CMON: Difference between revisions

From CLONWiki
Jump to navigation Jump to search
 
(18 intermediate revisions by 3 users not shown)
Line 3: Line 3:
== Algoritms and Structure ==
== Algoritms and Structure ==


Code location: ''$CLON/src/cmon''. Following describes core components.
Code location: ''$CLON/src/cmon''. Following describes core libraries. Usually library 'xx' has xxinit() and xxlib() calls as an interface.


'''Pattern recognition''' library, located in ''prlib.s'':
'''Pattern recognition''' library, located in ''prlib.s'':
Line 14: Line 14:


'''Track reconstruction''' library, located in ''trlib.s''.
'''Track reconstruction''' library, located in ''trlib.s''.
* alu.c
* cmlib.c - main calls
* dcbrun.c - called from cmlib in the beginning of every run
* dcdcam.c - derives "dcam" for each hit of the candidate track (calls: scmatch, scbeta, scdtime, dcdocam, dclramb)
* dcdocam.c -  calculates distance to the hit wire (calls: none)
* dcdsect.c - draw sector geometry, tracks and hits
* dcerun.c
* dcfield.c - Performs 3-dim 2-nd order interpolation for each component of the magnetic field using Taylor series
* dcfinde.c - Identify electron for electroproduction, or other particle for photoproduction
* dcfit.c - Level = 2: Performs track fitting using the stored wire positions and momentum from the pattern recognition templates; Level = 4: Final fitting with corrected drift times
* dcfitsl.c - fit one track segment in a SuperLayer
* dchist.c
* dcicyl.c - Intersection of a Track with a Cylinder. Calculates intersection of track  with cylindrical surface of radius R.  The track is approximated by a cubic in the track length.  To improve stability, the coordinate system is shifted. First the Newton method (by P.D.) is used to find the intersection. Derivatives are calculated analytically using the parameters of cubic approximation. For input parameters given correctly and for the track not too steep to the surface the Newton method converges in 1-2 iterations and gives ultimate accuracy. For the reasons of safety of the procedure the old Half-Interval Division method is used if the first method is failed.
* dcipln.c - Intersection of a Track with a Plane. Calculates intersection of track  with a plane situated at the distance 'd' from the center of coordinates. The vector 'ortvec' is the unit length vector directed from the center of coordinates to the direction of the plane and orthogonal to it.  The track is approximated by a cubic in the track length.    To improve stability, the coordinate system is shifted. Normally the Newton method is used to find the intersection. Derivatives are calculated analytically using the parameters of cubic approximation. For input parameters given correctly and for the track not too steep to the surface the Newton method converges in 1-2 iterations and gives ultimate accuracy. For the reasons of safety of the procedure the old Half-Interval Division method is used if the first method is failed.
* dclast.c - finishing the program
* dcpdst.c - gives access to a event bank 'EVTB' for Level = 2 or 4
* dcrkuta.c - one-step tracking a particle through a magnetic field using Runge-Kutta method
* dcrotds.c - vector transformation from Detector to Sector Coordinate System and reverse { DCS <--> SCS }
* dcstat.c -  keep tallies on different types of analysis cuts; for multithreaded applications: allocate TRstat structure locally for every thread and use 'dcstatadd(TRstat *stat)' to accumulate all statistic together
* dcswim.c - track swimming (traj. propagation) in magnetic field. Calculates track points on detector planes
* dcswimvt.c - back swimming in the PT magnetic field; calculates track points on the target plane#1
* dctodst.c - writes the reconstructed event into DST
* dctrfit.c - perform a fit of the track candidate
* dctrmat.c - calculates transport matrix for one step; records matrix at point of layer penetration
* sclib.c - find SC slab "idsc" in data the track points to
* stlib.c - start counter library
* tglib.c - tagger library (rewritten a la offline version)
* trbos.c
* trconfig.c
* trinit.c
* trlib.c


== Road Maps ==
== Road Maps ==
Line 65: Line 96:
Default runIndexTable is ''RunIndex''. For some run periods calibration constants are stored under different runIndexTable name, for example for g8b runIndexTable was ''calib_user.RunIndexg8b''. In that case use command
Default runIndexTable is ''RunIndex''. For some run periods calibration constants are stored under different runIndexTable name, for example for g8b runIndexTable was ''calib_user.RunIndexg8b''. In that case use command


  db2clonmap.pl system=TAG_CALIB runIndexTable=calib_user.RunIndexg8b unMin=1 runMax=1000000
  db2clonmap.pl system=TAG_CALIB runIndexTable=calib_user.RunIndexg8b runMin=1 runMax=1000000
  db2clonmap.pl system=SC_CALIBRATIONS_V2 runIndexTable=calib_user.RunIndexg9 runMin=1 runMax=1000000
  db2clonmap.pl system=SC_CALIBRATIONS_V2 runIndexTable=calib_user.RunIndexg9 runMin=1 runMax=1000000



Latest revision as of 19:16, 1 June 2012

CMON is CLAS Online Data Monitoring program. It does real time event reconstruction using old 'sda' program rewritten to C and modified for multithreaded environment.

Algoritms and Structure

Code location: $CLON/src/cmon. Following describes core libraries. Usually library 'xx' has xxinit() and xxlib() calls as an interface.

Pattern recognition library, located in prlib.s:

  • ec_export.c - fills ECPI bank
  • eclib.c - EC (Forward Calorimeter) calorimeter reconstruction library
  • iclib.c - IC (Inner Calorimeter) reconstruction library
  • pr_export.c - interface to offline packages, in particular makes DC1 bank
  • prlib.c - Drift Chamber pattern recognition, in particular Road Finder
  • sglib.c - Drift Chamber Segment Finder

Track reconstruction library, located in trlib.s.

  • alu.c
  • cmlib.c - main calls
  • dcbrun.c - called from cmlib in the beginning of every run
  • dcdcam.c - derives "dcam" for each hit of the candidate track (calls: scmatch, scbeta, scdtime, dcdocam, dclramb)
  • dcdocam.c - calculates distance to the hit wire (calls: none)
  • dcdsect.c - draw sector geometry, tracks and hits
  • dcerun.c
  • dcfield.c - Performs 3-dim 2-nd order interpolation for each component of the magnetic field using Taylor series
  • dcfinde.c - Identify electron for electroproduction, or other particle for photoproduction
  • dcfit.c - Level = 2: Performs track fitting using the stored wire positions and momentum from the pattern recognition templates; Level = 4: Final fitting with corrected drift times
  • dcfitsl.c - fit one track segment in a SuperLayer
  • dchist.c
  • dcicyl.c - Intersection of a Track with a Cylinder. Calculates intersection of track with cylindrical surface of radius R. The track is approximated by a cubic in the track length. To improve stability, the coordinate system is shifted. First the Newton method (by P.D.) is used to find the intersection. Derivatives are calculated analytically using the parameters of cubic approximation. For input parameters given correctly and for the track not too steep to the surface the Newton method converges in 1-2 iterations and gives ultimate accuracy. For the reasons of safety of the procedure the old Half-Interval Division method is used if the first method is failed.
  • dcipln.c - Intersection of a Track with a Plane. Calculates intersection of track with a plane situated at the distance 'd' from the center of coordinates. The vector 'ortvec' is the unit length vector directed from the center of coordinates to the direction of the plane and orthogonal to it. The track is approximated by a cubic in the track length. To improve stability, the coordinate system is shifted. Normally the Newton method is used to find the intersection. Derivatives are calculated analytically using the parameters of cubic approximation. For input parameters given correctly and for the track not too steep to the surface the Newton method converges in 1-2 iterations and gives ultimate accuracy. For the reasons of safety of the procedure the old Half-Interval Division method is used if the first method is failed.
  • dclast.c - finishing the program
  • dcpdst.c - gives access to a event bank 'EVTB' for Level = 2 or 4
  • dcrkuta.c - one-step tracking a particle through a magnetic field using Runge-Kutta method
  • dcrotds.c - vector transformation from Detector to Sector Coordinate System and reverse { DCS <--> SCS }
  • dcstat.c - keep tallies on different types of analysis cuts; for multithreaded applications: allocate TRstat structure locally for every thread and use 'dcstatadd(TRstat *stat)' to accumulate all statistic together
  • dcswim.c - track swimming (traj. propagation) in magnetic field. Calculates track points on detector planes
  • dcswimvt.c - back swimming in the PT magnetic field; calculates track points on the target plane#1
  • dctodst.c - writes the reconstructed event into DST
  • dctrfit.c - perform a fit of the track candidate
  • dctrmat.c - calculates transport matrix for one step; records matrix at point of layer penetration
  • sclib.c - find SC slab "idsc" in data the track points to
  • stlib.c - start counter library
  • tglib.c - tagger library (rewritten a la offline version)
  • trbos.c
  • trconfig.c
  • trinit.c
  • trlib.c

Road Maps

Road maps are used by Pattern Recognition. To generate road map following command must be used:

makefulldict 

If ran without parameters it will show usage:

Usage: makefulldict IWmin IWmax Pmin Pmax Charge [THETAmin THETAmax]
                       GeV/c                            Degree
Output will go to the prlink.bos file in the current directory.

Go to $CLON_PARMS/cmon/prlib/DICT/ directory and adjust local cmon.config file. Then run makefullduct, for example:

makefulldict 1 130 0.5 1.0 -1.

It will generate roads scaning starting point from wire 1 to wire 130, momentum from 0.5GeV/c to 1.0GeV/c for negative particles.

Several dictionaries can be merged by running

mergedict

with following usage:

Usage: mergedict <dictfilename1> <dictfilename2>
Output will go to the prlink.bos file in the current directory.

Calibration Constants

Calibration constants are stored in offline database. It can be accessed at this link.

Calibration constants must be extracted from offline database and stored in a form of map files to be used by CMON. To extract information from database to maps, db2clonmap.pl script must be used. It can be run with following parameters:

db2clonmap.pl runMin=<minimum run> runMax=<maximum run> \
      [system=<system name>] \
      [skip_run_control=<non-zero to skip RUN_CONTROL system>] \
      [runIndexTable=<run index table name>] \
      [time=<date/time of validity>] \
      [hostname=<db server hostname>] \
      [quiet=<non-zero to supress printout>] \
      [help=<non-zero for usage message]

For example, to extract tagger calibration parameters goto $CLON_PARMS/Maps and run

db2clonmap.pl system=TAG_CALIB runMin=1 runMax=1000000

it will create subdirectory Maps_1-1000000, create TAG_CALIB.map inside and fill it up.

Default runIndexTable is RunIndex. For some run periods calibration constants are stored under different runIndexTable name, for example for g8b runIndexTable was calib_user.RunIndexg8b. In that case use command

db2clonmap.pl system=TAG_CALIB runIndexTable=calib_user.RunIndexg8b runMin=1 runMax=1000000
db2clonmap.pl system=SC_CALIBRATIONS_V2 runIndexTable=calib_user.RunIndexg9 runMin=1 runMax=1000000

To extract entire database run following command (it is important to use 'skip_run_control=1', otherwise it takes forever)

db2clonmap.pl skip_run_control=1 runMin=1 runMax=1000000

After it finished, move all *.map files to $CLON_PARMS/Maps.

Calibration databases can be dumped in a form of ascii file using following procedure:

$CLAS_TOOLS/caldb/caldb_show_constants_run.pl s=TAG_CALIB ss=tag_t i=ci r=40000 > /home/boiarino/z123

JUST FOR INFORMATION, DO NOT DO THAT: To write it back remove first two lines and type:

$CLAS_TOOLS/caldb/caldb_write_and_link.pl s=TAG_CALIB ss=tag_t i=ci min=40000 max=50000 ci="test1" f=/home/boiarino/z123