Valid8TeVVertex
Intro
Planning of 8 TeV data start-up.
Main topics:
- MinBias measurement and data/MC comparison
- Extrapolation and study at higher pile-up
Focusing on three main domains:
- Vertex efficiency and related (split, fakes, etc..)
- Vertex position resolution and error estimation
- Identification efficiency for Hard-scattering vertex
Data and MC samples
Summary table of datasets (description below)
Type |
D3PD dataset |
Comment |
AOD dataset name |
Notes |
DATA |
user.spagan.data11_7TeV.00188949.calibration_VdM.merge.VTX_MON.x152_m849.v2.0/ |
7 TeV mu scan for low mu data |
data11_7TeV.00188951.physics_MinBias.merge.AOD.f403_m975 |
17.0.2.10 D3PD |
DATA |
user.spagan.data11_7TeV.00188951.calibration_VdM.merge.VTX_MON.x123_m849.v2.0/ |
7 TeV mu scan for low mu data |
data11_7TeV.00188949.physics_MinBias.merge.AOD.f415_m1025 |
17.0.2.10 D3PD |
MC |
user.spagan.mc10_7TeV.105000.pythia_minbias_inelastic.merge.VTXD3PD.e723_s932_s946_r2302_r2300.v4.0/ |
7 TeV mc10b sample with the D3PD format for validation of D3PD |
mc10_7TeV.105000.pythia_minbias_inelastic.merge.AOD.e723_s932_s946_r2302_r2300/ |
TrackD3PDMaker-01-01-12 |
Data
mu-scan data
Data from the mu Scan performed in Spetember 2011 can be used to get low mu data.
The LB range where the muScan was performed are in the
elog entry. Here for
completeness:
mu scan #1: Fill 2086, run 188949, ATLAS LB 91-126,9:45 < t < 10:03 h.
Toroid off. Single-sided test scan to exercise procedure.
Three problems found (scan protocol, BSSP, transition time); the first
2 were then
fixed, the third one we need to live with.
mu scan #2: Fill 2086, run 188949, ATLAS LB 138-160,10:23 < t < 10:41 h.
Toroid ramping. Single-sided test scan to exercise procedure.
All lumi data should be valid except forLUCID (and perhaps TILE) because of the
varying toroid fill.
mu scan #3: Fill 2086, run 188951, ATLAS LB 129-182, 13:17-14:06 h.
Toroid at nominal field. Double-sided scan, nominal conditions.
All lumi data should be valid.
The corresponding dataset should be:
data11_7TeV.00188951.physics_MinBias.merge.AOD.f403_m975
data11_7TeV.00188949.physics_MinBias.merge.AOD.f415_m1025
In both cases you should notice that at some point the LB numbering
change. These "pseudo-LB" number the phases of the scan at fix mu. If
you dump info on a
D3PD, have a look at ei_actualIntPerXing to
understand the corresponding mu value.
See also
ELOG entry (Internal LBL)
List of Performance tests
Vertex efficiency and related (split, fakes, etc..)
Main focus on determining key vertex reconstruction performance in a very low pile-up regime.
MinBias measurement and data/MC comparison
* Data
- Measurement of vertex reconstruction efficiency as function of good tracks. Uses very-low bias trigger. Assumes collision happened.
Denominator: good tracks with requirements w.r.t. Beamspot; numerator: events with >=1 reconstructed vertex.
See MinBias note:
http://arxiv.org/abs/arXiv:1012.5104
https://cdsweb.cern.ch/record/1271677?ln=en (section 6)
* MC
- Truth-based tool on MC samples. Comparison with 7 TeV results with homogeneous triggers (sample request in progress).
* Data/MC
- Basic plots: NTracks/vertex; Sumpt2; NVertices (after mu-weighting for High-mu); ...
- Same technique on MC as in data for data/MC comparison.
- Vertex fitting efficiency from split technique
Vertex position resolution and error estimation
* Data, MC, Data/MC
- Split-vertex method. Compare non beam-constrained vertices in data,MC with beam-constrained vertex resolution from truth.
Identification efficiency for Hard-scattering vertex
Extrapolation and study at higher pile-up
MinBias measurement and data/MC comparison
* Data
- NVtx-mu fit with toy-derived function
* MC
- Truth-based tool on MC samples. Comparison with 7 TeV results with homogeneous triggers (sample request in progress).
* Data/MC
- NVtx-mu fit comparison.
Vertex position resolution and error estimation
* MC
- Resolution using truth-info; study degradation as function of pile-up
Identification efficiency for Hard-scattering vertex
- Mis-id probability extrapolation using measured SumPt2 of MinBias interaction and MC-based sumpt2 spectra of Hard-scattering interactions.
Analysis code
Collect utility scripts, documentation and specific analysis macros in the package:
svn+ssh://svn.cern.ch/reps/atlasusr/spagan/Vertex/Valid_8TeV/trunk
Use scripts/checkout.sh from your test area to check out all needed dependencies.
- D3PD Production:
Truth information,
naming convention compatibility with analysis macros
Using new D3PD framework. Currently
TrackD3PDMaker-01-01-13
(requires >=
InDetRecExample-01-23-59
and
D3PDMakerConfig-00-02-90
)
Submitting D3PD using PUStudies/run/script/submitPAthena (action: trfvtx). Source of transform command in
PUStudies/share/VtxGroupD3PD_trf_v1.txt
For running D3PD with athena standalone, have a look at
this twiki.
- MC Truth-based analysis of efficiency, fakes, splits
Using TrackingAna.exe from the PUStudies package:
svn+ssh://svn.cern.ch/reps/atlasusr/spagan/Vertex/PUStudies/trunk/
See for instructions:
https://twiki.cern.ch/twiki/bin/view/Main/SPaganGrisoVtxPileupStudies
- MC truth-based analysis of resolution, pulls (k-factors)
Same software as above (PUStudies package) produces desired histograms.
- Data-driven analysis of resolution + vertex fitting efficiency using split technique
Using the macro in the folder macros/EnhancedPrimaryVertexMonitoring of the package:
svn+ssh://svn.cern.ch/reps/atlasoff/InnerDetector/InDetMonitoring/InDetGlobalMonitoring/
(Use makefile to compile, then runs over D3PD. May need some adjustments for setting input files, etc..)
- Data/MC comparison and NVtx-mu fit
Write new code based on
macros/drawing/plot_NVtxMu_mc_data11_comparison.C
Select important distribution to compare.
Improve fit modeling of NVtx-mu
- Identification efficiency given the pT spectrum of Minimum-Bias as function of pile-up
Can use the script from PUStudies package
svn+ssh://svn.cern.ch/reps/atlasusr/spagan/Vertex/PUStudies/trunk/
in macros/toy/sumpt2.C
- Toy MC for shadowing effect as function of mu given "masking PDF".
Need to write it, will store it in PUSTudies package, directory
macros/toy/
MC Samples and Triggers
- Minimum Bias 8 TeV. ND, SD, DD separately. No pile-up. 100k events. Best tune available (Py8 A2M? Py6 AMBT2?).
- [optional] Minimum Bias 8 TeV. Pile-up up to mu~30 with ~flat mu distribution.
- Very low-bias trigger for vertex reco efficiency
- Random trigger
Comments/special considerations
- Triggers must be implemented in MC to check efficiency/fakes.
- No truth slimming for pile-up samples
--
SPaganGriso - 09-Mar-2012