D0 global tracking: Previously Active Projects

(Some are still relevant!)

H. Greenlee
20jul01


These may or may not be undergoing active development. If no one is listed under personnel, then that project is open. Some projects need help even though some people are signed up. Click on the individual project to find out additional details about that project. Many of these refer to the GTR tracking package which is no longer used in track finding but is used in track fitting and refitting.


Other Older Projects:



Integration

Personnel: D. Adams, H. Greenlee, V. Kuznetsov.

Test changes to tracking code. Test & measure performance for each weekly release.


Analysis tools.

Personnel: D. Adams

Develop a comprehensive suite of ROOT macros to replace the paw macros in gtr_lyze. This has been started by Dave Adams, but more could be done. (see gtr_root).


Unpacking & clustering.

Personnel: M. Hildreth (CFT), Y. Kulik (SMT).

Develop/debug unpacking, clustering, calibration etc. for real data. This is normally the responsibility of the detector subgroups.


Paths for real data.

Personnel: M. Hildreth (CFT), Y. Kulik (SMT), D. Adams.

Develop paths that can work for partial detector commissioning data and later full detector. We currently have the following paths.

As more of the detector is instrumented, additional paths will be required.


Measure detector performance.

Personnel: M. Hildreth (CFT), Y. Kulik (SMT).

Using commissioning, cosmic, or physics data to measure the performance of individual detectors (light yield, pulse heights, noise, efficiency, etc.).


Performance analysis with real data.

Personnel: M. Hildreth, A. Kharchilava (CFT), Y. Kulik (SMT), D. Adams.

Measure the performance of the tracking software for real data.


Monte Carlo performance analysis.

Personnel: Various

Analyze Monte Carlo data sample of various types with a view to identifying specific problems and improving tracking software performance. We measure the following aspects of tracking performance.

Historically, the tracking group has concentrated on analyzing MC Z->mu mu with overlaid minimum bias background events with emphasis on the muons. There is a need to extend these analyses to other kinds of data, including "jetty" data and all particle types.


Debugging tools

Personnel: D. Adams.

As part of the attack on the problem of understanding and improving tracking (in)efficiency, there is a need to develop additional debugging tools, including printouts and event displays. One goal is to be able to quickly identify why any given Monte Carlo track failed to be reconstructed.


Improved CFT clusters

Personnel: J. Krane

Develop new cft trf clusters that correctly model odd/even effects within fiber doublets, and model the degradation of CFT cluster resolution (spoiled clusters) with increasing fiber occupancy.


Central CFT performace for high track densities

Personnel: M. Hildreth, J. Krane, D. Adams

Why does the high-pT track efficiency drop and the misrecos and fakes rise as background events are added?


low pT inefficiency

Personnel: M. Hildreth, J. Krane, D. Adams, R. Gelhaus.

Understand why chisquares increase and tracking efficiency drops at low pT, and for particle types other than muons.


Alignment.

Personnel: A. Kharachilava (CFT), Y. Kulik (SMT)

Deriving detailed alignment constants is the responsibility of the alignment group. However, the tracking group has worked on some aspects of the alignment problem, including the following. These tasks are now mostly done.


H-disk extension.

Personnel: V. Kuznetsov, H. Greenlee

Extend forward and gap tracks to H disks. Infrastructure and paths for this exists, except for non-uniform magnetic field propagators.


H-disk extension.

Personnel: V. Kuznetsov

Develop tracking algorithm starting with the smt in the overlap region.


Extend SMT tracks into CFT.

Personnel: V. Kuznetsov

Extend SMT and gap tracks that cross CFT layers.


Propagation outside of tracking volumne.

Personnel: V. Kuznetsov, G. Hasketh.

Propagate outside the tracking volume using interacting propagator. Need to correctly describe geometry and magnetic field. This is done for the central region. This still needs to be done for the forward region.


Fast propagators for non-uniform magnetic field.

Personnel: H. Greenlee

Need fast propagation in non-uniform magnetic field for track-finding with H-disks. Would like better (more accurate) non-uniform magnetic field propagator for overlap tracking. This is now done except for DCA propagators. There is also a need to optimize the magnetic field interface. A plan exists, but no one has identified to do it.


Propagator verification.

Personnel: R. Gelhaus, G. Hasketh

Verify and compare trf propagators (especially interacting propagator) against d0gstar.


Speed up overlap tracking.

Personnel: H. Greenlee, Y. Kulik, D. Adams, V. Kuznetsov

Total global tracking time is currently dominated by the overlap tracking step. Despite the number of people listed, we could use new ideas here. Some ideas which have been thought about, but not yet tried are as follows:


CFT paths with missing hits.

Personnel:

Add proper misses to CFT layers. Develop paths for CFT which allow for misses or CFT cluster inefficiency. SMT paths already allow inefficiency.


Reco_analyze

Personnel: V. Kuznetsov

Add global tracking information to reco_analyze ntuple (done).


Global cuts.

Personnel: H. Greenlee

Develop a mechanism for specifying certain cuts, such as minimum track pT globally (via rcp or obs), instead of in many individual .obs files. This is now partially done, global cuts values are now stored in ObjTable, but global cuts need to be implemented in many trf components (checkers, cluster filters, etc.)


Global objects.

Personnel: S. Kulik, H. Greenlee

Globally define magnetic field, propagator, beam position, and other objects usefully withing and outside of tracking (mostly done).


Optimal fits.

Personnel: Y. Kulik, D. Adams.

Currently, GTrack's only have optimal fits at ends of tracks. Develop methods for getting optimal fits at all layers (desirable for alignment purposes). (Is this the same as Kalman Smoothing?)

Lately, Dave Adams has developed a class to refit tracks and calculate residuals at each layer. This mostly meets this need.


Save misses on GTracks.

Personnel:

Modify GTrackState to store misses, and make them persistent. Put misses on GTracks when converting PTracks. This is partially done, but misses on GTracks are not yet persistent.


Final refit chisquare

Personnel:

Replace hard-coded final refit chisquare cut with one that is specified by .obs or .rcp. Make value of cut available to cut recorder.


mikeh@fnal.gov