 

Taus (Alexei)

  • Based on nanoAOD

  • Run on 2022 data

  • Tested it yesterday

  • Run totally in notebooks, no prep for the students

 

Generators (Dominic)

  • Dont need to download madgraph beforehand

  • Brief Rivet intro, not covered by Aliya

  • All exercises on Tuesday (Dominic has shifts)

  • Mikel Morentin as 2nd facilitator, maybe can use another

  • Runs entirely on lxplus/with CERN account

 

REANA (Pallibi)

  • Within 2 hours

  • Previous exercise exists

  • Call the exercise “REANA (CAT)”

 

Gitlab CI/CD (Aliya)

  • Ask students to setup gitlab ssh keys in pre-exercise

  • Would benefit from another facilitator (Matthias: Lukas Ebeling, PhD student from UHH can help; in contact now with Aliya)

  • Call the exercise “Gitlab CI/CD (CAT)”

 

Call the exercise “Applying central corrections (CAT)”

 

Add to the pre-exerises: 

  • Eduroam

    • Science hot spot also available, but may not be great

    • DESY guest network?

  • Set up the grid, can run a simple script

There are minutes attached to this event. Show them.
    • 09:00 09:10
      Introduction 10m
      Speakers: Alexander Grohsjean (CMS - University of Hamburg), Alexander Grohsjean (University of Hamburg), Juliette Alimena (CMS (CMS Fachgruppe Searches)), Maria Aldaya Martin (CMS (CMS Fachgruppe TOP)), Matthias Schroeder (Universität Hamburg)
    • 09:10 09:15
      Combine - limit setting 5m
      Speaker: Kuan-Yu Lin (CMS (CMS Fachgruppe Searches))
    • 09:15 09:20
      Combine - unfolding 5m
      Speakers: Beatriz Ribeiro Lopes (CMS (CMS Fachgruppe TOP)), Jan Kieseler (KIT), Jan Kieseler (KIT)
    • 09:20 09:25
      Machine learning 5m
      Speakers: Joern Bach (CMS DESY), Matthias Komm (DESY (Deutsches Elektronen Synchrotron))
    • 09:25 09:30
      Tracking 5m
      Speakers: Ms Brunella D'Anzi (University of Bari Aldo Moro), Jan Gerrit Schulz (CMS (CMS-Experiment)), Marco Musich
    • 09:30 09:35
      E/gamma 5m
      Speakers: Alberto Belvedere (CMS (CMS Fachgruppe QCD)), Ms Ying An (CMS (CMS Fachgruppe TOP))
    • 09:35 09:40
      Muons 5m
      Speaker: Matteo Bonanomi (UNI/EXP (Uni Hamburg, Institut fur Experimentalphysik))

      Every physics analysis builds under the assumption that Monte Carlo simulations reliably describe the Data collected at the LHC.

      Despite this assumption, it is quite hard to mimic in the Monte Carlo all the conditions of Data taking exactly. As a result of this, the selection efficiency of the different objects reconstructed by the experiment has some differences between Data and simulation.

      In this exercise we focus on the muons, a fundamental object for many physics analyses, and the second letter in the CMS acronym ;)

      We are going to see how the CMS Muon Physics Objects Group (POG) measures selection efficiencies in Data and simulation, learning the basics of a technique used in many other POGs: the Tag and Probe.

      After measuring the efficiencies, we will derive the Scale Factors, i.e. ratios between data and MC efficiencies, which are applied as correction factors to our simulations to improve their fidelity in the description of the Data.

    • 09:40 09:45
      Jets 5m
      Speakers: Andreas Hinzmann (CMS (CMS Fachgruppe Searches)), Dennis Schwarz (Austrian Academy of Sciences)

      Reuse the previous iteration of the exercises.

      Explain concepts in pieces (probably 30% of the time).

      - First exercises are in Jupyter with matplotlib.
      - Some exercises with CMSSW, e.g. dumpEventContent
      - Some exercises in C++, e.g. MET filters
      - covering an understanding of AOD/MiniAOD/NanoAOD chain, to understand where are the algorithms leading to the NanoAOD content

      Need 1-2 extra tutors.

    • 09:45 09:50
      Flavor tagging 5m
      Speakers: Philipp Gadow (UNI/EXP (Uni Hamburg, Institut fur Experimentalphysik))), Uttiya Sarkar
    • 09:50 09:55
      Taus 5m
      Speakers: Alexei Raspiareza (CMS (CMS Fachgruppe HIGGS)), Andrea Cardini (None)
    • 09:55 10:00
      Generators 5m
      Speaker: Dominic Stafford (CMS (CMS Fachgruppe Searches))
    • 10:00 10:05
      REANA 5m
      Speakers: Aliya Nigamova (University of Hamburg), Pallabi Das (Princeton University)
    • 10:05 10:10
      Gitlab CI 5m
      Speaker: Aliya Nigamova (University of Hamburg)
    • 10:10 10:15
      Applying central corrections 5m
      Speaker: Daniel Savoiu (UNI/EXP (Uni Hamburg, Institut fur Experimentalphysik))
    • 10:15 10:20
      Outreach 5m
      Speaker: Freya Blekman (DESY/University of Hamburg)

      In this exercise the basics of science communication and outreach are introduced, both for general particle physics and for CMS in general. One of the important aspects of science communication (and communication in general ) is stakeholder adjustment, and this is what this exercise focuses on. When finishing the exercise the participants will be aware of the activities in outreach and science communication in CMS, what opportunities there are for physicists at all career levels in CMS, and how to develop a communication plan for a physics result with a specific target audience in mind.

       

      Outreach short exercise planned as follows:

       

      (15 mins/presentation) - introduction on general ideas Science communication and Outreach in hep 

      (15 mins/presentation) - introduction CMS outreach and communications

      (60 mins/small teams) - develop communications strategy for a representative high-impact CMS result (we will use most recent CERN courier "energy frontiers" article by CMS), in small teams, and all for (randomly selected) different audiences. This article will also be given as 'preparation/homework' to the participants but is short enough to read on the day.

      (30 mins/remaining time/group discussion) - compare strategies between small teams, discuss pros/cons, discuss how to use in future.

      ---

      facilitators: 

      F. Blekman

      Likely one more facilitator needed, still investigating with CMS communications team if someone is avialable to come to Hamburg

       

      --- 
      Computing needs: minimal (room with projector needed)