Weekly Meeting Minutes - May 06, 2005
Meeting Agenda
Attending: E. Clinton, R. Demirchyan, M. Ito, D. McNulty, E. Pasyuk, M. Payen, A. Teymurazyan, M. Wood
By phone: A. Bernstein
Summary of Meeting Discussions:
All. Nominate secretary: Dustin McNulty
All. Approval of last week's minutes:
Unanimously Approved. You can find the previous week's minutes here.
All. New Business: Itemized below
1: Mike's Vito analysis Status:
Mike has written a summary of veto-related analysis studies done to date; this can be found here. Below are a list of comments and discussions about the veto analysis:
Need to add time walk correction (not a big effect--from Dan Pomeroy's analysis)
Should calibrate ADC scale to MeV (using minimum ionizing peak...) so can look at energy spectrum...
There appears to be signs of a time-walk shift in the veto paddles around beam hole; this was seen in the snake scan example of reconstructed veto position shown in above link.
Mike showed plots of accidental conversion (or probability of false veto) which exhibited the expected 1% average after a series of cuts, although there was some drop-off at the edges. Mike says this is due to his y-cut; plots found here exhibit a non-zero-sloped behavior between the "Veto_y minus Hycal_y fit_center" and Hycal_y. This could indicate that the speed of light constant in the scintillator is not properly calibrated, furthermore the notched paddles (around beamhole) have different slopes than non-notched paddles--thus there are counter dependent velocity constants due to geometry effects on reflection.
Aron suggested that Mike work with Yelena on determining the number of electrons to photons during Compton runs.
2:Dustin shows timing-related histograms from pi0 analysis. These plots can be found here
The first plot shows the number of fully reconstructed tagged photons found per event for a typical run (run05058--Carbon production)
The second plot shows the distribution of tagged photons (associated with the pi0 pair) for which there were 2, 1, or zero time matchings with the clusters in the pair. The relative ratios of 2, 1, and 0 matches are: 0.75, 0.21, 0.04. Note that this uses a time matching of +/-30ns between clusters and+/-30ns between cluster(s) and tagger.
The third plot shows the same above distributions for all Carbon runs (for Dustin's analysis this is 218 runs)--with roughly the same results for the relative ratios as quoted above.
The last plot shows the pi0 pair invariant mass distribution from events for which there were 0 time matches between both clusters and tagged photon. The statistics here are ~88,000 events which is ~4% of total Carbon statistics.
Dustin also presented his main algorithm for applying cuts and filling nTuple in the context of determining the pi0 yield as a function of pi0 production angle from the 86 Lead runs used in his analysis. The slide describing the algorithm and cuts used can be found here on the first page. The remainder of this document shows results from pi0 analysis from 4 different fiducial regions of hycal pairs: both in Glass only, both in Crystal only, mixed--one in each, and all channels. The last page shows the background (sideband-type) subtracted dn/dtheta for Lead target.
3: The final topic discussed concerned the "Timing Anomalies" that are showing up in our analysis.
Mark presented the diagnosis of the latest anomaly plaguing the epics_events table in the primex book_keeping database (depicted in this plot), and it turns out that this problem occurs anytime the raw data is analyzed on a file-by-file bases as opposed to analyzing at once all the files associated with a particular run as was demonstrated by this plot.
The following will summarize the causes and solutions of the 3 Time-Anomalies presently in our data set:
1: time in EVENTID wrong at beginning of each file (with the exception of the first file).
cause: Code in trigger package thinks the beginning of each job (file) is the beginning of the run.
sol'n: Use "time" in TRIGGER bank, which is not a reconstructed "time" as with time in EVENTID bank.
2: timing glitch in live1, and live2 of TRIGGER bank
cause: Sometimes carry bit comes early on bits 16 and 24 of TriggerSupervisor.
sol'n: Use the event context to flag the anomaly--meaning easily take note when a sudden large jump back and/or ahead in time occurs and correct for it.
3: More subtle timing glitch in live1 of TRIGGER bank
cause: Sometimes there is a double carry-out on bit 8 in TriggerSupervisor.
sol'n: Mark says maybe too small to worry about, otherwise use event context.
All. Discussions for next time:
Eric will present some information about the bit-8 timing anomaly.
Dustin will present more timing histograms related in pi0 analysis.