[Halld-offline] FCAL smearing changes

[David Lawrence]

Hi All,

    I have just committed a change to the repository implementing a 
discussion during the offline meeting yesterday about how FCAL hit data 
is kept. Here's the short version:

1. Changes had been committed back in July to move the random number 
generation used in applying  detector resolution to the individual FCAL 
hits from the reconstruction into the mcsmear program.

2. Among the changes was an duplication of the FcalHit structure in the 
HDDM data model to a parallel FcalTruthHit structure. This allowed 
mcsmear to place both the smeared and unsmeared(truth) values in the 
same output file.

3. The original implementation allowed hdgeant to keep filling only the 
FcalHit structures, leaving the FcalTruthHit structures empty. The 
mcsmear program would then copy the FcaHit data into FcalTruthHit and 
smear the FcalHit.

It became clear pretty quickly during the discussion that this was a 
poor design choice as it left too many easy opportunities for operator 
error. Hence, the changes.

Now, hdgeant fills only the FcalTruthHit structures and leaves FcalHit 
empty for mcsmear to fill later. The main functional change for end 
users is that if they want to do reconstruction with the unsmeared data 
they have to set the DEFTAG:DFCALHit configuration parameter to TRUTH 
either by passing -PDEFTAG:DFCALHit=TRUTH to the reconstruction program 
on the command line, or adding a line:

DEFTAG:DFCALHit  TRUTH

to their jana configuration file.

A quick scan of a few events with hd_dump after implementing these 
changes indicated it was working as expected (the code changes were 
relatively minor). If anyone has any trouble with FCAL reconstruction 
after updating these changes or has any questions, please let me know.

-David

-------- Original Message --------
Subject: 	r5540 - in trunk/src/programs/Simulation: HDGeant mcsmear
Date: 	Thu, 24 Sep 2009 14:23:49 -0400
From: 	Hall-D.SVN.Repository at jlab.org
To: 	davidl at jlab.org, brash at pcs.cnu.edu, wolin at jlab.org, 
zisis at uregina.ca, mashephe at indiana.edu, remitche at indiana.edu, 
zihlmann at jlab.org, somov at jlab.org, staylor at jlab.org



Author: davidl
Date: 2009-09-24 14:23:46 -0400 (Thu, 24 Sep 2009)
New Revision: 5540

Modified:
   trunk/src/programs/Simulation/HDGeant/hitFCal.c
   trunk/src/programs/Simulation/mcsmear/smear.cc
Log:
Modify how FCAL smearing data is transported from hdgeant through the reconstruction.

hdgeant now fills the FcalTruthHit structures (instead of the FcalHit structures). The mcsmear program copies the values from FcalTruthHits to FcalHits simultaneously smearing the energy and time of indivdual hits and applying a threshold cut on the energy.

This corrects a poor design choice made earlier in which the truth information was carried from hdgeant in FcalHits and then mcsmear copied it into FcalTruthHits by mcsmear. Now, the truth info always exists only in the branch with "truth" in the name.





-- 

------------------------------------------------------------------------
 David Lawrence Ph.D.
 Staff Scientist                 Office: (757)269-5567   [[[  [   [ [       
 Jefferson Lab                   Pager:  (757)584-5567   [  [ [ [ [ [   
 http://www.jlab.org/~davidl     davidl at jlab.org         [[[  [[ [[ [[[
------------------------------------------------------------------------