[BLAST_ANAWARE] beginning to worry about quality of crunched runs and our listing

From: Aaron Joseph Maschinot (ajmasch@MIT.EDU)
Date: Sat Jul 24 2004 - 16:52:24 EDT


b/c of the recent "no-ADC" data runs, i've been looking through the data
files. i've found some bad things...

1) karen estimated that the no-ADC data runs started at run 9243 and go
   through run 9263. however, runs 9241 and 9242 both have something
   wrong with them, too; the vector polarizations from these runs are
   just wrong. maybe the ADC was "sort of" signaling in these runs.
   anyway, for the present, i think we should not include them in the
   analysis.

2) more scary is the fact that some files appear not to be recording
   charge data. up to just a couple of minutes ago, i thought that root
   would issue a warning if no charge was found. however, run 9232 has
   no charge in it but root doesn't give any warning about this fact. if
   other such "no-charge-but-root-doesn't-mind" files exist, then they
   will clearly have a BIG impact on polarization.

   for example, the vector dilution of runs 9230 9231 9233 9234 is
   "normal" (around 0.57 or so). however, if i include run 9232 in the
   listing (as it is listed in pol.log), the dilutions become -0.137 and
   0.062!!! admittedly, i haven't been looking at each data file,
   one-on-one, in order to see if there is anything wrong with it (i.e.
   no charge, no ADC, etc.); it's kind of hard to do that with 200-300
   files. but, clearly, even one such bad file can mess up three others
   drastically.

3) runs 9098, 9099, 9101, and 9102, though listed as "good" files, all
   have no ADCs. the log-book says that the right-sector WC's didn't have
   any HV around this time. maybe this somehow affected the runs.
   however, these files' charges have been being taken into account for
   total charge to normalize by.

i'm going to update pol.log. however, it seems to me that a file-by-file
check is necessary. maybe a scanning short scanning program could be
written that will, for each file, make sure that it has charge, TDCs,
ADCs, etc. it seems that, do to the large number of incoming files, it
would seem appropriate for the shift taker to run such a file (else
everyone else who does analysis will have to individually check ALL the
files).

aaron



This archive was generated by hypermail 2.1.2 : Mon Feb 24 2014 - 14:07:31 EST