MINUTES of the CALCOM Analysis meeting April 11, 1997 ====================================================== Agenda: V. Burkert CED study of tracks in CLAS J. Manak Results of de-MUXing for RECSIS analysis A. Coleman EC track analysis C. Smith Update on EC gain matching and calibration Alan Coleman presented preliminary results of a study that uses the EC inner and EC outer information to measure particle directions. This capability could be of interest for the track matching with the drift chamber as well as for selecting neutral particles coming from the target. More studies are needed to better understand the angle resolution resolution and how well events can be matched using the drift chamber information. The next meeting will be Friday April 18, 1:30pm, room A110. Volker Burkert -------------------------------------------------------------------- The following are individual reports: Volker Burkert: ============== 200 events from run 781 were studied using the ced display. The goal was to find all tracks in CLAS which an ultimate reconstruction code should be able to identify and reconstruct (this is a lot easier done by eye than by software!). Trajectories were defined simply as having sufficiently many hits that would line up in such a way that a fitting routine using a global approach to pattern recognition - in this case my eyes - should identify them as coming from the target. The result was that in 200 events about 50% more tracks were found than RECSIS reconstructed. Speculation is that the pattern recognition in RECSIS/DC is not adequate if the drift chambers have significant local inefficiencies. Currently the pattern recognition in RECSIS is done locally, were each superlayer is treated separately and a mimimum number of hits per superlayer is required. A "global" strategy should be able to find more tracks in cases were the drift chambers are not fully efficient or electronics channels are missing. Obviously, such an approach would have consequences as to the way the left/right ambiguities are resolved, the use of track segments in the analysis, and on the track 3-momentum resolution. Joe Manak: ========= Results of a study were presented which showed improvements in the track yield after the improved demuxing algorithm and a new (correct) hardware map was applied to the raw data. The new demuxing algorithm resulted in a 30% improvement the number of events in with tracks for run 682, with the track yield going up from 10.6% to 13.4% of the total number of events. Additional improvements in the yield were obtained by decreasing the number of hits required to form a track segment in each superlayer from 3 in the innermost layer to 2, and from 4 to 3 segments in the outer layers . With all improvements (demux, correct hardware map, and decreasing the number of hits to make a track segment) the yield for run 682 was 18.4%. A similar study was also performed for run 781 and the yield increased from 15% to 29.5%. Cole Smith: ========== A mistake was corrected in the ECCL.bfp calibration bank. Attenuation lengths for Sectors 5 and 6 had been interchanged. Correction was made 22:20 10-Apr. Analysis of low threshold Zero-B runs continues. As of 21:30 Apr 4, calibration bank ECCA.bfp was updated with new ADC gains derived from fits to light attenuation of MIP pixel events as a function of pixel distance from PMT. Looser pixel cuts were used to increase number of events/pixel. Previous method used mean ADC corrected for attenuation using cosmic-ray test data. Both methods show the same overall trend in tube-to-tube gain fluctuations, except for some of the shorter strips. Fitted gains should be more accurate since attenuation correction is measured in-situ. However, comparison with expected attenuation shows some residual systematic shifts possibly related to looser pixel cuts allowing higher fraction of non-MIP tracks. Statistics presently limit fitted gain uncertainties to 5-10%.