[Clas_offline] Bug in CLAS-6 MC ST reconstruction (C++ code only)

[Paul Mattione]

I'm glad this works for you, but it seems odd.  My understanding is that gsim takes some shortcuts on simulating the full ADC & TDC response of the detector, and so needs special calibration constants that take these into account (I don't know the details on this).  

Also, what run number are you using for the geometry?  Did you copy all of the geometry calibration constants into run 10 as well?  There may be something I'm missing, but I'm not convinced that this will fully work for normal configurations of the software.  

The software was intended to be used such that the MC head bank has the run number you wish to use for the geometry, and (when it's working) the reconstruction automatically grabs the other (non-geometry) calibration constants from runs between 1 and 10 instead.  If someone could get this working for the new start counter code, I would greatly appreciate it.  

Thanks.  

 - Paul

On Apr 11, 2014, at 4:01 PM, Michael C. Kunkel wrote:

> Greetings,
> 
> I am not sure if what I am going to say applies to your email, but for my simulations, I copy the values of the ST, DC and SC from the calibration run into run 10 and then solely simulate with run 10. I have never had incorrect timing with this method (+- 0.05 ns difference between MC and Data). The reason I do not copy all the calibration run into run 10 is because GSIM needs the values for the CC and EC to be of the original run 10.
> 
> I hope this helps.
> 
> BR
> MK
> 
> 
> On 4/11/14 3:51 PM, Paul Mattione wrote:
>> I've found a bug in the C++ version of the new start counter reconstruction code for CLAS-6.  It only effects reconstruction of Monte Carlo data: ST hits are being reconstructed using the calibration constants from the run number in the HEAD bank rather than run number 1 (resulting in hit times that were off by ~700 ns).
>> 
>> I modified my personal reconstruction code to use separate run #'s for the ST geometry and calibration constants, but reconstruction crashes because no calibration constants have ever been defined for runs 1-10 for some of the constants that are only needed by the C++ reconstruction code (e.g. STN_CALIB / adc2edep / reg1limit, needed by SThit::GetEnergyDeposit() in st/stpp_SThit.cc).
>> 
>> Could someone with advanced knowledge of the start counter simulation and reconstruction codes set the missing calibration constants for runs 1-10?
>> 
>> Until this is fixed, I recommend running your Monte Carlo simulations using the Fortran reconstruction code ("set st_type 1;" in your tcl file).
>> 
>>  - Paul
>> 
>> FOR DEVELOPERS:
>> 
>> My code modifications to use separate run #'s for the ST geometry and calibration constants can be found on ifarm1102 at (note that this is g13-era svn/scons software, so a straight copy may break your build):
>> 
>> Code: /w/hallb/clasg13/pmatt/svnroot_clas6/g13a_pass2v2_64bit/trunk/
>> Environment Variables: /home/pmatt/env_jlab_clas6_mccook_centos62
>> 
>> In this trunk, I made small modifications to the following files:
>> 
>> detector/st/stpp_JCalibration.*
>> detector/st/st_brun.F
>> detector/st/stpp_initialize.cc
>> detector/st/stpp_STposition.*
>> calibration/tag_calib/tagraw.cxx
>> 
>> If you want to test using your own source code, I recommend diff'ing these files with your current source code and only copying the changes that look related to this issue.  Thank you for helping with this.
>> 
>> 
>> _______________________________________________
>> Clas_offline mailing list
>> Clas_offline at jlab.org
>> https://mailman.jlab.org/mailman/listinfo/clas_offline