On Mon, 5 Jun 2000, Ian Bearden wrote: > Of course we do need to >compare the various methods in detail, but it seems to me that this task >is only made easier by having code which works (and which several people >know about) in BRAT. I also am not sure which optimizations you are >speaking of, do you mean that there are parameters within the code that >have to be tuned? The working vertexing routine using MTPC1 tracks has been in the brat cvs since last Oct. Of course there are parameters have to be tuned in any reasonable vertexing routines. "What" tracks to used for vertexing, check distance to the vertex position of the tracks and delete if it's larger than the "cut" parameter and do vertexing again, "when" do you stop the iterations,... >Is this possible? I thought that for most cases even the "rich man's" >vertex method (using tracks in MTPC1) does not give a good measure of x >and y. The vertexing routine (BrVertex) using MTPC1 tracks gives a very good resolution in y and z. (RMS ~ 1 - 2mm)(see the figure (old) http://www.rhic.bnl.gov/brahms/WWW/private/computing/jhlee/cross/vertex_distribution.ps for MRS at 90 degree using central events.) By the way, the cpu time overhead using track information for vertexing is NOT an issue. >In any event, my >guess is that the "known" x and y position from RHIC will be better than >we can determine with any method, That's exactly the point. If the x and y information from RHIC is really accurate, it's good to calculate x and y positions of as well as z positions of a vertex to check our spectrometers are well-aligned with the RHIC/Beam coordinate. How do we know the mean of x and y distributions will be "zeros" unless you know the detectors are perfectly aligned with the RHIC/beam coordinate? That's what I meant "calibration". They are the only reference points (Vx and Vy) where the detector geometry and the beam geometry tie together in software. >It seems we *must* know how the >vertex is determined as well as the errors associated with the >determination. As far as I understand, the histogramming method is more difficult to estimate the errors on vertex determination. In any case, I just hope that the method we are going to use works well for us whatever it is. JH
This archive was generated by hypermail 2b29 : Mon Jun 05 2000 - 18:21:20 EDT