Since Hiro/catalin is about to redo the auau ltr+gtrs I promised to look into some issues with fitted tracks, and with the the dst. After several discussion with Kris, and also with Catalin, Hiro it has become clear that there are conceptual problems with the way with we analyze FFS data. The issue comes from the fact that since both FFS and FS tracks are picked from the same Dst class namely BdstFsTrack , but has been subjected to different data handling due to the refitting, the FFS tracks that belong to a FS tracks are quite different from FFS tracks that are NOT in the full FS. A couple of point n The projections to the vertex is different (only T1,T2 are used for fit.) The momentum is determined from just one measurement (D2) while the full fs uses 4-5 chambers and the 1-2 additional mom measurements. n Thus the two classes of FFS tracks have different basics (cuts, momentum resolution) which is not a proper way to do the analysis. Most of the confusion comes from using only one track representation in the dst's. Were we to include in the dst (default root file) a branch FS (with only pure FS tracks- requires bfs) and a branch FFS that has all FFS the dst separation would be clean (and the files probably not to much bigger). I am sure this probably what was intended in the original design of Bdst classes that has both BdstFfsTrack and BdstFsTrack. The drawback is of course that the subsequent steps will have to do analysis with two different classes; I do not think this is a big deal - there must already be different flags depending on what acceptance is used, whether there are tracks in T5 etc. There are some relatively simple issues with the refitting of FS and FFS storing data in the same variables in the gtr files ; this can easily be solved by just adding some data members. Thus my conclusions are: - add some data members for the refitted values in FFS and FS - Add the FSS branch to the dst produced following the gtr's - Separate the analysis according to FFS and FS. - This should be included in the upcoming final(?) pass of the auau data and the cucu 200 gev data. Please comment to this proposal. Flemming Some additional Notes. First the simple stuff. -- I want to store the chisq and the fit parameters a fT1Track. This is straightforward for full Fs tracks, -- it is also simple for ffs tracks . All that is needed is additional parameters for the tracks, just a few elements. The complication is that I discovered that the refit set's the ffs momentum; Unfortunately the parameter is the D2 matched track mom, and I do not think one wants to overwrite this value. Somehow the proper solution is to add a specfic momentum to the ffstrack, called fFittedMomentum, and load that in the fit-module. Some of the needed changed for the track fit parameters and separation of FFS and FS parameters in the brat data classes have already been made and in CVS. ---------------------------------------------------------------- Flemming Videbaek Physics Department Brookhaven National Laboratory e-mail: videbaek@bnl.gov phone: 631-344-4106 _______________________________________________ Brahms-dev-l mailing list Brahms-dev-l@lists.bnl.gov http://lists.bnl.gov/mailman/listinfo/brahms-dev-lReceived on Sat Oct 15 14:47:30 2005
This archive was generated by hypermail 2.1.8 : Sat Oct 15 2005 - 14:47:41 EDT