GSI Forum: Bugs, Fixes, Releases » Bug in PndPidEmcBayesAssociatorTask?

Home » PANDA » PandaRoot » Bugs, Fixes, Releases » Bug in PndPidEmcBayesAssociatorTask?

Show: Today's Messages :: Polls :: Message Navigator

Bug in PndPidEmcBayesAssociatorTask? [message #13027]

Fri, 03 February 2012 14:15

Andreas Pitka
Messages: 38
Registered: November 2011
Location: Gießen

continuous participant

From: *cb.uni-bonn.de

Hi,

there seems to be some problem in the PndPidEmcBayesAssociatorTask, seems like calculating the log of a not defined variable or log of zero.

===========================================================
#10 0xb69f06c8 in log10 () from /lib/libm.so.6
#11 0xb66bcb3d in TMath::Log10(double) ()
from /data_dirac/apitka/fairsoft/jan12/tools/root/lib/libMathCore.so
#12 0xb0a5310a in PndPidEmcBayesAssociatorTask::GetPdf (this=0x8b58f30,
ppin=6.62114954, thin=65.4592667, phin=-31.6744957, z20in=0.995157063,
z53in=0.0010235901, LATin=0, EPin=0.0368693545, charge=1, proba=0xbfc95868)
at /data_dirac/apitka/Pandaroot/Trunk/trunk/pid/PidClassifier/PndPidEmcBaye sAssociatorTask.cxx:209
#13 0xb0a52dc1 in PndPidEmcBayesAssociatorTask::DoPidMatch (this=0x8b58f30,
pidcand=0x1083ad38, prob=0x1083b528)
at /data_dirac/apitka/Pandaroot/Trunk/trunk/pid/PidClassifier/PndPidEmcBaye sAssociatorTask.cxx:151
#14 0xb0a52c17 in PndPidEmcBayesAssociatorTask::Exec (this=0x8b58f30,
option=0xb37fd4d0 "")
at /data_dirac/apitka/Pandaroot/Trunk/trunk/pid/PidClassifier/PndPidEmcBaye sAssociatorTask.cxx:124
#15 0xb72c01a8 in TTask::ExecuteTasks(char const*) ()
from /data_dirac/apitka/fairsoft/jan12/tools/root/lib/libCore.so.5.32
#16 0xb72bf65e in TTask::ExecuteTask(char const*) ()
from /data_dirac/apitka/fairsoft/jan12/tools/root/lib/libCore.so.5.32
#17 0xb372449b in FairRunAna::Run (this=0x8085df8, Ev_start=0, Ev_end=100)
at /data_dirac/apitka/Pandaroot/Trunk/trunk/base/FairRunAna.cxx:382
===========================================================

Best Regards

Andreas

ps i am using the macros from macro/pid and revision 14319

Report message to a moderator

Re: Bug in PndPidEmcBayesAssociatorTask? [message #13031 is a reply to message #13027]

Fri, 03 February 2012 16:52

Ronald Kunne
Messages: 32
Registered: October 2009

continuous participant

From: *in2p3.fr

The problem seems to be that the Lateral Moment = 0, for the EMC cluster, which is strange.

Andreas Pitka wrote on Fri, 03 February 2012 14:15

LATin=0

I can't change the code, but you can cure it by changing lines 205-209 of trunk/pid/PidClassifier/PndPidEmcBayesAssociatorTask.cxx :

// trafo that spreads out the parameters
   pp  = rangePconst0 + rangePconst1 * TMath::Log10(ppin);
   Z20 = -TMath::Log10(1-z20in);
   Z53 = -TMath::Log10(z53in);
   LAT = -TMath::Log10(LATin);

into

// trafo that spreads out the parameters
   if( ppin<0.000001)  ppin=0.000001;
   if(z20in>0.999999) z20in=0.999999;
   if(z53in<0.000001) z53in=0.000001;
   if(LATin<0.000001) LATin=0.000001;

   pp  = rangePconst0 + rangePconst1 * TMath::Log10(ppin);
   Z20 = -TMath::Log10(1-z20in);
   Z53 = -TMath::Log10(z53in);
   LAT = -TMath::Log10(LATin);

This checks for zeroes in all four logarithms.

Friendly greetings,
Ronald Kunne

Report message to a moderator