Thanks for the nice answer, Konstantin. We are starting to think about calibrations over here, primarily tof calibrations, but since Flemming has implemeted the new parameter file I have been thinking about TPC "calibrations". The things I had been thinking about was : 1st loop 1) look at sync triggers to get the noisy pads Her special care should perhaps be taken to not target pads that hav irregularities for low time bins, so maybe just look at time bin 50-80. 2) look at trigger whatever to get the dead/ low performance pads. 2nd loop After removing these pads 3) do clustering on sync events to get a feel for if there is any real clusters/ back ground tracks - I would not expect any. 4) Use the pads not in any real clusters from 3 to get an estimate for the level of white noise in the TPCs somehow 5) do tracking on trigger 6 events to find the offset in hits for each row. I am currently looking at this. This is to fine tune the time offsets in each pad (at the level of ~0.1 cm) - this should improve the chi^2 very much. 3rd loop after adding info from 2nd loop to detectorparameters 6) test that row offsets are small and that chi^2 or confidence level is reasonable (tests withe noise) and eventually that chi^2 of cluster fits are reasonabl (long way to go). Experience will determine how often we need to do each of these calibrations. Question : 1) where does the drift velocity fit in ? 2) I don't want to program something that is already there ? Can your program do the first step here, Konstatin ? 3) Could we settle on some sort of standard calibration module so that we toss all the simple calibrations into a maybe more specialised module container and says go. Ingredients I can think of is : BrEventManager - for selecting triggers. How many loops A loop counter with an increment and set method. Maybe a switch for how severe calibrations are needed (light/medium/heavy) so that maybe every run has light every 10th has medium and so forth. So we would have BrTOFCalibrationModule that inherits from BrCalibrationModule and some manager that creates it for each run and looks to see what the max number of calibrations needed is. It knows if the calibration should be light etc. and then handles maybe through another calss the opening and closing of files. In the Event method of the calibration modules would then be if ( fLoop == 3 && fCalibType = BrCalibrationModule::hard ). And maybe the calibrations would be written to the BrXXXDetectorParameter in the EndRun after each loop or a new method EndLoop could be made. I can already forsee that if a calibration needs multiple runs my little scheeme does not work, but maybe someone could come up with something better. We would not mind making a quick draft for the class here at NBI. This is also a call for all that are doing calibrations or have done. Please commit your code and tell us where to look by emailing this list. What do you think Peter :-) --------------------------------- )-: |Peter H L Christiansen aka PAN @ NBI | |EMAIL : pchristi@nbi.dk | |OFFICE : Tb1 @ NBI | |PHONE : 353 25269 <- New!! | |SNAIL : Sdr. Fasanvej 14 ST 2000 F | |PHONE : 38 872042 | :-D --------------------------------- \-:
This archive was generated by hypermail 2b29 : Wed Oct 18 2000 - 13:13:52 EDT