| Updates in PndTrack - PndTrackCand [message #11095] |
Wed, 13 October 2010 19:41  |
StefanoSpataro
Messages: 2736
Registered: June 2005
Location: Torino
|
first-grade participant |
From: *180-80-r.retail.telecomitalia.it
|
|
Dear all,
after the EVO session of this morning, we have decided to apply the following scheme to the tracking code:
- PndTrackCand will store ONLY hit informations (no momentum anymore)
- PndTrack will store only the track informations (i.e. track parameters at the first and last hit), and an index of the corresponding PndTrackCand
- each tracking algorithm will register two separated TClonesArray, one for PndTrack objects, the other for PndTrackCand
- the link between PndTrack and PndTrackCand will be done using FairLink code
We have decided to apply the changes in three steps:
- Each tracking algorithm will write both PndTrack and PndTrackCand TClonesArray
- The momentum information will be removed from PndTrackCand
- The PndTrackCand object will be removed from PndTrack, leaving only the TCA index
We will proceed step by step, so that we will always be able to use track informations for analysis and tests.
Let's start with step 1.
The people involved should be:
MVD: Tobias/Ralf
STT: Gianluigi/Lia
TPC: Felix/Sebastian(?)/Christian(?)
GEM: Radek
LHE: Stefano
STT+MVD: Gianluigi
Global: Radex
I hope I have forgotten nobody.
When the first step change (i.e. writing two different TClonesArray made of PndTrack and PndTrackCand) will be applied, we cpuld move to the second step.
I would like to ask to developers to write in this forum the "OK" once the change will be done in the trunk, so that we could go on. It would be nice to have step 1 ready before next evo meeting (3rd november).
|
|
|
|
|
|
|
|
|
|
| Re: Updates in PndTrack - PndTrackCand [message #11240 is a reply to message #11239] |
Sun, 28 November 2010 14:08  |
StefanoSpataro
Messages: 2736
Registered: June 2005
Location: Torino
|
first-grade participant |
From: *4-87-r.retail.telecomitalia.it
|
|
Hi Sebastian,
| Sebastian Neubert wrote on Sun, 28 November 2010 13:23 |
There are still seed values for the momentum and position in the PndTrackCand ... right?
|
At present yes,
Christian has added them on 20/07/2009 regardless the original design, but I don't know for which reason.
They will be removed in a short time, because useless for our design.
| Quote: |
The PndTracks are filled with content in a common TrackFitting task... right?
|
Not exactly. PndTracks are filled by several track finding tasks which should also give prefit parameters (i.e. from helix assumption).
| Quote: |
Is there a an agreed Track Merging strategy, yet? In that case I suppose there should be a mechanism to join several PndTrackCands. As far as I can tell this is not there ... right?
|
There is a track merging strategy, but at present there is no code doing it for all the detectors because each track finding/fitting is giving different track/candidate objects, even if the object design was fixed three years ago. This is what we are trying to solve.
| Quote: |
Anyhow. My current picture is this: The pattern reco (for each subsystem) will spit out a PndTrackCand array plus an PndTrack array. The Tracks will be linked too the cands but they will not contain fitted parameters yet.
|
This I have not understood.
Your finding must give also the prefit values, and put these values in the PndTrack (if not you cannot extrapolate and merge to different detectors)
Hoping it helps a bit...
|
|
|
|
|
|
|
|
| Removing Seed from PndTrackCand [message #11290 is a reply to message #11095] |
Thu, 09 December 2010 10:33 |
StefanoSpataro
Messages: 2736
Registered: June 2005
Location: Torino
|
first-grade participant |
From: *to.infn.it
|
|
Dear all,
considering that almost everybody has updated the tracking classes to store a PndtrackCand and also a PndTrack TClonesArrays, I would move to the second step -> removing the seeds from PndTrackCand.
This means deleting the following data members:
TVector3 fPosSeed;
TVector3 fDirSeed;
double fQoverPseed;
and the correlated member functions.
I would like to ask to tracking developers to apply this modification in their code, considering that the seed should be taken from the PndTrack object.
If not, I can just delete all the calls to SetSeed and GetSeed, but it is important that the code knows where to take those values properly.
Waiting for feedback, this is the responsible list:
MVD: Tobias/Ralf
STT: Gianluigi/Lia
TPC: Felix/Sebastian
GEM: Radek
LHE: Stefano
STT+MVD: Gianluigi
Global: Radek
LHE is not using seeds as originally planned, therefore it is ready for 3rd step.
[Updated on: Thu, 09 December 2010 10:36] Report message to a moderator |
|
|
|
GSI Forum
Search
Help
Members
Login
Home
