Run II Reconstruction Program Status

Page updated: September 20, 2002

Run II Reconstruction Program Status - Version p11

Latest certified version: p11.12.01	Next pass release: p11.13.00
Status: Certified	Not yet scheduled

Performance • Upcoming Features • Known Problems • History
How to run RECO • Report RECO problems

Overview

Significant improvements were made in p11 concerning detector alignment and calibration, tracking and muon identification. The first certified version was p11.08.00. The current version is p11.11.00.

A short list of improvements is included below:

Detectors
- Global and internal alignment of SMT, CFT, MUON
- Improve SMT Lorentz corrections (from alignment studies)
- Tuned SMT / CFT cluster errors
- Calorimeter non-linear corrections
Tracking
- Provide “best” tracks in “401 chunk”
- Turn on F-disks
Vertexing
- Start using global tracks for data (“401”) for default vertexing
- Ability to find vertices with all types of tracks
Electrons - no major changes
Muons
- Use survey-based muon geometry
- Allow "scintillator-only segments“, filling gaps in the wire chamber coverage
- Allowing BC-only tracks, increasing coverage by not requiring A layer
- Use global matching with the full GTrack and Muon::Track error
- Use calibrated PDT t0
- Fix charge (use correct sense of field based on db info)
Jets / MET - Use calorimeter non-linearity corrections
Taus - no major changes
BC jets - no major changes

Performance

The Algorithm and Object ID groups have studied the performance of p11 reconstruction (tracking (data, MC), vertexing, electrons, muons, jets, missing Et, jet energy scale).

Some overall statistics (speed, memory usage, etc.) (p11.09.00, p11.11.00) and standard plots (p11.08.00, p11.09.00, p11.10.00, p11.11.00) are available.

On September 18, 2002, a MiniWorkshop on Tracking Algorithm Studies was held. Results presented there are based on p11.11.00.

Upcoming patches, etc.

Modify semreco and caltrack_extra to handle the fact that the offline calorimeter geometry is not aligned with the tracking detectors. An ad hoc correction will be made in these packages to take care of the global offset.

Known Problems

Major error, Annoying feature, Functionality to be added
Scheduled for next release, Problem fixed.
(common = occurs within 100 events, rare = occurs within 1000 events)

Status	Description	Fixed in version
	L3PhysicsResults missing in the L3Chunk
	Calorimeter not aligned with trackers (3cm shift).	Partial patch for semreco / caltrack_extra scheduled for p11.13.00
	reco_analyze crashes on MC farms due to bug requiring network connection (common)	reco_analyze being phased out
	p11 reco_analyze crashes when processing p10 reco files in bcjet_analyze (common)	p12
	reco_analyze crashes with a floating-point exception on d0mino, using the non-optimized build. Temporary solution: use maxopt build; i.e. runrecoanalyze -maxopt ...	reco_analyze being phased out
	There is no rcp to allow users to reprocess Monte Carlo events in SAM. Users can find a working rcp on d0mino, in ~melanson/projects/reco/p11.11.00/d0reco/rcp/runD0recoSAM_mc_reprocess.rcp. Copy this file to your local d0reco/rcp area and use the runreco options "-localrcp -localbuild -format=mc_reprocess"	p13
	Analyzing thumbnail (vertex) requires access to run config db	p13
	p11.11.00 maxopt build failed on d0mino. Users can check out the latest version of AATrack for p11 (addpkg AATrack p11-br) and build a local version of RECO.	p11.12.01
	EM HMatrices not tuned for p11. When available, will improve electron ID efficiency.	p11.12.00
	Calorimeter non-linearity correction wrong for data (introduced in p11.10.00 when attempting to fix for Monte Carlo). (common)	p11.11.00
	Data taken with lower calorimeter thresholds ("1.5 sigma") increase CPU time per event and reco_analyze root tuple size	p11.11.00
	RECO crashes in bc_muTagreco when processing some Monte Carlo samples, due to using NULL pointer. (rare)	p11.11.00
	Floating point exception in hmreco::Helqmaker (rare)	p11.11.00
	Bug in emreco causing wrong Hmatrix chisq (rare)	p11.11.00
	d0_analyze fails to run in p11.10.00	p11.10.01
	Incorrect SMT Lorentz corrections when processing Monte Carlo generated with p10 (common)	p11.11.00
	RECO crashes in muonid with a divide by zero. (p11.10.00 only on d0mino)	p11.10.00
	Incorrect MET blocks in reco_analyze	p11.10.00
	cps_anal crashes (disabled in reco_analyze)	p11.10.00
	Calorimeter non-linearity corrections not handled properly for MC	p11.10.00
	Muon index problems resulting in no MSC, MDT hits	p11.09.00
	No CPS clusters	p11.09.00
	Incorrect ADC to GEV constants for ICD	p11.09.00
	Incorrect readout map for ICD	p11.09.00
	Bug in SMT hit error matrix	p11.09.00

History of Major New Features

p11.12.01
- Reinstate CFT axial track block in reco_analyze root tuple
p11.12.00
- New HMatrix training - will improve electron id efficiency
- vertexing - change PV cuts (tighten)
- caladdress - Fix obscure problem with Run I channels
- muonid - Fix thumbnail index packing
- Fix problems with tracking algorithm studies (rcp, AATrack)
- muon - include patches to help with timing studies
- CFT axial-only tracks restored to reco_analyze root tuple
p11.11.00
- Enable offline threshold cuts for calorimeter and set to 2.5 sigma
- Fix calorimeter non-linear corrections (problem introduced in p11.10.00)
- Implement TMB+ (TMB + CalDataChunk + MuoTrackChunk)
- Fix bug in emreco resulting in incorrect HMatrix chisq
- Fix bug in SMT Lorentz corrections for Monte Carlo (long standing problem)
- Add support for all existing tracking algorithms and provide tools for TRAC studies for p13
- Add support for SMT / CFT calibration db access (use format = "data_db")
- Bug fixes to deal with reported floating point exceptions
p11.10.01
- d0_analyze - fix OBJECTS file.
p11.10.00
- Fix ICD mapping.
- Fix calorimeter non-linear corrections for Monte Carlo.
- Fix CPS reco_analyze package and re-enable.
- Fix bug causing connections to db to crash.
- Fix thumbnail bugs for EM reco.
- Fix MET block in reco_analyze.
  - NB - the reco_analyze root tuple block called OLDMET is now called METMET.
- Remove some noisy printout from CFT.
p11.09.00
- Fix muon index bugs. Solves problem with missing MSC, MDT hits.
- Fix CPS clustering bug. Solves problem of no CPS clusters.
- The following changes were made to speed up processing
  - Eliminate the smt central track finding and corresponding cft extensions (chunks 230, 211, and 333) in reco.
  - Raise threshold on extrapolating tracks to muon system ("muo_analyze") to PT > 10 GeV.
  - Disable V0 searching in reco_analyze.
  - Drop several GTR blocks in reco_analyze
- Disable CPS in reco_analyze. cps_anal is crashing.
- Drop SFTDigiChunk, in preparation for DST in p12.
p11.08.02
- Use new post-June 2002 SMT thresholds
p11.08.01
- No changes to RECO
p11.08.00
- First certified version of p11

Detailed history of changes in p11

p11.12.01
- gtr_analyze - reinstate CFT axial track block in reco_analyze root tuple
p11.12.00
- AATrack - Fix bugs uncovered by TRAC studies
- caladdress - Fix obscure problem with Run I channels
- elastic_reco - Fix rcp
- event_info - Fix L2 bit mask bug; fix L1 bit info initialization.
- hmreco - Tuned H-Matrices
- muonid - Fix TMB bug
- muon_raw - Contain development essential for timing studies
- muon_hit - Contain development essential for timing studies
- d0reco - Fix rcp's for TRAC studies
- vtrkimp - Select tracks with impact parameter significance less than 3 for PV.
- valgimp - Max. chi-squared contribution of a track set to 10 down from 100 for PV
- gtr_analyze - CFT axial-only tracks restored to reco_analyze root tuple
p11.11.00
- bc_muTagreco - protect against using NULL pointer
- calunpdata
  - Fix non-linear corrections for data
  - Introduce offline threshold cut (set to 2.5 sigma)
  - Make nlc corrections use "double" instead of "int"
- caltables - Make nlc corrections use "double" instead of "int"
- emreco - Fix bug returning incorrect HMatrix values
- hmreco - Protect against negative energies
- gtrfit - updated fitters (support all tracking algorithms)
- smt_hitalgs
  - Support 1D clusters
  - Fix Lorentz corrections for Monte Carlo
- silicon_geometry
  - Support 1D clusters
  - Fix Lorentz corrections for Monte Carlo
- smt_lpos - Support 1D clusters
- smt_hitconverter - Support 1D clusters
- smttrack_reco - Support 1D clusters
- smt_path - Add OBSNAME
- cft_obs - Add OBSNAME
- gtr_find - Support for all tracking algorithms
- elastic_reco - Update
- gtr_htf - Update
- AATrack - First version
- gtr_analyze - Support for all tracking algorithms
- calibration_management - Enable CFT and SMT prd servers
- cft_unpdata - Provide rcp to connect to calibration db
- d0reco
  - Support tracking combinations for TRAC
  - Support SMT / CFT calibration db connection
- reco_analyze
  - Support tracking combinations for TRAC
- d0_analyze
  - Support tracking combinations for TRAC
p11.10.00
- caladdress
  - Fix phi swap for ICD
  - Return invalid for Run I physics addressed are input
- caltables - fix non-linear corrections for Monte Carlo processing
- cps_anal - fix crashes
- d0omCORBA - fix bug resulting in crashes if db connection opened/closed often
- d0stream - associated with d0omCORBA patch
- em_analyze
  - No pointer to CaldataChunk is removed
  - Road info is also available without the SEMparticleChunk
- em_evt - EM id is saved in thumbnail chunk (EM part)
- muon_raw - fix usage of useDB variable
- reco_analyze
  - Re-enable CPS
  - Start using met_analyze
- sftdigi - remove annoying print statements
- muonid - fix bugs in thumbnail
- chpart_reco - fill chi2/ndof
p11.09.00
- cal_nlc - Fix bug in component test (p11-br-04).
- caltrack_extra - Fix LinkIndex from TrackCalExtraChunk (p11br-01)
- cpsdigi - Fix duplicate rcp problem in itest.
- muo_analyze
  - Adjust PT cut to 10 and extrapolate only to C layer.
  - Fix bug causing MakeClass to fail (p11-br-11).
- smt_hit - Fix bug related to error matrix (p11-08-00-g-br).
- thumbnail
  - Fix packing of Global quantities (p11-br-10).
  - Fix ThumbNailBcJetUtil component test (p11-br-11).
- vertex_analyze - Turn of V0 finding, to speed up processing.
- caladdress
- caltables
  - Fix ICD ADC to GEV constants.
  - Changes needed for non-linearity corrections in pileup (p11-br-02).
- calunpdata - Fix ICD electronics map.
- bcjet_evt - Fix range problem in bcTmbObj (p11-br-08).
- cps_reco - Fix thresholds, allowing the finding of clusters.
- d0reco
  - Drop SFTDigiChunk
  - Change default processing of data to use faster GTR paths.
  - Include new prejun02 format (old, slow paths).
- muon_index - Fix bugs causing no MSC, MDT hits.
- smtraw - Include new post-June02 shutdown pedestals.
- Speed up GTR and GTR_ANALYZE
  - cft_obs - Modify cft stereo paths to extend tracks with fewer than eight axial hits.
  - cft_trf -
    - Modify detectors to add doublet information to cft misses.
    - Modify checker CftCheckMissLhood to have ability to separately check axial and stereo miss probability.
    - Add checker CftCheckHitCount (can count axial & stereo hits separately).
  - cft_trfcl_filter - Fix component test.
  - cft_trfclus - Add (optional) doublet information to CftMiss.
  - gtr_analyze - Modify package rcp to record fewer ntuple blocks.
  - gtr_evt - Modify MissRecorder to store updated CftMiss parameters.
  - gtr_find - Add new rcps GtrFindPkg_rdata_fast*.rcp to run faster by omitting smt central steps.
  - smt_path - Add smt_to_cft_overlap.obs.

This page maintained by Harry Melanson