This is to announce that I have installed an upgraded version of
the monitor (I'll call it version 2.1). Here are the main changes:
-- There is now a 4 ms coincidence cut for the combined burst search,
in order to get rid of inter-SM muons. (This would have gotten
rid of 1 hit in last week's candidate burst, lowering its multiplicity
below threshold).
-- Intermediate event info i.e. all hits that were cut in the time interval,
is now also included in the output text file (and the ntuples).
This info will follow the non-cut hit burst info in LARGE_BURST_*_*.TXT.
The columns labeled "Type" and "Cut" indicate whether the event was
included in the burst search or not:
Roughly,
Type = 0 means good event info
Type = 1 means bad raw data
Type = 2 means bad reconstructed info
A '*' in the Cut column means that the event was included in the burst
search. Hits with no '*' were cut for one reason or another (coincidence
with neighbor, below energy threshold, etc.)
If you're on the alarm mailing list, you should check out some
examples of these files in $1$DIA2:[SCRA.DISKGC] to familiarize
yourself with the new format.
* A lot of the code was restructured to accommodate these changes,
since you need to keep a lot more info in memory to do the inter-SM
coincidence cut and keep the non-cut hit info, but unless you are Alec
you probably don't care about the details. I did have to decrease the
amount of info saved for every hit that is written into the ntuples: raw ADC
and TDC values and some intermediate variables were sacrificed.
Also, I added a few extra goodies:
* Ntuples are now written for combined bursts as well (all hits,
cut or not, are included over an extended time period).
* A kumac file is now written for all bursts, allowing easy PAWing of
the ntuples. It automatically inputs the ntuple file, and prints out
boxes, erptimes, energies and positions of events during the time
interval of the burst (allowing you to look at both cut and non-cut
events). Check out $1$DIA2:[SCRA.DISKGC]LARGE_BURST.KUMAC for an
example.
* I followed Alec's suggestion: running LARGE_BURST.COM
defines a logical, LB$NTUPLE, which points to the most recent ntuple file.
I also fixed various random bugs, such as a 1-day offset in the UT time and
lots of other minor things you don't want to hear about.
Anyway, I have tested this new version on a number of disk files, but
due to the non-trivial restructuring of the code, there may be some
bugs left and I expect it to be a bit less stable than the currently
running version, at least for a while. I'll keep a sharp eye on it.
(Erik, I apologize in advance in case you get woken up!)
Kate.