Re: [Brahms-dev-l] new brat version 2.15.10 & vertex browser in brahms_app/DB_util

Hello.  I have a few comments (or suggestion) about calibration and 
database since I can see that you made scripts, which easily calibrate 
every runs we have.  Try not to submit calibrations for every single run 
to our databases if not necessary (I know it is not necessary for many 
of global detectors.)  It will create more problems (and/or headaches 
for a few people) than it will solve.  For global detectors, every 20 
runs is more than enough (except the runs immediately after some change 
in hardware.)  Also,  (particulary but not necessary limited to  global 
detectors), please do not commit calibration with range. This also tends 
to introduce some screaming in a few people's office.   (For example, if 
you use 3 runs from 10000-10002, just commit it as run 10000.) 

Hiro


Djamel Ouerdane wrote:

>Hi all,
>
>I've just committed some updates to the BB calibration scripts and vertex 
>modules. 
>
>* BB vertex:
>The BB vertex modules deal with all tubes as a 4th method 
>(cleaned up the hack introduced by Hiro or Flemming) and this method is 
>the second best (after the small tube vertex). I also added a vertex 
>calibration browser in brahms_app/DB_util: VtxDbBrowser. Works like the 
>other ones.
>
>* BB calibration:
>Basically, I put an option in the BB calibration module to bypass the 
>fitting routines in the Finish() methods. This option is false by default 
>(= old behaviour). This option is useful if you use the new scripts in
><brat-tree>/scripts/calib/bb:
>
>1st step
>--------
>there are a masterBBCal.sh and updated versions of Bb*.C's To start up a
>data scan, you first have to pick up the *.sh script and put it into your
>analysis directory, let's call it e.g. ~/bbcal
>
>cd ~/bbcal
>
>in ~/bbcal, you must do: 
>
>mkdir adcgain deltatime slewing vtxoffset
>
>cp <brat-install-tree>/share/scripts/calib/bb/BbAdcGain.C adcgain/
>cp <brat-install-tree>/share/scripts/calib/bb/BbDeltaTime.C deltatime/
>cp <brat-install-tree>/share/scripts/calib/bb/BbSlewing.C slewing/
>cp <brat-install-tree>/share/scripts/calib/bb/BbVtxOffset.C vtxoffset/
>(you must pick up the latest version, otherwise, you won't make them work 
>properly)
>
>then, now that you have everything ready:
>
>./masterBBCal.sh -h 
>(for some help)
>
>./masterBBCal.sh -b 10251 -t adcgain
>(for adcgain cal on run 10251)
>
>./masterBBCal.sh -b 9930 -e 9980 -t deltatime
>(for deltatime cal on runs between 9930 and 9980)
>
>The script masteBBCal.sh will generate lsf scripts, launch jobs for each
>sequence of each run from available farm nodes. At the end, the histogram
>sequence files will be merged by an application called mergehistfile that
>I put in brat/application/misc and that should be somewhere in your $PATH
>
>
>then you will have the choice between transferring the final histogram 
>file to another machine or keep it where it is.
>
>2nd step
>-------- 
>you should collect all histogram files you have produced with
>masterBBCal.sh (whatever the calibration type) in one spot. Let's call
>this directory ~/bbcal/hists (but in the future this will be less messy).
>You should keep the names of these files as they were set by the master
>script, don't change the filenames for the following to work.
>
>
>go to ~/bbcal/hists
>cp <brat-install-tree>/share/scripts/calib/bb/bb*.C .
>(I mean bb in small letters)
>
>or alternatively, set your macro path in ~/.rootrc to point to the brat 
>script dir.
>
>Now do
>
>bratroot -l bbCalibration.C
>
>and follow instructions. This is very easy to use and I need adventurers 
>to test the whole thing for bugs and imnprovements. If you make it work 
>without any problem and want to commit something, make sure you havethe 
>permission (your ~/.bratdbrc should be set up correctly, ask an expert - 
>Flemming, Kris, maybe me I guess, if you can be allowed to commit stuff if 
>you cannot already). 
>
>---------------------
>More doc will follow soon with window snapshots to illustrate the 
>different steps. I can tell you that I'm able to process a lot of 
>calibrations in a reasonable short time with full control on fitting, 
>monitoring, etc. The only drawback is that the scripts are using the 
>standard algorithms for peak finding, etc, but that's not so hard to 
>change for experienced developers, the functions in the scripts are easy 
>to debug and the whole stuff is highly "segmented" into small functions.
>
>
>The TOF stuff will follow soon, when I make sure the official gtr script 
>does what it's supposed to do.
>
>As usual, if something goes wrong because of my changes, tell me asap!
>
>Cheers!
>Djam
>
>PS: sorry for the long email :)
>
>  
>


_______________________________________________
Brahms-dev-l mailing list
Brahms-dev-l@lists.bnl.gov
http://lists.bnl.gov/mailman/listinfo/brahms-dev-l