muid_eff modifications
- v05-02-00_prel_test : Sept. 27, 2010:
Added preliminary (for testing) efficiencies for Run2b2 MC.
- v05-02-00 : May 29, 2010:
Add new SPCs for Summer 2010 (Run IIa + Run IIb)
- For Run IIa:
o recertified efficiencies for data and MC
o all this for the updated track ID working points
o same new isolation working points implemented (as for Run IIb in December)
- For Run IIb:
o all the data until Summer 2010 is included in the certification
o the data efficiencies are split in the three epochs:
* Run IIb1, Run IIb2, Run IIb3
o the MC efficiencies are provided for old MC (p20.09.xx) ONLY for the time being
o the MC effiencies are split into the same three data epochs as data, which is
necessary because of the different lumi profile of the data in those epochs.
The SPCs are fit for splitting into epochs and simultaneous processing of Run IIa+b.
The corresponding tags to use this set of SPCs are:
- caf_eff_util p21-br-20
- eff_utils p21-br-29
- caf_utils p21-br-130
- vjets_cafe v05-04-00
(the ones specified or later)
N.B.
1) We believe that for most of the analyses on D0 facing notable background from
QCD multijet production the advised way to apply isolation criteria is in two
steps:
1. deltaR(mu,jet) > 0.5
2. posterior isolation (e.g. TopScaledMedium)
This creates a clear and well-defined topology. The sample passing 1. only but
not 2. should be used as a model for QCD multijet background (and is thus
kinematically and topologically similar to those QCD events passing 1. and 2.,
simply inverting 2. could result in a QCD multijet sample which is not similar
to the one passing the full selection criteria.
Because the isolation is applied in 2 steps, we provide data/MC scale factors
for:
1. dR(mu,jet)
2. isolation | dR,
where "isolation|dR" is a so-called posterior isolation working point, e.g.
TopScaledTight|dR, which means that the efficiencies we provide assume that
cut 1., dR > 0.5 was passed. This is reflected in the standard setup in
vjets_cafe/config/VJetsMuonStandard.config
If your analysis has a strong reason to deviate from this scheme, we are ready
to provide SPC for the plain isolation (i.e. w/o the "|dR") on a private basis.
2) The Single Muon OR trigger efficiencies are now derived also w/r/t the updated
track ID and new isolation working points. Please use the most recent spc in
the caf_trigger package.
- v05-01-02 : Jan. 25, 2010:
Update SF-s for the Summer 2009 Extended dataset (Run IIb2):
- new luminosity read-back code, recovered o(1%) of total statistics, so only "cosmetic" changes due to this;
- produced with p21-br-13 branch of the muonid package, and thus changes of o(1%) result for the muon-ID SF-s only.
- new EfficiencyVersion: 14
- updated run range
- removed tight quality for muon-ID
- v05-01-01 : December 16, 2009:
Updated l1, l2, and l3 trigger terms. This is for new track quality def's only. Revert to v05-01-00 or earlier for old track quality def's.
- v05-01-00 : December 14, 2009:
New track definitions available, new isolation available.
Update of SFs for both old and new tracks definitions (on the summer09 extended dataset).
The majority of these changes are for the re-certification of the Summer 2009 extended dataset:
- one additional track quality: 'tracknewmedium'
- track ID efficiency measurement further unbiased by calorimeter isolation requirement on the probe - 4 additional isolation requirements with much less lumi dependence and less relying on calo isolation:
* TrkTightScaled
* TrkLooseScaledL
* TrkTight
* TrkLoose
- events with lumi=0 are excluded from the procedure
- luminosity reweighting implemented:
- requires the weight variable on each of the tree (see README in lumiProfile folder how to get it)
- can be switched on by passing the --lumi flag to Certif_x
- luminosity parametrisation implemented
- factorisable efficiency measurements, i.e.
eff_ijk = eff_ij x eff_k,
where eff_ij and eff_k are derived on the full sample (and are uncorrelated).
To ensure proper normalisation, introduce:
* absolute
* normalised
efficiency measurements.
- improved code legibility
- fixed overall efficiency bug (histo->Integral() instead of ->GetEntries() )
- added possibility to force splitting of *offline* SFs into v15 and v16 by passing the '--split_v15v16' flag.
- added field ObjVersion for each of the spc-s to keep apart new and old track quality definition
The actual changes to the SF-s and their parametrisation are:
- muon ID: no changes other than binning
- track ID: SF-s for two versions of ObjVersion: 2 (1) for new (old) definition. Parametrisation in:
eta x z0 (as before, but changed binning)
eta x lumi (normalised to 1 to preserve overall normalisation)
- dR isolation: parametrisation in:
|eta_CFT| x lumi (old parametrisation was in dR and thus all SF=1 by definition)
- "posterior" isolation (i.e. NPTight|dR): parametrisation in:
|eta_CFT| x dR x pT
- v04-03-00 : November 7, 2008:
New spc files for the winter 2009 dataset are now available.
There is a slight change to the medium muon definition that should result in a 1-2% increase in the efficiency for medium (and mediumnseg3) muons in the central region. This medium definition change is the reason why the updated MuonSelector in caf_util is needed. (Note that no change has been made - yet - to the medium muon definition for run2a.)
As a reminder, two parameterizations are also now available for the isolation spcs. The "old" parameterization, in pT and the number of jets, is still available. There is also a parameterization in pT and dR to the nearest jet available. It was shown a while ago that this new parameterization works better for final states with a lot of jets. (Note that this new isolation spc parameterization is not yet available for run2a. We hope to provide this soon.)
- v04-02-01 : June 6, 2008:
Updated muon spc files have been released in muid_eff v04-02-01. These spc files cover the full summer 2008 data set. Changes since the last release include:
- Updating to the data quality in dq_defs v2008-05-01a.
- Updating to muo_cert/wzreco v07-00-01, which includes a more complete implementation of the v16 trigger.
- muid, tracking, and isolation efficiencies are no longer divided by trigger epoch, they are averaged over the full run2b data set. (Trigger efficiencies are still divided by trigger epoch.)
The previous release (v04-02-00) broke up muid, track, and isolation efficiencies by trigger epoch. However, this strategy does not work with the current caf_eff_utils::MuonCorr code (as several users have already found). So, anyone using muid_eff v04-02-00 should upgrade to v04-02-01. Any problems should be reported to the muon ID group.
There are various technical reasons that make dividing muid, tracking, and isolation efficiencies by run number (such as into preshutdown 2007 and postshutdown 2007 epochs) difficult. If there is a strong demand for such a split, the muon ID group will try to pursue it further.
- v04-02-00 : May 27, 2008:
The first iteration of the certification for the summer 2008 run2b data sample has been released as muid_eff v04-02-00.
This version includes the data quality from dq_defs v2008-05-01. All efficiencies are separated into v15 and v16 trigger lists ranges. The v15 range includes both pre-shutdown 2007 and post-shutdown 2007 data.
The implementation of the v16 trigger is not yet complete. Currently, only those trigger terms that were already present in v15 are implemented. The implementation of the v16 trigger is expected to take another week, at which point updated spc files will be released.
Any questions or issues should be directed to the muon ID conveners.
Mike and Boris
- v04-01-01 : Jan 18, 2007:
A new version of the spc files for muons (for run2b preshutdown 2007 data) has been released in muid_eff v04-01-01. This version corrects a bug in the L2 STT efficiency calculation. The L2 STT efficiency present in previous versions of muid_eff is overestimated. (Note that this is an issue only for run2b data - run2a is unaffected.)
Ken Herner has taken a look at these new spc files for one specific trigger and found roughly a 5% effect. The effect will probably be different for different triggers, but you can expect that there will be a noticeable effect (unless your trigger doesn't have an STT requirement at L2). Work is ongoing (by Jorge Benitez) to redo the spc files for the muon trigger OR.
This version of muid_eff also updates the data quality used to that for the fixed run2b sample (dq_defs v2007-10-01-fixed). The previous release of muid_eff used the unfixed data quality. This change is not expected to have any noticeable effects on analysis.
- v04-00-07 : Nov 6, 2007:
A new version of the muid_eff package is available that contains spc files for the run2b single muon trigger or (thanks to Liang Li).
Here is the files that have changed in this new version:
- results_muor/eff_SingleMuonOR_SingleMuonTriggersOR_2D_phi_deteta_Binned.spc - this is the 2D spc that has both run 2a and run 2b results
- results_muor/eff_SingleMuonOR_SingleMuonTriggersOR_3D_phi_deteta_z_Binned.spc - this is the 3D spc that has only run 2b results
- results_muor/triggers.list.run2a - contains list of triggers in the run2a or, renamed from triggers.list
- results_muor/triggers.list.run2b - contains a list of triggers in the run2b or
- results_muor/goodRunsList.run2a - list of runs used in the run2a or, renamed from goodRunsList
- results_muor/goodRunsList.run2b - list of runs used in the run2b or
- v04-00-02 : June 16, 2007:
The package muid_eff version v04-00-02 has been updated with (preliminary) spc files for run2b.
Change relative to previous version:
- up to date data quality => expect efficiency change at the level of 0.2%
- bugfix: deltaR for isolation computed with jet with pt>15 GeV (bug was consistent between data and MC)
- change binning for isolation
- v04-00-01 : May 16, 2007:
The package muid_eff version v04-00-01 has been updated with (preliminary) spc files for run2b.
Change relative to previous version:
- fix metada bug for run2b efficiencies
- nEfficency computation for the entire summer 2007 p20 data sample.
Expected changes in the near future:
-up to date run2b data quality.
- v03-00-00 :
A new version of files for muon related efficiencies is available: muid_eff v03-00-00.
This should be the final "certified" muid efficiencies for p17.
For preliminary results, the old (v02-00-00) pre-certified efficiencies are still valid. The difference between old and new are smaller than the systematic uncertainties.
Summary of the changes:
- updated sample and selection do derive efficiencies
- trigger efficiencies are broken down into more trigger lists
- Id efficiencies are stable at the level of <0.3% with respect to v02-00-00
- (average) trigger efficiencies are stable at the level of ~0.5% with respect to v02-00-00
Going into Details:
- Reskimmed the Zmumu sample to derive efficiencies and used cable swap fixed data.
- a few (small) bias removed thanks to this new selection.
- update data quality dqdefs 2006/10/10
- efficiencies are stable at the level of <0.3% for Id
- the trigger efficiencies have been splitted into more trigger list ranges:
- [8, 12.99] [13,14.99] for L1 L2 L3 muon
- [8, 12.99], [13, 13.99], [14,14.89], [14.90,14.99] for L1, L3 track
- average efficiencies are stable at the level of ~0.2-05% for triggers. Larger differences (~2%) can be expected when analyzing a very specific time period with single muon trigger.
For more information, plots, see Wiki page
shary,cooljoe