|
|
|
|
|
|
| Re: PndTrack and LheTrack [message #8928 is a reply to message #8915] |
Mon, 29 June 2009 17:02   |
StefanoSpataro
Messages: 2736
Registered: June 2005
Location: Torino
|
first-grade participant |
From: *to.infn.it
|
|
Dear all,
I have fallen into two considerations, while converting my lhetrack code inside PndTrack scheme.
The object PndTrackCand is stored inside PndTrack. But had we agreed that the PndTrackcand should be stored inside its own TClonesArray, and then PndTrack should point to it storing just an index? Or do I remember badly?
Another thing, I have seen in PndTrack there are no fRefIndex or similar data members, to store the index of some preliminary object. In my case, once I am storing the tracks coming out from genfit, I would like also to store an index refering to the prefit Track with lhe helix fit. Of course, if the PndTrackCand are stored in their TCA, I should just use that index (not existing at the moment). If PndTrackIndex is inside PndTrack, I still need this new integer.
For the moment I will add this index, I would like hearing comments about this point.
Regards
|
|
|
|
| Re: PndTrack and LheTrack [message #8931 is a reply to message #8928] |
Mon, 29 June 2009 18:35 |
StefanoSpataro
Messages: 2736
Registered: June 2005
Location: Torino
|
first-grade participant |
From: *to.infn.it
|
|
Dear all,
after some fights (and I won!), now even genfit tracks using lhetrack are written as PndTrack.
This means that in the file one can find, as TClonesArray:
LheTrack -> lhe prefit (helix)
LheGenTrack -> genfit refitted tracks
Both are TCA made of PndTrack, as agreed.
There is still the PndTrackCand issue to solve, and of the ref index. At the moment no index is stored inside the PndTrack.
Now, some considerations:
It would be also interesting to agree on a "flag" scheme, considering that the PndTrack stores a fFlag, and that we all should use the same values. Any suggestions? From genfit I have seen the flag is always 0, but maybe because this is the initialization value (that I have set in PndTrack).
And I have seen that half the tracks I have have NDF = 0 (even if there is some value of momentum inside). This is another quite interesting point to understand.
Moreover, in PndTrack we are storing the PidHipothesis (I suppose from the genfit fit), then maybe this info zould be propagated from the kalman task to the outgoing Track object, in order to fill also this variable.
And it would be nice if the output of the kalman fitter, exceptions, and reco hits, could be reduced a lot, depending on the Verbose flag. I think there is too much unwanted I/O, and this could also slow down a bit the reconstruction, producing also big log files.
Regards
|
|
|
|
|
|
|
|
|
|
GSI Forum
Search
Help
Members
Login
Home
