Minutes of the CALCOM analysis meeting, August 29, 1997. ======================================================= Agenda: A. Vlassov - Update on Cherenkov counter analysis L.C. Smith - Update on EC minimum ionizing particle calibration K. Joo - Some event reconstruction issues M. Kossov - Artificial NN approach to hit-based tracking ====================================================================== Due to the CLAS collaboration meeting there will be no CALCOM analysis meeting next week. Note: Hall B will be accessible from Friday September 5, 8am, until September 18. This is the only time period available to get all the sytems fully operational and checked out before the run. Volker Burkert ===================================================================== Individual reports: A. Vlassov: ---------- The analysis of Cerenkov efficiency data, based on hit-based tracking reconstruction for run #3166, was presented. The distribution of nondetected elastically scattered electrons was shown as a function of Cerenkov detector segment number and the Y-coordinate of CC matched point. It was shown, that * Most of inefficient events located around of the detector boundaries. * The efficiency of Cerenkov detector inside the region, which has to be effective due to GEANT estimation, is 94 % - 98 % depending on the different possible regions. ( The accuracy of hit-based tracking matching is not very good, and significant uncertainties in the projected hit positions may be present. ) * The electron efficiency decreases with increasing electron scattering angle. It was mentioned that the using of EC information can purify the selected electrons (from, say, pions) and it can helped to estimate Cerenkov detector efficiency a bit more accurately. Home page : http://www.cebaf.gov/~vlassov/ L.C. Smith: ---------- Calibration of the electromagnetic calorimeters involves two stages: 1) Hardware gain matching of PMTs for trigger uniformity 2) Determination of software gains of PMTs which optimize energy resolution and linearity of electromagnetic showers. Current strategy is to accomplish stage 1 using a low-threshold trigger to catch minimum-ionizing particles (MIP) passing through single pixels viewed by each PMT. Stage 2 requires a self-consistent minimization procedure using showers from electrons or pi-zero decay photons, in which the known electron energy from tracking or the invariant mass of the pi0 would constrain the fit. The viability of the MIP approach has been demonstrated in test setups using muons from cosmic-rays, where very clean MIP peaks were seen. However the vertical orientation of the calorimeters in Hall B is not optimum for cosmic-rays. Therefore data were taken in June to evaluate the possibility of using minimum-ionizing pions for this purpose. Runs were taken at 4.045 GeV with the torus field on (75%) and off. Preliminary analysis of the torus on runs show MIP-like peaks but with large tails presumably due to hadronic interactions. Although tube-to-tube gain variations are easily determined from fitting to these peaks, there are systematic theta-dependent shifts of 20-30% in the overall gains which are not seen in the cosmic-ray data. Tracking analysis of these runs shows a 4:1 ratio of positive to negative tracks entering the calorimeter within a momentum acceptance of 0.6-3.0 GeV. Possibly a large fraction of these are protons, which are not minimum ionizing at the low end of this range. CELEG/GSIM simulations are proceeding to estimate the proton:pion ratio as a function of theta to see whether this can account for the gains shifts. K. Joo: ------- (e,e'p) hydrogen elastic data were studied using run 3166 with 1.645 GeV beam energy. In this analysis, EC and negative tracks sin DC were used as electron id and positive tracks were used as proton id. By using kinematic constraints for hydrogen elastic (e,e'p) such as hadronic invariant mass, missing mass spectrum and missing momentum spectrum, 2935 events were reconstructed. Several kinematic cross checks were presented and they showed good evidences for hydrogen elastic (e,e'p) data. The beam energy was reconstructed using scattering angles for protons and electrons The peak value is 1.672 GeV and relative error is 1.64 % compared with the 1.645 GeV from accelerator MCC measurement. This analysis shows possible charge-dependence of the momentum measurements. Negative charge tracks shows 5% positive shift and positive charge tracks show 5% negative shift. This observation is worthwhile to be explored. Vertex coordinate reconstructions were also studied using current recsis. Z coordinate from both negative and positive tracks has strong dependency on scattering angles and negative and positive tracks shows opposite trends. It is possible that the current version of code for z coordinate reconstruction is not written in the way it should be or magnetic field might affect this results. --------------------------------------------------- M. Kossov: --------- Newral Net algorithms for segment finding (dcm_segm) and segments linking (dcm_link) in DC have been described. The frst algorithm searches for clusters of wires at different angular progections (in respect to the "line" of the layer) and organizes a competition of different progections to exclude intermediate "fake" segments. The algorithm is linear, fast and templet independent. The idea is similar to the noise reduction of Dave Heddle but has an opposite gole. Each segment is characterized by a central point and a segment direction. The second algorithm is based on the discovery of "magic circles" which approximate a distribution of the crossing points of the extrapolation lines of segments. The information collected in the "magic circles" helps to reconstruct correct track inspite of the correlated background around the "real" segment (forces crossing on the "magic circles"). ----------------------------------------------------------