Summary of changes:
- Change sign of deflection length (previously inconsistent with TRAP model,
to be clarified which one is used in reality)
- decrease windows for track matching, possible due to better tracklet parameters
- Change handling of output tree (not owned by current directory after creation)
- check for a = 0 in pt-calculation
- added GetLabel to AliTRDtrackletGTU
- change CookLabel according to other changes (offset of 1 in label)
Update of the tag system. 1. Correct implementation of the event ID - period, orbit ID and bunch-crossing number. 2. Adding the information relevant to the reconstruction of the first LHC data which will be taken with a bunch-srossing trigger. The information is related to same basic quantities related to the activity in SPD and VZERO. This information should be enough in order to identify the events on which one would run the full aliroot reconstruction. 3. A bug fix in the trigger and trigger-cluster selection in AliEventTagCuts.cxx (Panos)
major TRD reconstruction update
- update clusterizer for flexible switch between methods to calculate
r-phi cluster position. Currently the following methods are supported:
- COG - center of gravity
- LUT - look up table [default]
- GAUS - gauss approximation [testing phase]
- move position calculation related code from clusterizer to the
cluster
itself
- steering options for cluster position calculation
implemented at Reconstructor level.
- user help via static method
AliTRDReconstructor::Options() function
- consistent pad row cross implementation in the
tracklet. Reference
radial position for pads crossing pad row is
recalculated based on z
cluster charge distribution.
- miscellaneous
major TRD reconstruction update
- update clusterizer for flexible switch between methods to calculate
r-phi cluster position. Currently the following methods are supported:
- COG - center of gravity
- LUT - look up table [default]
- GAUS - gauss approximation [testing phase - to production]
- move position calculation related code from clusterizer to the cluster
itself
- steering options for cluster position calculation implemented at Reconstructor level.
- user help via static method AliTRDReconstructor::Options() function
- consistent pad row cross implementation in the tracklet. Reference
radial position for pads crossing pad row is recalculated based on z
cluster charge distribution.
- miscellaneous
The variable fSeed is not really needed as the fitter is called in a way that the seeding is done only once. Apparently its presence caused a bug when the fitter is used to fit set of tracks and the fit fails at some point. Then fSeed was not properly reset and all the consequtive fits used the seed at the point of failure. Thanks for Dariusz for reporting the problem.
All:
- Speed up the computation/selection of primary track candidates (by up to a factor 100)
- Improve computation of initial covariance matrices for Kalman
- Use specific covariance matrices when there is no magnetic field
- fix potential bug by reordering cluster/track selection
- fix selection of track candidates when there is no magnetic field
- new protection against tracking divergences (see below)
AliMUONRecoParam:
- change Cosmic settings
- New parameter: maximum number of trigger tracks above which the tracking is cancelled
- New parameter: maximum number of track candidates above which the tracking abort
Several changes, mostly technical in nature: some mother volumes have been
added to increase the speed of overlap checking (and some typos 'heigth' -->
'height').
The functional changes are an increased tolerance between the ladder and the
mounting blocks to remove overlaps with full misalignment and a fix to put
back the piece of cooling tube that was missing at the end of the ladders
The cooling tube code has been simplified in the process.
New classes to get a much improved performance in the computation of
the contours of the various elements in the tracker (manu, buspatch,
detection element, chamber). With this improved algorithm, we can now
stop to use the padstore.root file (which used to contain pre-computed
contours, as that computation was really long) in mchview.
At the same time we achieve a better separation of the contour computation
and graphical representation. So AliMUONPainterContourMaker and
AliMUONPainterContour classes are removed, and the contour and
contour making are now part of the more general geometry library,
instead of the graphics one.
Modifications for TOFnoiseda:
- acquisition (matching) window now read from TOFFEE
- threshold set in a Config file in OCDB (TOF/Calib/ConfigNoise)
- modified test histograms added
marian [Mon, 6 Apr 2009 14:21:02 +0000 (14:21 +0000)]
TPCCEda.cxx.diff Simplification of the AMORE DB part
TPCPEDESTALda.cxx.diff Simplification of the AMORE DB part
TPCPULSERda.cxx.diff Simplification of the AMORE DB part
Last set of modifications for reducing size of SDD correction maps in OCDB. Flag to control the use of these maps moved from AliITSCalibrationSDD to AliITSRecoParam (F. Prino)
Initial version of the new RCU format decoder. The user interface is the same as for the fast version of the old format one. The decoder will be tested further by Magnus and as soon as we get confidence in the RCU format and the decoder itself, we will remove the fast version of the old decoder and merge the slow one into the same code.
Most of these updates are related to the offline software needed for the pixel trigger, for instance:
* Managing pixel trigger conditions data.
* Simulating generation of fast-or signals from SPD.
* Simulating pixel trigger response.
See also class schematics at http://tydes.home.cern.ch/tydes/doc/PitOffline
The treatment of dead/noisy pixels has also changed, both in the simulation and the reconstruction step. During simulation, it will be possible (via AliITSSimuParam) to choose whether dead/noisy pixels should be removed/added. Default is to add noisy pixels (which may be none, as one likes), but not to remove dead pixels. A similar choice for local reconstruction (via AliITSRecoParam) removes by default both noisy and dead pixels before the clusterization step. This is the safe way to proceed. However, in case of real data, one might speed up by skipping the removal of dead pixels, since they could not appear - by definition - in this case. With this strategy there will hopefully be less confusion on the meaning of dead and noisy pixels.
AliITS:
* CreateTriggerDetector is passing the AliITSTriggerConditions to AliITSTrigger upon creation.
* In HitsToDigits and SDigitsToDigits, AliITSDetTypeSim object (which takes care of generating the AliITSFOSignalsSPD) will be called to reset the fast-or signals before the event and to add noise and store them after wards (using the AliITSLoader).
* Digits2Raw will take the AliITSFOSignalsSPD from the AliITSLoader in order to pass them to AliITSDDLRawData::RawDataSPD, where they will be added to the simulated raw data.
AliITSClusterFinderV2SPD:
* The two FindClustersSPD methods (one for raw data, one for digits) are now filling the FastOrFiredMap (TBits) member of AliITSDetTypeRec. Fast-or signals come either directly from raw data or from AliITSFOSignalsSPD object in digits file.
* Before clustering, there is a possibility to remove noisy and dead pixels from the fired pixel map. Default is to remove both noisy and dead, but may be changed via AliITSRecoParam. In case of real data one might want to speed up by skipping removing dead pixels (which anyway should not appear in this case).
* Henrik's comment: The name of this class is a bit misleading, since it does several things apart from clustering. It might be discussed if one should separate into something like AliITSLocalReconstructionSPD and AliITSClusterFinderSPD. For instance, with the existing code, it is a bit strange that the cluster finder needs to know how the raw data is written, what a digit is, and so on.
AliITSDDLRawData:
* Modified to be able to write fast-or signals into raw data. The signals are fed from outside to this class (see AliITS::Digits2Raw).
* The procedure that converts digits into raw data (inside GetDigitsSPD method) was reimplemented, since the old code was unreadable.
AliITSDetTypeRec:
* Added pointer to AliITSFOSignalsSPD. This is used to let the AliITSClusterFinderV2SPD access the fast-or signals generated during simulation.
* Added pointer to AliITSTriggerConditions. This is used to remove the inactive chips from the fFastOrFiredMap (TBits), needed in real data. This is done in method RemoveFastOrFiredInActive. The fFastorFiredMap will eventually go into the AliMultiplicity class in the ESD.
AliITSDetTypeSim:
* Added AliITSFOGeneratorSPD object to take care of simulating the generation of fast-or signals.
* Added pointer to noisy calibration objects read from OCDB.
* Added pointer to AliITSTriggerConditions, needed by the AliITSFOGeneratorSPD object.
* GetCalibration method reads AliITSFOEfficiencySPD and AliITSFONoiseSPD objects from OCDB and passes them to the AliITSFOGeneratorSPD object.
* Takes care of storing the generated fast-or signals via AliITSLoader.
AliITSLoader:
* Added an AliBaseLoader for the AliITSFOSignalsSPD object that should be stored in the 'ITS.Digits.root' file. It is placed as a second loader (AliObjectLoader in this case) in the digits data loader, next to the digits tree loader (AliTreeLoader). Its name is the same as the class - 'AliITSFOSignalsSPD'.
* The Load/UnloadDigits methods are presently not taking care of this new loader. It is not clear which is the best way to solve this. It may require changes to the AliLoader class itself.
AliITSRawStreamSPD:
* Added method GetDDLID, which may be useful in places where there is no access to the AliRawReader object (just calls the AliRawReader::GetDDLID).
* Minor modifications to parsing of extra calibration headers.
AliITSsimulationSPD:
* Method RemoveDeadPixels slightly modified. Does not require the AliITSmodule as input anymore.
* Method AddNoisyPixels added. This will allow the possibility to explicitly add noisy pixels as digits, which could then also give rise to fast-or signals. For each noisy pixel, a noise 10 times the threshold will be added to the map.
* For each digit found in method FrompListToDigits (basically SDigits to Digits), the AliITSDetTypeSim::ProcessSPDDigitForFastOr will be called.
AliITSTrigger:
* Added object AliITSTriggerFOProcessor as member. This will take care of simulating the response of the pixel trigger system upon receiving a set of fast-or signals. It will need an AliITSTriggerConditions object, which normally is read from OCDB.
* The CreateInputs method will create inputs based on the AliITSTriggerConditions object.
* The Trigger method reads the AliITSFOSignalsSPD via AliITSLoader and asks the AliITSTriggerFOProcessor for a decision (yes/no) from each of the defined output lines. If yes, SetInput will be called.
ITSSPDSCANda:
* Modified initial comments, since this DA should not be used for run type DAQ_FO_UNIF_SCAN, as originally planned.
* Removed the temporary patch of reading chip present information, which is no longer needed.
* Mean threshold scan header format changed. The current DAC value will now be replaced by RowSpan and current DAC value should be taken from TPLow, assuming it is the same for all 6 half-staves of a half-sector.
AliITSFOEfficiencySPD:
* New class used to store information about the fast-or efficiency. The values stored are the probability that a pixel hit will generate a fast-or signal. In this base class there is one value for each pixel chip. The daughter class AliITSFOEfficiencySPDColumn may be used if there should be even a difference per pixel chip column.
* Will be put in OCDB as "ITS/Calib/SPDFOEfficiency". A new file with 100% efficiency for all is put as default.
AliITSFOEfficiencySPDColumn:
* New class that inherits from AliITSFOEfficiencySPD. May be used instead of base class, when the effiency values differ between single pixel chip columns.
AliITSFONoiseSPD:
* New class used to store information about the fast-or noise. The values stored are the probability that a pixel chip will generate a fast-or signal originating from random noise.
* Will be put in OCDB as "ITS/Calib/SPDFONoise". A new file with 0% noise for all is put as default.
AliITSFOGeneratorSPD:
* New class taking care of generation of fast-or signals in simulation.
* The ProcessPixelHit method should be called for each fired pixel found in the digitization procedure. The decision if the pixel will generate a fast-or signal will be based on an AliITSFOEfficiencySPD object which gives the probability.
* After digitization, method ProcessNoise should be called, which will add noise according to probabilities taken from an AliITSFONoiseSPD object.
* The final result is a filled AliITSFOSignalsSPD object, a member of this class.
AliITSFOSignalsSPD:
* New class that is able to store 1200 bits per event, each saying wether a pixel chip gave a fast-or signal or not. This object will be stored in the 'ITS.Digits.root' file during simulation.
AliITSTriggerConditions:
* New class that represents the conditions of the pixel trigger system. This object will contain the same information as is being sent by the pixel trigger system at start-of-run to DCS file exchange server, and taken offline by the shuttle. Preprocessor code will store the object in OCDB as "ITS/Calib/PITConditions". A new file with some default conditions is placed there.
* Data members: run number, firmware version, input config version, parameters version, a list of AliITSTriggerAlgorithmConditions objects, a list of inactive chips.
AliITSTriggerAlgorithmConditions:
* New class that holds the information regarding a single pixel trigger algorithm.
* Data members: label, id, description, number of parameters, parameter values, parameter names.
AliITSTriggerFOProcessor:
* New class that is able to process the AliITSFOSignalsSPD in order to evaluate if a given pixel trigger output is activated.
* Uses an AliITSTriggerConditions object to know the parameters corresponding to the algorithms in use.
* New algorithms should be added here as soon as they are implemented online. Only a couple of the existing online algorithms have been implemented here so far.
* The two algorithms used in the old AliITSTrigger code, have been adapted to the new framework and are added to the AliITSTriggerConditions object in ITS/Calib/PITConditions.
AliITSSimuParam:
* Added flag saying whether noisy pixels should be added at the level of digitization (which could then also lead to additional fast-or signals).
* Added flag saying whether dead pixels should be removed at the level of digitization.
AliITSRecoParam:
* Added flag saying whether noisy pixels should be removed at the level of local reconstruction (default is true).
* Added flag saying whether dead pixels should be removed at the level of local reconstruction (default is true, but may for instance be put to false to speed up in case of real data, since we then know the dead pixels would not appear in any case).
---------------
REMOVED CLASSES
---------------
AliITSFastOrCalibrationSPD:
* This class is removed, since it is no longer needed. The TBits of fast-or configured chips will be part of AliITSTriggerConditions from now on (instead represented as inactive chips). The calibration file, which before was in "ITS/Calib/SPDFastOr" is removed as well.
Most of these updates are related to the offline software needed for the pixel trigger, for instance:
* Managing pixel trigger conditions data.
* Simulating generation of fast-or signals from SPD.
* Simulating pixel trigger response.
See also class schematics at http://tydes.home.cern.ch/tydes/doc/PitOffline
The treatment of dead/noisy pixels has also changed, both in the simulation and the reconstruction step. During simulation, it will be possible (via AliITSSimuParam) to choose whether dead/noisy pixels should be removed/added. Default is to add noisy pixels (which may be none, as one likes), but not to remove dead pixels. A similar choice for local reconstruction (via AliITSRecoParam) removes by default both noisy and dead pixels before the clusterization step. This is the safe way to proceed. However, in case of real data, one might speed up by skipping the removal of dead pixels, since they could not appear - by definition - in this case. With this strategy there will hopefully be less confusion on the meaning of dead and noisy pixels.
AliITS:
* CreateTriggerDetector is passing the AliITSTriggerConditions to AliITSTrigger upon creation.
* In HitsToDigits and SDigitsToDigits, AliITSDetTypeSim object (which takes care of generating the AliITSFOSignalsSPD) will be called to reset the fast-or signals before the event and to add noise and store them after wards (using the AliITSLoader).
* Digits2Raw will take the AliITSFOSignalsSPD from the AliITSLoader in order to pass them to AliITSDDLRawData::RawDataSPD, where they will be added to the simulated raw data.
AliITSClusterFinderV2SPD:
* The two FindClustersSPD methods (one for raw data, one for digits) are now filling the FastOrFiredMap (TBits) member of AliITSDetTypeRec. Fast-or signals come either directly from raw data or from AliITSFOSignalsSPD object in digits file.
* Before clustering, there is a possibility to remove noisy and dead pixels from the fired pixel map. Default is to remove both noisy and dead, but may be changed via AliITSRecoParam. In case of real data one might want to speed up by skipping removing dead pixels (which anyway should not appear in this case).
* Henrik's comment: The name of this class is a bit misleading, since it does several things apart from clustering. It might be discussed if one should separate into something like AliITSLocalReconstructionSPD and AliITSClusterFinderSPD. For instance, with the existing code, it is a bit strange that the cluster finder needs to know how the raw data is written, what a digit is, and so on.
AliITSDDLRawData:
* Modified to be able to write fast-or signals into raw data. The signals are fed from outside to this class (see AliITS::Digits2Raw).
* The procedure that converts digits into raw data (inside GetDigitsSPD method) was reimplemented, since the old code was unreadable.
AliITSDetTypeRec:
* Added pointer to AliITSFOSignalsSPD. This is used to let the AliITSClusterFinderV2SPD access the fast-or signals generated during simulation.
* Added pointer to AliITSTriggerConditions. This is used to remove the inactive chips from the fFastOrFiredMap (TBits), needed in real data. This is done in method RemoveFastOrFiredInActive. The fFastorFiredMap will eventually go into the AliMultiplicity class in the ESD.
AliITSDetTypeSim:
* Added AliITSFOGeneratorSPD object to take care of simulating the generation of fast-or signals.
* Added pointer to noisy calibration objects read from OCDB.
* Added pointer to AliITSTriggerConditions, needed by the AliITSFOGeneratorSPD object.
* GetCalibration method reads AliITSFOEfficiencySPD and AliITSFONoiseSPD objects from OCDB and passes them to the AliITSFOGeneratorSPD object.
* Takes care of storing the generated fast-or signals via AliITSLoader.
AliITSLoader:
* Added an AliBaseLoader for the AliITSFOSignalsSPD object that should be stored in the 'ITS.Digits.root' file. It is placed as a second loader (AliObjectLoader in this case) in the digits data loader, next to the digits tree loader (AliTreeLoader). Its name is the same as the class - 'AliITSFOSignalsSPD'.
* The Load/UnloadDigits methods are presently not taking care of this new loader. It is not clear which is the best way to solve this. It may require changes to the AliLoader class itself.
AliITSRawStreamSPD:
* Added method GetDDLID, which may be useful in places where there is no access to the AliRawReader object (just calls the AliRawReader::GetDDLID).
* Minor modifications to parsing of extra calibration headers.
AliITSsimulationSPD:
* Method RemoveDeadPixels slightly modified. Does not require the AliITSmodule as input anymore.
* Method AddNoisyPixels added. This will allow the possibility to explicitly add noisy pixels as digits, which could then also give rise to fast-or signals. For each noisy pixel, a noise 10 times the threshold will be added to the map.
* For each digit found in method FrompListToDigits (basically SDigits to Digits), the AliITSDetTypeSim::ProcessSPDDigitForFastOr will be called.
AliITSTrigger:
* Added object AliITSTriggerFOProcessor as member. This will take care of simulating the response of the pixel trigger system upon receiving a set of fast-or signals. It will need an AliITSTriggerConditions object, which normally is read from OCDB.
* The CreateInputs method will create inputs based on the AliITSTriggerConditions object.
* The Trigger method reads the AliITSFOSignalsSPD via AliITSLoader and asks the AliITSTriggerFOProcessor for a decision (yes/no) from each of the defined output lines. If yes, SetInput will be called.
ITSSPDSCANda:
* Modified initial comments, since this DA should not be used for run type DAQ_FO_UNIF_SCAN, as originally planned.
* Removed the temporary patch of reading chip present information, which is no longer needed.
* Mean threshold scan header format changed. The current DAC value will now be replaced by RowSpan and current DAC value should be taken from TPLow, assuming it is the same for all 6 half-staves of a half-sector.
AliITSFOEfficiencySPD:
* New class used to store information about the fast-or efficiency. The values stored are the probability that a pixel hit will generate a fast-or signal. In this base class there is one value for each pixel chip. The daughter class AliITSFOEfficiencySPDColumn may be used if there should be even a difference per pixel chip column.
* Will be put in OCDB as "ITS/Calib/SPDFOEfficiency". A new file with 100% efficiency for all is put as default.
AliITSFOEfficiencySPDColumn:
* New class that inherits from AliITSFOEfficiencySPD. May be used instead of base class, when the effiency values differ between single pixel chip columns.
AliITSFONoiseSPD:
* New class used to store information about the fast-or noise. The values stored are the probability that a pixel chip will generate a fast-or signal originating from random noise.
* Will be put in OCDB as "ITS/Calib/SPDFONoise". A new file with 0% noise for all is put as default.
AliITSFOGeneratorSPD:
* New class taking care of generation of fast-or signals in simulation.
* The ProcessPixelHit method should be called for each fired pixel found in the digitization procedure. The decision if the pixel will generate a fast-or signal will be based on an AliITSFOEfficiencySPD object which gives the probability.
* After digitization, method ProcessNoise should be called, which will add noise according to probabilities taken from an AliITSFONoiseSPD object.
* The final result is a filled AliITSFOSignalsSPD object, a member of this class.
AliITSFOSignalsSPD:
* New class that is able to store 1200 bits per event, each saying wether a pixel chip gave a fast-or signal or not. This object will be stored in the 'ITS.Digits.root' file during simulation.
AliITSTriggerConditions:
* New class that represents the conditions of the pixel trigger system. This object will contain the same information as is being sent by the pixel trigger system at start-of-run to DCS file exchange server, and taken offline by the shuttle. Preprocessor code will store the object in OCDB as "ITS/Calib/PITConditions". A new file with some default conditions is placed there.
* Data members: run number, firmware version, input config version, parameters version, a list of AliITSTriggerAlgorithmConditions objects, a list of inactive chips.
AliITSTriggerAlgorithmConditions:
* New class that holds the information regarding a single pixel trigger algorithm.
* Data members: label, id, description, number of parameters, parameter values, parameter names.
AliITSTriggerFOProcessor:
* New class that is able to process the AliITSFOSignalsSPD in order to evaluate if a given pixel trigger output is activated.
* Uses an AliITSTriggerConditions object to know the parameters corresponding to the algorithms in use.
* New algorithms should be added here as soon as they are implemented online. Only a couple of the existing online algorithms have been implemented here so far.
* The two algorithms used in the old AliITSTrigger code, have been adapted to the new framework and are added to the AliITSTriggerConditions object in ITS/Calib/PITConditions.
AliITSSimuParam:
* Added flag saying whether noisy pixels should be added at the level of digitization (which could then also lead to additional fast-or signals).
* Added flag saying whether dead pixels should be removed at the level of digitization.
AliITSRecoParam:
* Added flag saying whether noisy pixels should be removed at the level of local reconstruction (default is true).
* Added flag saying whether dead pixels should be removed at the level of local reconstruction (default is true, but may for instance be put to false to speed up in case of real data, since we then know the dead pixels would not appear in any case).
---------------
REMOVED CLASSES
---------------
AliITSFastOrCalibrationSPD:
* This class is removed, since it is no longer needed. The TBits of fast-or configured chips will be part of AliITSTriggerConditions from now on (instead represented as inactive chips). The calibration file, which before was in "ITS/Calib/SPDFastOr" is removed as well.
ivana [Fri, 3 Apr 2009 14:17:12 +0000 (14:17 +0000)]
Changes in tracker raw data decoding and mchview :
- raw data decoder now keep tracks of the total number of (parity,
glitch, padding) errors, so we can easily sort events based on this
- AliMUONRawStreamTrackerHP no longer derives from VRawStreamTracker
(which will be phased out at some point)
- merging of AliMUONTrackerRawDataMaker and
AliMUONTrackerCalibratedDataMaker classes into a single one
AliMUONTrackerDataMaker, which, in addition, skip buspatches with
parity error, and can read from memory
Other changes related to the change of derivation in
AliMUONRawStreamTrackerHP and to the fact that we no longer allow the
switch between old and new (HP) decoders (the usage of the former is now
deprecated)
git://git.uio.no / u / mrichter / AliRoot.git / log