Comparing 6.1.6 100K tests, file by file.
SL3 vs SL5 One seems to have different numbers of events: br042856.0102phys.
SL4 vs SL5 Two files seem to disagree about the number of events--possible crashes? br03b001.0001phys and br042856.0102phys Mostly no differences at the K-test of .999999, and 17 of 24 comparisons turned out to be identical.
SL3 vs SL4 One run seems to disagree about the number of events: br03b001.0001
From the overlap I surmise that the br03b001.0001 crashed in Run4, and the br042856.0102 crashed in Run5. Unfortunately I can't find the log files for the Run5 set.
Update: I reran the br03b001.0001 in SL44 and the files agree much better now: no differences between 4 and 5 and only a bin shift between 3 and 4.
I reran the br042856.0102phys on SL5 and the files agree much better now.
Summary is in LIST 12 of the 22 files show no differences at the .999999 test level. SL4/SL4 and SL5/SL5 comparisons are now all the same at the .999999 level. I see that a couple of times I rated histograms as "Looks OK" that I rated as "StatLow" (low statistics) another time.
Modified 22-March-2010 at 09:59
http://hep.physics.wisc.edu/~jnb/cdfcode/22Mar2010
| Previous notes | Next notes | Main slide directory |