Hi I have tried to look a bit at the different steps in the data selection in the MRS for run 5692 (90 deg. 350A) : The idea was to try and find out where there could be some big efficiency problems, i.e., set some rough limits. What I did was to look at the number of tracks per abs(vtxZ)<15 trigger 6 event with some additional cuts. The summary plots can be seen at : http://www.nbi.dk/~pchristi/BRAHMS/SPECTRA/mrs.html The 2 plots are : plot 1 = average tracks pr event accepted in TPM1, TPM2, MRS, (TEST), STATUS, TOF, PID TEST means momentum cut where applicaple, STATUS means require status == 1, TOF means require valid tof hit, PID means vtx cut, dy cut, slat cut, mass2 cut plot 2 = comparison of cuts (both axis are cuts) so the diagonal shows how many events that cut rejected and the off diagonal shows the correlation with other cuts. if 4 bins (vtx cut elipse, tof dy<3cm, tof slat (max 1), mass2 cut (no p dependence) if 5 bins (vtx y cut, vtx z cut, tof dy<3cm, tof slat (max 1), mass2 cut (no p dependence) When I compare 5% central with 10-20% central I find that the number of good tracks pr TPM2 track is constant (0.32+-1%) for both settings so that once you have a TPM2 track the rest is indpendent of centrality. If I compare the ratio between TPM2 tracks in the 2 settings I get : 0.544/0.869 = 0.626 which is very close to the ratio of our published dN/deta = 377/625 = 0.603 leaving room only for a small (~5%) centrality dependece on the number of TPM2 tracks (=yield) between 10-20 and 0-5 % centrality bins. In general there is a big loss going from TPM2 to good tracks, only 32% are good tracks. The steps are as follows : TPM2 to MRS : 0.8688 -> 0.4971 = 43 % loss Is that ok ? MRS to Status : 0.4971 -> 0.3821 = 23 % loss Only 5% is not a very severe fiducial cut (1 cm = geometrical 20%) that will be changed. The last 5% is almost all a status cut which means the difference between a 4 and a 5 sigma cut. Is this momentum dependent ? Status to TOF : 0.3821 -> 0.3369 = 12 % loss (missing slats ~ 2%) Geometry would suggest some amount like 5% ? The rest is hard to understand. Is there a momentum dependence ? Tof to Good : 0.3369 -> 0.2797 = 17% loss Almost entirely due to vertex cut, see below Ian suggested that I made a test with different momentum cuts. The biggest difference from these tests are the vertex cut. Before I turn to this let me just mention that the difference between the status cuts for 0.3-0.6 amd 1.2-1.5 is only 2 % so there could be an effect (there likely is) but it is small. A big effect 10 % is the vertex cuts see the pictures. There are 2 reasons for a large difference. The ok one is that the background should be bigger at low p. The not so ok is that our tracks could be worse determined at low p because of multiple scatterings and so the cuts are stricter there. At high p the cut has little effect and most of all the vertex cuts are uncorrelated. At low p vtx Z and vtx Y are very correlated but there are much more data cut away by vtx Y cut. Both cuts are 3 sigma, but dz is dominated by the bb vertex uncertainty and dy is dominated by the tracking quality. The way I interpret this is that the correlated stuff is background. The small ~1% extra in the vtx z cut is likely due to 1% of the BB vtx being really off. The big difference in the vtxY cut is due to 2 things I believe. The distribution is not gaussian (drift velocity change ?) (3%) and depends on momentum (0-3%). This seems to be a significant effect and maybe a losser momentum dependent cut should be applied. I don;t think we can drop the dy cut since it is a notorios background proton remover. The results indicate that the background level is around 10% 0.3 for low pt dropping to 0 above 1 GeV. My next plans are to go through the big drops and see if I can understand those and start to look at the FS in the same way. Cheers Peter -- :-) --------------------------- )-: Peter H L Christiansen @ NBI EMAIL : pchristi@nbi.dk OFFICE : Tb1@NBI (353 25269) HOME : Hjertensfrydsgade 3, st PHONE : 33330493(New)/ 40840492(mob.) :-D --------------------------- \-:
This archive was generated by hypermail 2b30 : Wed Jun 19 2002 - 13:35:01 EDT