Hi I have rewritten the clustering and the track follow module code. The idea is to include it in the upcoming brat 2, but still keep the old stuff in the beginning. I hope that interested people will have a look and play a little bit with it and some of their own local tracking code. Questions, comments and suggestions are as always welcome. Especially from the tpc and dc guys. NB! The new classes does not work with the track matching yet, so only local tracking can be done NB! I am on vacation the next 2 weeks but will read my email next weekend. Below is a short description of how to get it and how to run it. // You need brat version BRAT-1-16-8 (look in brat-config) // Checkout the programs from CVS cvs -d /afs/rhic/brahms/BRAHMS_CVS co brahms_app/pc_app/tpc cd brahms_app/pc_app/tpc/ClusterTest/ make // Tru doing the hit finding TestMainModule testnewcluster.cc -i <raw_data> -o reduced.root -H hist.root -e 100 // Look at the hit positions etc. TestMainModule testoutputdata.cc -i reduced.root -e 1 > out.txt cd ../TrackTest/ make // modify usetpclibs source usetpclibs // if you don't have any better copy the geometry and magnet file from // TpcMon cp ../TpcMon/90deg.* . // do the tracking TestMainModule testtracking.cc -i ../ClusterTest/reduced.root -o reducedTracks.root -e 100 -H hist.root cd ../TpcMon make personal // modify usetpclibs or cp ../TrackTest/usetpclibs . source usetpclibs // go to your favourite data files dir and do fx to look at the data with // the Alv Kjetil monitor /nfs/hehi/hehi25/hehi/pchristi/brahms_app/pc_app/tpc/TpcMon/tpc3d The idea behind rewriting was : * use better base classes * make it brat code compliant * make it easier to read and modify * remove what was never used Most of this have been done. The code (and structure) have been discussed with Christian but I still have a lot of small comments from him on details that I need to add after my vacation (2 weeks). I would also like to rewrite BrTPCPreProcess to make it more flexible. I have tested it with 1000 events and it did not crash or run out of track candidate space (I have seen it do that for larger search widths). The histograms looks reasonable. The bad chi**2 for TPM2 is well understood as coming from these time row effects. The bad chi**2 for TPM1 propably have most of the origin in the way we calculate errors so that big blops like we see in TPM1 should have the smallest error, but we know that in general the temporal resolution is really bad. As a final comment I would like to point out that there are 2 small test programs not mentioned above. The first one is in ClusterTest and uses the BrTpcFitCluster.h .cxx (I know there are some non optimal things in the class) and compares the weighted mean with a 2d MINUIT fit. It runs with TestMainModule testfitcluster.cc -i reduced.root and is preliminary but might be useful at a later stage. The second one simulates a straight track at (pad, time) = (0, 0) with some errors and look at what we get from the fit, and see that they agree. It is in TrackTest and called testmomres.cc Cheers Peter -- :-) --------------------------------- )-: |Peter H L Christiansen aka PAN @ NBI | |EMAIL : pchristi@nbi.dk | |OFFICE : Tb1 @ NBI | |PHONE : 353 25269 | |SNAIL : Hans Tavsensgade 35, 4th | |PHONE : 35 349336 | :-D --------------------------------- \-:
This archive was generated by hypermail 2b29 : Sun Jun 17 2001 - 14:58:56 EDT