2012 datasets - initial considerations

[katilp]

Notes for CMS 2012 data:

Collision data: dataset=/*/*2012*22Jan2013*/AOD (total of 1.1 PB out of which 510 TB Parked - see https://cds.cern.ch/record/1480607/files/DP2012_022.pdf and also https://profmattstrassler.com/articles-and-posts/lhcposts/triggering-advances-in-2012/data-parking-at-cms/ :smiley: ) A: from the start to May 6: 45 TB (including 2 TB Parked) B: from May 12 to June 18: 230 TB (95 TB Parked) C: from July 1 to September 27: 354 TB (155 TB Parked) D: from September 28 to December 5: 477 TB (260 Parked)

NB (dates approximate, some double-counting in data volumes because of special HZZ, TOPElePlusJets and TOPMuPlusJets processings with 42.9 TB)

TBD:

• [x] Check if it is justified not to include Parked datasets
• [x] Decide whether release A+B+C (appr 14fb-1) or B+C (appr 13fb-1) or C+D (appr 16 fb-1): all these combinations less than 0.5 PB (without Parked)

MC dataset=/*/*Summer12_DR53X*V7*/AODSIM (total of 2 PB) or dataset=/*/*Summer12_DR53X*V19*/AODSIM (total of 1 PB)

TBD:

• [x] check with the experts if these are overlapping:

No, the legacy campaign for the 2012 MC (called Summer12DR53X) is the one using the GT: START53_V19 (757.6 TB)

In that campaign there were also some additional special production (mostly run-dependent) using other GTs:

• START53_V19F: BPH run-dependent DIGI-RECO in 5.3. (162.8 TB)
• START53_V19E : digi-reco in 5.3 with specific Muon alignement for EXO (1.3 TB)
• START53_V7N : run dependent for h2gg (254.6 TB)

Open question for V19D (38 TB)

(From Gianluca Germinara and ppd)

• [x] check if there is any obvious rule to decide what to leave out
• [x] check with the 2011 "algorithm" how to divide these in categories
• [x] check with @tiborsimko that all 2011 procedures scale with the increased number of datasets

[katilp]

Include Parked data to the release

[RaoOfPhysics]

@katilp: I'm probably missing something, but I thought I should confirm anyway that you mean 0.5 TB for the second TBD and not 0.5 PB.

[katilp]

@RaoOfPhysics: good point, thanks; I mean 0.5 PB

[katilp]

Recapitulating numbers for the resource request:

Data: 2012 data taking was divided in four runs: RunA, RunB, RunC and RunD with a total of 1.1 PB out of which a part will be released in 2017. The released data will be at maximum 831 TB - in case of releasing RunC + RunD - and at minimum 477 TB - if only RunD is released , other combinations are possible, and to be defined considering the best possible software compatibility with the MC samples. In long-term, CMS may consider releasing the full data, if so decided by the collaboration board, but this will not happen in 2017.

MC: The legacy campaign for the 2012 MC (called Summer12DR53X) is the one using the GT: START53_V19 and sums up to 757.6 TB. In addition, some special productions should be kept (run dependent production for B physics with 162.8 TB, special alignment for muons with 1.3 TB and Higgs to gg with 254.6 TB), summing up to a total of 1.2 PB.

CMS therefore needs for the long-term preservation and open access of these samples 2 PB of disk space in eospublic at CERN, to be served through xrootd (or direct download) from opendata.cern.ch.

[katilp]

Storage space for 2 PB OK for 2017, so we can proceed 😃

[katilp]

Listing of https://cmsweb.cern.ch/das/request?view=plain&limit=3000&instance=prod%2Fglobal&input=dataset%3D%2F*%2FSummer12_DR53XV19*%2FAODSIM

Summer12_DR53X-PU_S10_START52_V19_listing.pdf

[katilp]

For the Run periods to be released, a good choice could be RunB + RunC. It would amount to 584 TB, and include most of the data that were included in the Higgs discovery analysis. To be discussed with the CMS physics coordination.

Update: From Higgs -> 4l point of view, no contradiction for RunB + RunC. RunD has some more pile-up, but would still be good as well. (from Andre Mendes)

https://cmsweb.cern.ch/das/request?view=list&instance=prod%2Fglobal&input=dataset%3D%2F*%2F2012B22Jan2013*%2FAOD+ 31 datasets Run2012B-22Jan2013_listing.pdf

https://cmsweb.cern.ch/das/request?view=list&instance=prod%2Fglobal&input=dataset%3D%2F*%2F2012C22Jan2013*%2FAOD+ 39 datasets Run2012C-22Jan2013_listing.pdf

NB the list of triggers in each dataset in https://fwyzard.web.cern.ch/fwyzard/hlt/2012/dataset

Confirmed that they are all proper datasets for physics, information also from http://inspirehep.net/record/1467921/files/10.1016_j.nuclphysbps.2015.09.144.pdf (Dataset definition for CMS operations and physics analyses)

In particular from RunB and RunC:

• HTMHTParked (Parked dataset motivated by Susy hadronic searches)
• HcalNZS (technical trigger with HLT_HcalNZS, HLT_HcalPhiSym, HLT_HcalUTCA)
• NoBPTX (technical trigger)
• VBF1Parked (motivated by Vector Boson Fusion)

In addition from RunC:

• LP_ZeroBias
• LP_ExclEGMU
• LP_Jets1
• LP_Jets2
• LP_MinBias1
• LP_MinBias2
• LP_MinBias3
• LP_RomanPots

-> all LP_ datasets have only 5 runs 198899-199903, https://twiki.cern.ch/twiki/bin/view/CMS/CertificationCollisions12 indicates that they are Totem runs: "LHC fill: 2836 Runs: 198899 198900 198901 198902 198903 Comment: Totem run - Special trigger menu - Run 198898 is flagged as cosmic but it is a collision run"

Update Nov 4 2016: The LP_ datasets are the CMS part of the common CMS-TOTEM runs. The data can be analysed only when combined with separate TOTEM data. Acoording to the FSQ PAG conveners, we can leave them out of the release.

[katilp]

The release date will be decided together with the CMS physics coordination making sure that all CMS 8 TeV key analyses will have been published before.

[katilp]

NB: no DoubleMu in 22Jan2013 reprocessing, checking this with Higgs POG, Muon PAG and PPD. Earlier reprocessing available: RunB: https://cmsweb.cern.ch/das/request?view=list&limit=50&instance=prod%2Fglobal&input=dataset%3D%2FDoubleMu%2F2012B%2FAOD RunC: https://cmsweb.cern.ch/das/request?view=list&limit=50&instance=prod%2Fglobal&input=dataset%3D%2FDoubleMu%2F2012C%2FAOD

Explanation: No need for DoubeleMu as DoubleMuParked contains DoubleMu