This is a list of improvements of marple with respect to the release of ARTE. The ARTE subrelease notes contain information about main changes as well without technicalities.
- VERSION 3.04
- Included in ARTE version -03-09-r5 (9.11.2001); could be compiled with ARTE-03-01- ...
- marple.hh
- - add MARPLEIDRTRARICH method
- RcPoint.hh
- alsoRcPoint.icc, RcClusPoint, RcMatchPoint.C, RcPoint.C, RcSegClusBuilder.C, RcSegPoint.C , maFindTarget.C , matchotrecal.C, matchsvdrich.C - speeding up routines : use fast algorithm of Matrix
- RcMsOfGesl.C, marpleinit.C
- - tuning prolongation algorithm in case no chambers in magnet
- matchsvdpc.C
- - warning for efficiency estimations
- marpleCmd.C
- - warning for MASVDECALNORTRA
- VERSION 3.03
- Included in ARTE version -03-09-r1 (9.07.2001); could be compiled with ARTE-03-01- ...
- maRichSegment.C
- - polishing
- marplekuip.cdf, marpleCmd.C
- - 2 new commands :
- MASVDECALET ET
- MASVDECALNORTRA nmSVD nmECAL
- marpleinit.C
- - new variable MARPLE_SEET
- matchrtrarich.C
- - new algorithm RTRA-RIH match
- matchsvdecal.C
- - check if SVD or ECAL are matched to RTRA already if required(defined by MASVDECALNORTRA)
- VERSION 3.02
- Included in ARTE version -03-08-r3 (27.11.2000); could be compiled with ARTE-03-01- ...
- marple.hh
- - include enumeration type MARPLETYPEOFMATCH for DQ, and switches of reconstruction.
- RcMatchConst.C
- - add call of DQ_RECO_marple_evnt which pass variable to Data Quality package.
- MARPLETYPEOFMATCH
- - type and global variable maPresentMatch to specify which algorithm is working at the moment. To be used internally and in DQ. The algorithms corrected are in : matchotrecal.C, matchotrrich.C, matchrichecal.C, matchsvdecal.C, matchsvdot.C, matchsvdpc.C,matchsvdrich.C
- RcPoint.C
- - supress number of messages
- matchsvdpc.C
- - momentum estimated by minimizing dTX in the middle of magnet is saved as momentum of match ( before it was RPNT-PC momentum)
- rcmesm.C
- - check on existance of required superlayer is added
- VERSION 3.01
- Included in ARTE version -03-08-r1 (16.10.2000); could be compiled with ARTE-03-01- ...
- RcSegRpntBuilder.C
- - in case chi2 is specified the algorithm of matchInTheMiddle() takes estimation of momentum correctly (bug fixed)
- VERSION 3.00
- Included in ARTE version -03-07-r5 (7.08.2000); could be compiled with ARTE-03-01- ...
- matchsvdpc
- - !!!!
new function to match SVD segments with RPNT created at first point of PC segment before propogation through magnet. Matching is done in efficient way and 5 sets of constants are available. In this case magnet information is not taken into account. See Match SVD - PC-OTR for more details.- matchotrecal
- - soft cut in deltaX reduced to 15 cm;
- matchotrrich
- - soft cut in deltaTX reduced to 8mrad; in deltaTY to 25 mrad;
- matchsvdot
- - soft cuts reduced to deltaX < 3; soft deltaTX increased to 15 mrad; small bug corrected;
- matchsvdecalt
- - same as matchsvdecal_() but one can match clusters with Et (Ktrig) less then threshold; soft cuts are reduced to deltax < 3, deltaY < 5;
- MARPLE_RISEID
- - cut on RICH identification with RISE can be tunned from kumac with kuip command MARPLE/MARICHSIGMASQR <Float>. the default is set to 50.
- maRichSegment
- - function to identified RICH segment. if it returns 1 MARPLE supposes that rseg is of RISE type.
- RcMatchBuilder, RcMatchPoint
- - to speed up algorithms most of functions return the constant reference to the vector of objects;
- RcMatchBuilder::saveRSEG
- - set not used variable to 0 to avoid NonValid numbers in RSEG; rseg.nvar is filled with number of parameters (X,Y,TX,TY,PE) used for chi2 calculations; the match segment is connected to MTRA if both original parts refers to the same MTRA. Only one common MTRA is taken into account.
- RcSegRpntBuilder
- - new class which do match between RcSegPoint and RcRpntPoint. It is needed since the algorithm to match is not standard.
- RcMsOfGesl
- - average Energy losses of the particles in the GESL are saved as well. The valies are fixed in the Look-Up table in the rcmesm.C
- rcmesm
- - C++ function replaces Fortran one. Function takes one more parameter as well. The MS values are fixed in Look-Up table and not calculated from geometry anymore.
- RcPoint
- - new function correctEnergyLoss is added to recover energy looses during tracing through the medium. The prolongate() is corrected accordingly. new function estimateParFromPoint is introduced. It allows to estimate unknown parameter using the another RcPoint. The variables to take into account { X, Y, TX or TY\} could be choosen. The function is not optimized yet; little bug in MS correction fixed; some time optimization; if the initialization of marple is not performed the program does not crash anymore but returns a lot of warnings. The RKMAIN is used for propogation in this case.
- Rc*Point
- - const mtraList(), hitbList()
- RcMatchPoint
- - delta* member data could be changed by setDelta* functions; ostream operator takes care of delta* as well.
- RcRpntPoint
- - new class to operate with RPNT ArteTable
- maFindTarget
- - if track far away take (0,0,-3) for first iteration; consider targets only if the GTAR->eff > 0 and x, y < 1.5cm
- VERSION 2.13
- Included in ARTE version -03-06-r9 (29.05.2000); could be compiled with ARTE-03-01- ...
- matchsvdrich.C
- - the cut on RICH hypothesis for given momentum is reduced to 5 sigma. Also a small bux is fixed.
- VERSION 2.12
- Included in ARTE version -03-06-r6 (04.05.2000); could be compiled with ARTE-03-01- ...
- RcMatchPoint
- - slight time optimization, loop is based on iterators.
- matchsvdot,matchotrrich, matchotrecal
- - optimization of parameters using "golden minimum bias run" 14577. The alighment of OTR was taken into account which allows to reduce a lot of background. Misalighment on y,ty is still seen.
- VERSION 2.11
- Included in ARTE version -03-06-r4 (11.04.2000); could be compiled with ARTE-03-01- ...
- RcMatchBuilder
- - create RSEG and relations only for matches which are not marked negative.
- matchsvdrich
- - uses RISE identification to reduce background.
- VERSION 2.10
- Included in ARTE version -03-06-r2 (16.03.2000); could be compiled with ARTE-03-01- ...
- RcMatchConst
- - correction for the case when points are located at different Z position
- RcMatchConst, RcMatchBuilder, RcSegClusBuilder
- - if the chi2 > RcMatchConst::maxConsideredChi2( =1000) the chi2 cut is not performed and no check on number of combination with given object is performed
- RcMatchPoint
- - New data members are added to the class for convenience. They keep information about difference between parameters of the objects to be matched. The functions of classes RcMatchBuilder, RcSegClusBuilder are affected as well.
- RcSegClusBuilder
- - new member function stat() - check MARPLE_DEBUG flag automatically. All match functions are corrected respectively.
- matchsvdecal, matchrichecal
- - Only clusters with energy > 1 GeV are considered by match.
- matchsvdmt
- - chi2 cut is removed to take into account CATS segments. Probably will be restored in future version of ARTE. Soft constants are added.
- matchsvdot
- - from this version it takes OTR tracks with magn bit set into consideration as well. In case the OTR momentum is not available the programs estimates it.
- matchsvdrich
- - estimation of momentum is performed only for the case magnet ON.
- VERSION 2.09
- Included in ARTE version -03-06-r1 (25.02.2000); could be compiled with ARTE-03-01- ...
- the standard match routines consider only segments from standalone reconstruction, which does have a positive rseg->cmp (Rsegc::bitsign == 0)
- matchsvdotr - cut on number of combination with given track is omitted
- matchotrecal - segments from magnet chambers are taken into account; the cut on \delta X reduced to 4 cm; for better efficiency the TYotr set to Yotr/Zotr (temporary)
- mamakent, RcMatchBuilder, RcTuples - debug option tuning
- VERSION 2.08
- Included in ARTE version -03-05-r2 (31.10.2000); could be compiled with ARTE-03-01- ...
- changes for native hera-b compiler
- matchotrecal - is added
- VERSION 2.07
- Included in ARTE version -03-05-r1 (18.12.1999); could be compiled with ARTE-03-01- ...
- matchotrrich
- - is added
- matchsvdot
- - updated for real data, the momentum estimation bug fixed the momentum estimation bug fixed. consider CATS segments as well as HOLMES one.
- RcSegClusBuilder
- - time optimization; Bug in matchWithMomentum is removed (not serious, it was softer rejection)
- matchsvdecal.C
- - cut on | delta(y) | is 2.5 cm instead of 4 cm;
- matchsvdrich.C
- - no cut on quality of RICH segments is performed
- VERSION 2.06
- Included in ARTE version -03-04-r1 (11.11.1999); could be compiled with ARTE-03-01- ...
- CATS segments are considered by matches
- The momentum estimation VDS-RICH is tuned.
- VERSION 2.05
- Included in ARTE version -03-01-r2 (27.09.1999); could be compiled with ARTE-03-01- ...
- Time optimization is performed. Most affected functions :
- RcPoint::estimateParFrom..(..)
- RcPoint::lineswim(..)
- RcPoint::swimRunge(..)
- the macro MARPLEPERFORMACE for time optimization is added
- RcClusPoint - estimateAngles function is removed
- VERSION 2.03
- Included in ARTE version -03-01-r1 (09.08.1999); could be compiled with ARTE-02-04 - ...
- The code completely corrected for new C++ interface.
- CPU time consume optimization
- fortran interface to manipulate MARPLE objects became obsolete.
- maTakeRccl includes standard cut on clusters
- The new convention about RSEG->cmp flag is fulfill.
- VERSION 2.02
- Included in ARTE version-02-08-r1 (18.06.1999); could be compiled with ARTE-02-04 - ...
- matchrichecal
- - a estimation of momentum using coordinates of cluster and angles of RICH segments is performed.
- RcMatchBuilder
- - MARPLE can save match information into database if MARPLE_MAKERSEG flag set to 1 ( by means of MARPLE/MAKERSEG kuip command)
- RcMatchPoint
- - quality of match is calculated using probability function from CERNLIB ( prob)
- mamakentup
- - MARPLE can create ntuples with match information if MARPLE_DEBUG > 10 ( set by MARPLE/DEBUG kuip command) ntuple numbers are different for different match
- VERSION 2.01
- Included in ARTE version -02-07-r1 (23.04.1999); could be compiled with ARTE-02-04 - ...
- In this version the following match functions are available :
(*) - has to be reoptimized
- matchrsersf
- - match of 2 different types of segments magnet ON/OFF
- matchsvdmt
- - match of SVD segments - Main Tracker segments magnet ON/OFF
- matchsvdot*
- - match of SVD segments - OuterTracker segments magnet ON
- matchsvdecal*
- - match of SVD segments - ECAL clusters magnet ON/OFF
- matchsenomag
- - match of SVD segments - ECAL clusters magnet OFF (optional)
- matchrichecal*
- - match of RICH segments - ECAL clusters magnet ON/OFF
- The important change is :
The $\chi^2$ value of matching is calculated using only specified parameters of match (e.g. x and y). The reduced covariant matrix is used (here $\sigma_{xx}, \sigma_{xy}, \sigma_{yy}$). This bug fixing requires a reoptimization of the $\chi^2$ value for all matching functions.- From this version MARPLE takes care of magnet presence using GESL(LGESL_FLD)->mfl information.
- New class :
RcMatchFlag is added to specify parameters to be matched. The external interface remains the same.- Other changes :
- RcClusPoint
- - is diagonal elements of covariant matrix in RCCL bank equals 0 MARPLE set space resolution using values from HERA-B 97-150 note
- RcSegPoint
- - correctly consider RICH segments as they filled in ARTE-02-06
- RcMsOfGesl
- - don't consider GESL with shape == 0
- RcMatchConst
- - new function energy_delta_cut_ok(..) introduces energy depended delta cut as const/sqrt(E) for operations with clusters
- RcMatchBuilder
- - new option to fill ntuples with 'pulses of match' is added
- RcSegClusBuilder
- - reorganization of calls of different match algorithms. (see MARPLE documentation for details); delta cut depends on energy of cluster as Const/sqrt(E)
- RcTuples
- - reorganization of /deltas/ common block
- VERSION 1.04
- Included in ARTE version -02-06-r1 (05.02.1999); Could be compiled with ARTE-02-04 -..
- main changes :
from this version it is based completely on "new" C++ ARTE interface which could conflict with old C++ interface if they used both in one function. Parameters of match remain unchanged since ARTE-01. In case of principle changing in other reconstruction packages or in digitization the efficiency of matching could drop.- new classes :
A possibility to match Segments and Clusters by using more advanced functions is implemented in RcSegClusBuilder- new functions :
(optimization of parameters was done for SUSI-CARE in ARTE-01-09)
- matchsvdecal\
- - allows to match SVD segments with ECAL clusters
- matchsenomag
- - the matching of SDV segments and ECAL clusters for case magnet OFF ( faster than matchsvdecal)
- development and bug fixing up to version 01.04
- ARTE-02-05-r3
- The marple package is not yet updated to the new C++ interface. It means that the present version of marple would not work correctly with many-to-many relations. It could result in wrong output of matching procedure. From other side the base function like prolongation or Kalman formalism doesn't use this feature, so they should work as expected. Anyway it is not recommended to use marple with the releases from ARTE-02-04-r1 to ARTE-02-05-r3 .
- ARTE-02-05-r2
- RcClusPoint.C - updates to the new compiler option -... ;
they was done to follow warning you got with some new compiler option.- ARTE-02-05-r1
- RcPoint.C, marpleinit.C - bug (unresolved sort function)
corrected again
- ARTE-02-04-r3
- package is updated for ARTE-02-04 and
refit algorithm is corrected according to the new meaning
of old GEDE parameters ( gede->cos,gede->sin)- ARTE-01-09-r1
- maFindTarget.C - bug corrected (important for analysis based on MC sample with one target)
RcPoint.C - corrected constructor, algo.h included to resolve sort on OSF platform
marpleinit.C - algo.h included to resolve sort on OSF platform- ARTE-01-08-r4
- rcmesm.F - bug fixed
- ARTE-01-08-r3
- : maFindTarget.C - bug corrected (important for analysis based on MC sample with one target)
- VERSION 1.03
- Included in ARTE version -01-08-r2 (05.06.1998); Could be compiled with ARTE-01-05 - ARTE-01-08 (depends on include system)
- New "supplied products" are added : matchsvdot_(), matchrsersf_(..) ;
- refit, filter, updateARTE functions of RcPoint class family are added, RcPoint::findProlongation() became obsolete;
- estimation of the RcPoint parameter using prolongation to the target is added (RcPoint::estimatePARfromTAR(), RcPoint::estimatePEfromTAR() RcClusPoint::estimateAngles() );
- most functions return error code, normaly zero;
- Runge-Kutta method is invoked in the magnet area for all segments even with zero momentum ( or $ Q/p = 0 $ );
- call rkmain with fast calculation of the derivative matrix. The default value of the first rkmain parameter equals 2;
- check the covariant matrix of object by RcPoint::checkAndCorrectCov() which set MARPLE_INFINITY instead of zero diagonal elements;
- RcMatchPoint constructor : add lineswim parameter to constructor of (RcPoint, RcPoint);
- RcHitBPoint class is added to RcPoint class family;
- VERSION 1.02
- Included in ARTE version -01-08-r1 (02.03.1998); Could be compiled with ARTE-01-05 - ARTE-01-07 with old system of includes ( see mail by Peter Weyers in the desy.hera-b.software news ) and without saveRelation.C function
- adaption to "ARTE-01-08 include's style";
- Fortran interface is added;
- RcMatchBuilder : functions which save relations RSEG-RSEG, RSEG-RCCL, RTRA-RCCL and RTRA-RSEG are added ( file saveRelation.C ); function clearAll() is added;
- constructors of RcSegPoint, RcClusPoint and RcTraPoint using only ArtePointer to the corresponding ARTE table are added;
- constructors of RcPoint class family create objects with zero value of parameters if they are unknown. The covariant matrix has infinity ( =ARTE_FLARGEST/100) diagonal elements and zero non-diagonal ones;
- comments in doc++ style are added;
- VERSION 1.01
- first version of MARPLE package. Included in ARTE version -01-07; Could be compiled with ARTE-01-05 and ARTE-01-06; Could be updated for ARTE-01-04 with slight correction;
generated by doc++