GSI Forum
GSI Helmholtzzentrum für Schwerionenforschung

Home » PANDA » PandaRoot » Tracking » cov[0][0] = 0 in Kalman.cxx
Re: cov[0][0] = 0 in Kalman.cxx [message #6982 is a reply to message #6981] Thu, 19 June 2008 15:03 Go to previous messageGo to previous message
Susanna Costanza is currently offline  Susanna Costanza
Messages: 33
Registered: January 2008
Location: Pavia
continuous participant
From: *pv.infn.it
Thank you for your explanation about the use of exceptions!

I did some tests... What I see is that throwing away all the events in which cov[0][0] is null results in a considerable loss of efficiency.
Instead, if you skip just that particular hit in which the covariance matrix has that problem, GEANE propagates to the next hit and sometimes the track returns to the right path. The resulting Kalman rep is quite good and the efficiency is higher.

Ok, I agree that cov[0][0] should never be 0 and we will take a look in order to understand why there is this strange behaviour for some events...
In the meantime, do you think it's a so bad idea to "convert" the exception into a return, in order to save a bigger number of tracks?

Ciao,
Susanna
 
Read Message
Read Message
Read Message
Read Message
Previous Topic: GEANE backward propagation
Next Topic: New Features in GENFIT
Goto Forum:
  


Current Time: Thu Jul 18 17:20:23 CEST 2024

Total time taken to generate the page: 0.00976 seconds