Common Samples Group

1
Common Samples Group

• Y. Gershtein

2
Goals

• The idea for this group comes from a grass-root
  movement for common data skims - we will try to
  maintain the spirit
• The groups charge is to find ways to reduce
  duplication of effort in physics groups
• We hope that this group will become a place
  where people from different physics groups can
  work together and share work on similar tasks

3
Tasks

• selective p13 DST reprocessing with p14
• fixing calorimeter and vertex in TMBs
• skimming p14 TMB (and moving from skimming to
  off-line inclusive streaming)
• common object certification and correction
• documenting and coordination production of a
  common use data sets both MC and data

4
(No Transcript)
5
Selective DST reprocessing
O. Atramentov, F. Deliot, YG, H. Greenlee, G.
Landsberg, M. Verzocchi

• Use remote farms to re-reconstruct p13 data from
  DST but if we want to finish some time next
  winter we can only process 1/3 of the data
• Define the selection based on
• benefit from reprocessing
• physics priority
• fraction of data selected
• New code - p14.05.00
• Events are selected based on p13 objects
• calorimeter problems and old vertex
• Complementary to p13 thumbnail fixing

6
Selection for Reprocessing
7
Fixing Thumbnails
S. Protopopesku, M. Verzocchi

• Fix calorimeter problems, redo vertexing with
  latest algorithms and re-reconstruct calorimeter
  objects
• p13
• Energy sharing tower 2 problems
• correct (p14) ICD weights
• cable swap
• Two pass vertexing and min-bias PV selection
• p14
• Tower 2 problems
• cable swap
• Two pass vertexing and min-bias PV selection

8
Fixing Thumbnails

• Fixer is ready to go
• lost a week chasing a nasty non-reproducible
  crash (which was causing p14.04 d0reco crashes,
  too)
• Will start fixing p14 TMBs first
• status will be posted on the CS web page
• expect to finish p14 by the end of October
• You can run the fixer by yourself
• instructions can be found on the CS web page

9
Skimming and Streaming
O. Atramentov, F. Deliot, YG, H. Greenlee, M.
Verzocchi

• Thumbnails will be skimmed
• inclusive skims
• (practically) no limit on number or size of
  skims
• Off-line exclusive streaming
• anticipate the need for future data reprocessing
• does not affect thumbnail skims
• will separate DSTs into four exclusive streams
• Stream A high pT electrons/muons 30
• Stream B high pT jets, MET, taus 20
• Stream C low pT muons 20
• Stream D the rest
  30
• separation is based on skim definitions (being
  finalized)
• considering writing out RAW data for Stream A

10
Stream A

• 1EM ID10,11 pT18, h8
• 2EM ID10,11 pT7 2
• 1MU loose, max(local, global)10 8
• 2MU loose
  5
• MUEM loose, ID10,11, pT5 GeV
• one track pT5 GeV
  6
• JES (tight version)
  1
• Minbais
  1
• Zero bias
  1
• J/psi electrons
  2

11
Stream B

• JT_65TT, JT_95TT, JT_125TT
  2
• MHT30_CJT5 3CJT5 monitor
  3
• 3jets pT20,15,15, eta_d12
• METtrk MET20, pT5
  8
• Tau2trk 10 GeV, 1-prong, NN0.3
• plus two tracks above 5 GeV
  11
• EM trk ID10,11 Et12,
• track pT 7 GeV matched to EM
• EM 2trk ID10,11 Et8, 2 trk pT5
• one matched to the EM
  6
• MU 2trk loose, central pT5,
• plus additional track above 5 GeV
  9
• JES (all)
  9

12
Stream C

• b-MU medium, central pT2 33

13
Common Corrections
A. Zatserklyaniy, F. Deliot, YG

• The idea is to move certified object corrections
  to a common package
• Make sure that different root-tuple makers
  (top_analyze, TMB_analyze, wz_analyze, etc) use
  the same corrections can have a well-defined
  versioning system
• Decouple the correction code from tuple-making
  code easier path from correction code into reco
  and better code development practices
• Gain ability to write out corrected objects in a
  common format (thumbnail for now)
• Object corrections on a root-tuple level
• Important for development but the code should
  be made available in the framework on a regular
  basis

14
Common Corrections

• For each object, a new chunk with corrected
  objects is added to the event
• Physics objects are modified so that the new
  object can be represented by it
• for example electron likelihood error on jet
  scale correction
• The root-tuple makers become much simpler they
  just need to take a chunk and put it into a root
  branch
• Thumbnails for the physics objects are modified
  so that the new fields get packed and the event
  can be written out this can become a platform
  for common sample definitions

15
Plan for Common Corrections

• Ideally
• d0correct contains only RCPs
• corrections are done by dedicated packages (like
  em and muons now)
• the versioning is ensured by a high level
  releases
• affect only high level packages like
  emreco/em_evt
• can be done quite frequently
• Near Future (next week)
• some of corrections are done by dedicated
  packages
• some will have to be applied in d0corr code
• versioning by tar-balls and clear CVS tags
  (i.e. d0cor-01-02 for all relevant packages)

16
Common MC Samples

• Need to have an inventory of existing MC and the
  requests from all physics groups
• Some of the samples can be shared (with proper
  parameter coordination)
• Drell-Yan
• QCD
• Wjets
• Looking for a person to organize this effort

17
Common Data Samples

• Work on data sample is an integral step of every
  analysis. With very little effort your sample and
  analysis code can become useful for other people
• good example is Marians di-muon sample for LP
• enjoy increased name recognition ?
• CS group can become a repository of knowledge
  about useful data samples and code
• easier start for newcomers
• environment which encourages people working on
  similar analyses in different physics groups to
  share the workload
• Leslie Groer and Ryan Hooper volunteered to
  coordinate creation of high pT di-EM and di-MU
  samples what about other final states?

18
Summary/Plans

• A lot of projects have started
• So far concentrated on getting fixed and skimmed
  data to collaboration
• The activity on common MC and data samples is
  ramping up
• Stay tuned!