[G12] A Kinematic Fitter issue SOLVED

[Michael C. Kunkel]

Greetings,

It took me some time, but I have solved a fitter issue. (There might be 
more)
First let me take you down the procedure. When I asked for everyone parm 
files, I did some digging and found out that my parms and the g11 parms 
were very similar. I then used the g11 parms on my tuner fitter and got 
good pulls and probabilities. (I will show plots at the meeting). I then 
used the fitter on Dustins page, which is the one I send to people and 
noticed the same discrepancy FSU reported.

I dug into the codes and found one fix so far.
In KinLine.C at line on or about 287 there is a line that reads
p_offset = 2.5*(pPars[part_index][sector-1][bin][2]);

change this to

p_offset = 1.0*(pPars[part_index][sector-1][bin][2]);

I have posted my plots to
https://jlabsvn.jlab.org/svnroot/clas/users/mkunkel/clas/kinematic_fitter/NEW_PLOTS/

Annotations:
Prob_diff_samecode_diffKinline.pdf --->Before fix
Prob_diff_samecode.pdf ---> After Fix
GP_PPip_tunechannel_multifalse_TUNEPULLS.pdf  --->Tune Pulls

I have not and am not planning on testing this with multi = true.


BR
MK