GSI Forum: Tracking » Macros for Mvd+Stt Pattern Recognition

Home » PANDA » PandaRoot » Tracking » Macros for Mvd+Stt Pattern Recognition

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

Macros for Mvd+Stt Pattern Recognition [message #11277]

Mon, 06 December 2010 23:26

Gianluigi Boca
Messages: 177
Registered: March 2004

first-grade participant

From: *gsi.de

hallo,
you can find in the new directory

$VMCWORKDIR/macro/sttmvdtracking

three macros for generating, digitizing and reconstructing events.

In particular in runreco.C you can find the Task for Pattenr Recognition with Mvd + Stt

In short time runreco.C will be expanded to include also the Kalman filter track fitting.

Please try it and report any problems you may encounter

Gianluigi

Report message to a moderator

Re: Macros for Mvd+Stt Pattern Recognition [message #11280 is a reply to message #11277]

Tue, 07 December 2010 15:13

StefanoSpataro
Messages: 2736
Registered: June 2005
Location: Torino

first-grade participant

From: *to.infn.it

Is there some reason to use straws_skewed_blocks_pipe_120cm.geo instead of straws_skewed_blocks_35cm_pipe.geo ? I supposed the latter is the default one.

Report message to a moderator

Re: Macros for Mvd+Stt Pattern Recognition [message #11282 is a reply to message #11280]

Tue, 07 December 2010 16:52

StefanoSpataro
Messages: 2736
Registered: June 2005
Location: Torino

first-grade participant

From: *to.infn.it

Now sttcombi macros in macro/pid folder are updated using STT+MVD tracking code from Gianluigi.
It seems to work, also the kalman and the pid correlation Smile

Report message to a moderator

Re: Macros for Mvd+Stt Pattern Recognition [message #11286 is a reply to message #11280]

Wed, 08 December 2010 14:54

Gianluigi Boca
Messages: 177
Registered: March 2004

first-grade participant

From: *gsi.de

Stefano Spataro wrote on Tue, 07 December 2010 15:13

Is there some reason to use straws_skewed_blocks_pipe_120cm.geo instead of straws_skewed_blocks_35cm_pipe.geo ? I supposed the latter is the default one.

Form the mechanical point of view, the 120 cm geometry is favoured by the Julich people.
From the STT+GEM detector performance point of view it is not still clear, there pro's and con's.

So, what do you suggest Stefano ? Would the usage of the 120 cm geometry be a problem of consistency with result obtained in the past ?

Report message to a moderator

Re: Macros for Mvd+Stt Pattern Recognition [message #11287 is a reply to message #11277]

Wed, 08 December 2010 14:58

Gianluigi Boca
Messages: 177
Registered: March 2004

first-grade participant

From: *gsi.de

Gianluigi Boca wrote on Mon, 06 December 2010 23:26

I uploaded a new veriosn of the same
$VMCWORKDIR/macro/sttmvdtracking/runreco.C
with also the Kalman filter.
Also Lia provided the Macro
$VMCWORKDIR/macro/sttmvdtracking/checkmomentum.C
to plot the reconstructed momentum after the Pattern Recognition but BEFORE the reconstructed momentum after the Kalman filter.

Gianluigi
Kalman filter and the

Report message to a moderator

Re: Macros for Mvd+Stt Pattern Recognition [message #11292 is a reply to message #11277]

Fri, 10 December 2010 09:53

Ralf Kliemt
Messages: 507
Registered: May 2007
Location: GSI, Darmstadt

first-grade participant

From: *cb.uni-bonn.de

Hello.

Trying to use the new PR I get for the Kalman fit status flag always -1.
With Lhe I have a nice mix of 1, -1 & -2.
What is going on?

Ralf.

Report message to a moderator

Re: Macros for Mvd+Stt Pattern Recognition [message #11293 is a reply to message #11292]

Fri, 10 December 2010 10:45

StefanoSpataro
Messages: 2736
Registered: June 2005
Location: Torino

first-grade participant

From: *to.infn.it

Hi,
Tobias has changed in some part of the code the way to access to detector ID in the PndTrackCand.
Before, the detId was taken from the enum in PndDetectorList -> kMVDStripPixel.

With the new changes the detector ID is the branch ID of the corresponding TClonesArray, i.e.:

FairRootManager::Instance()->GetBranchId("MVDStripPixel")

As far as I have checked, the code was changed in lhetrack, pid and recotasks. Probably there is still some unchanged part of the code in Stt and maybe also in other tracking classes.

Report message to a moderator

Re: Macros for Mvd+Stt Pattern Recognition [message #11294 is a reply to message #11293]

Fri, 10 December 2010 11:24

Lia Lavezzi
Messages: 291
Registered: May 2007
Location: Torino

first-grade participant

From: *47-151.net24.it

Hallo Ralf,
please use an older revision for the moment. I think r10451 should be fine. In the meantime we are working to make the stt code compatible with the new way of getting the detID. It shouldn' t take too long.
Cheers,
Lia.

Update: sorry, 10451 is not old enough, it has to be older than 10446. But keep the newest macro/pid macros for stt in order to use the stt + mvd PR!

[Updated on: Fri, 10 December 2010 12:23]

Report message to a moderator

Re: Macros for Mvd+Stt Pattern Recognition [message #11295 is a reply to message #11292]

Fri, 10 December 2010 15:01

Gianluigi Boca
Messages: 177
Registered: March 2004

first-grade participant

From: *gsi.de

Ralf Kliemt wrote on Fri, 10 December 2010 09:53

Hello.

Trying to use the new PR I get for the Kalman fit status flag always -1.
With Lhe I have a nice mix of 1, -1 & -2.
What is going on?

Ralf.

Hallo Ralf,
I have just put the new version of the STT and STT+MVD PR code. All
the hit enumerators have been substituted, so I believe the Macros run now.

Tschuess Gianluigi

Report message to a moderator

Re: Macros for Mvd+Stt Pattern Recognition [message #11310 is a reply to message #11295]

Mon, 13 December 2010 17:00

Ralf Kliemt
Messages: 507
Registered: May 2007
Location: GSI, Darmstadt

first-grade participant

From: *pool.mediaWays.net

Hi Gianluigi,

I windered why the Real PR for STT needs the Monte-Carlo Point array. Diving deeper I found that in PndSttTrackfinderReal::DoFind() (~line 475) you abort the PR for fake hits, because you cannot add a proper reference. Could you remove these restrictions, giving a plain -1 as reference?

Regards.
Ralf

Report message to a moderator

Re: Macros for Mvd+Stt Pattern Recognition [message #11328 is a reply to message #11310]

Wed, 15 December 2010 22:09

Gianluigi Boca
Messages: 177
Registered: March 2004

first-grade participant

From: *gsi.de

Ralf Kliemt wrote on Mon, 13 December 2010 17:00

Hi Ralf,
done ! Now the noise hits or anyway hits not belonging to any MC track are included in the STT Pattern Recognition. The number that usually connects a Hit to a MC track in my code now is set at -20.
Notice though that in the STT at any Hit it corresponds a MC Track since the noise hasn't been implemented yet.
So the PR results should be unchanged.

The new PndSttTrackFinderReal.cxx and PndSttMvdTracking.cxx
are in the repository.

Gianluigi

Report message to a moderator

Re: Macros for Mvd+Stt Pattern Recognition [message #11329 is a reply to message #11328]

Thu, 16 December 2010 10:02

StefanoSpataro
Messages: 2736
Registered: June 2005
Location: Torino

first-grade participant

From: *to.infn.it

Hi Gianluigi,
I have run the full sim chain in macro/pid, simulating 10k events with DPM to check the stability.

Everything is fine up to the stt pattern recognition (PndSttTrackFinderReal).
Once I add PndSttMvdTracking I have the system hanging on a well defined event (7434) for hours and then closing the root.
I have tried two times with the same effect.

This is the last message I got:

Found Tracks: 0 in event no. 7431
----------------
Found Tracks: 0 in event no. 7432
----------------
Found Tracks: 0 in event no. 7433
----------------
Warning: numerical instability (primal simplex, phase II)
Found Tracks: 0 in event no. 7434
----------------
something fishy is going on in PndSttTrkAssociatedParallelHitsToHelixQuater!Range in Fi (rad) is  3.454
something fishy is going on in PndSttTrkAssociatedParallelHitsToHelixQuater!Range in Fi (rad) is  3.454
something fishy is going on in PndSttTrkAssociatedParallelHitsToHelixQuater!Range in Fi (rad) is  3.454
something fishy is going on in PndSttTrkAssociatedParallelHitsToHelixQuater!Range in Fi (rad) is  3.454
something fishy is going on in PndSttTrkAssociatedParallelHitsToHelixQuater!Range in Fi (rad) is  3.454
something fishy is going on in PndSttTrkAssociatedParallelHitsToHelixQuater!Range in Fi (rad) is  3.454
something fishy is going on in PndSttTrkAssociatedParallelHitsToHelixQuater!Range in Fi (rad) is  3.454
something fishy is going on in PndSttTrkAssociatedParallelHitsToHelixQuater!Range in Fi (rad) is  3.454
something fishy is going on in PndSttTrkAssociatedParallelHitsToHelixQuater!Range in Fi (rad) is  3.454
something fishy is going on in PndSttTrkAssociatedParallelHitsToHelixQuater!Range in Fi (rad) is  3.454
something fishy is going on in PndSttTrkAssociatedParallelHitsToHelixQuater!Range in Fi (rad) is  3.454
something fishy is going on in PndSttTrkAssociatedParallelHitsToHelixQuater!Range in Fi (rad) is  3.454
something fishy is going on in PndSttTrkAssociatedParallelHitsToHelixQuater!Range in Fi (rad) is  3.454
something fishy is going on in PndSttTrkAssociatedParallelHitsToHelixQuater!Range in Fi (rad) is  3.454
something fishy is going on in PndSttTrkAssociatedParallelHitsToHelixQuater!Range in Fi (rad) is  3.454

Report message to a moderator

Re: Macros for Mvd+Stt Pattern Recognition [message #11342 is a reply to message #11329]

Fri, 17 December 2010 13:51

Gianluigi Boca
Messages: 177
Registered: March 2004

first-grade participant

From: *gsi.de

Stefano Spataro wrote on Thu, 16 December 2010 10:02

Found Tracks: 0 in event no. 7431
----------------
Found Tracks: 0 in event no. 7432
----------------
Found Tracks: 0 in event no. 7433
----------------
Warning: numerical instability (primal simplex, phase II)
Found Tracks: 0 in event no. 7434
----------------
something fishy is going on in PndSttTrkAssociatedParallelHitsToHelixQuater!Range in Fi (rad) is  3.454
something fishy is going on in PndSttTrkAssociatedParallelHitsToHelixQuater!Range in Fi (rad) is  3.454
something fishy is going on in PndSttTrkAssociatedParallelHitsToHelixQuater!Range in Fi (rad) is  3.454
something fishy is going on in PndSttTrkAssociatedParallelHitsToHelixQuater!Range in Fi (rad) is  3.454
something fishy is going on in PndSttTrkAssociatedParallelHitsToHelixQuater!Range in Fi (rad) is  3.454
something fishy is going on in PndSttTrkAssociatedParallelHitsToHelixQuater!Range in Fi (rad) is  3.454
something fishy is going on in PndSttTrkAssociatedParallelHitsToHelixQuater!Range in Fi (rad) is  3.454
something fishy is going on in PndSttTrkAssociatedParallelHitsToHelixQuater!Range in Fi (rad) is  3.454
something fishy is going on in PndSttTrkAssociatedParallelHitsToHelixQuater!Range in Fi (rad) is  3.454
something fishy is going on in PndSttTrkAssociatedParallelHitsToHelixQuater!Range in Fi (rad) is  3.454
something fishy is going on in PndSttTrkAssociatedParallelHitsToHelixQuater!Range in Fi (rad) is  3.454
something fishy is going on in PndSttTrkAssociatedParallelHitsToHelixQuater!Range in Fi (rad) is  3.454
something fishy is going on in PndSttTrkAssociatedParallelHitsToHelixQuater!Range in Fi (rad) is  3.454
something fishy is going on in PndSttTrkAssociatedParallelHitsToHelixQuater!Range in Fi (rad) is  3.454
something fishy is going on in PndSttTrkAssociatedParallelHitsToHelixQuater!Range in Fi (rad) is  3.454

Hi Stefano,
this is only to let the collaboration know that after we discussed on the phone, and after your attempts last night, it seems clear that the hanging of the program at events 7434 is
caused by a memory leak.

Instead the messages : "something fishy is going on etc.etc."
should never occur in principle.
For that purpose I sent you a modified version of
PndSttTrkAssociatedParallelHitsToHelixQuater
with some printouts added, in order to try understand this problem.
Gianluigi

Report message to a moderator

Previous Topic:	Template for association track from PR to MC track
Next Topic:	alternative to LHE tracking

Goto Forum:

-=] Back to Top [=-

[ Syndicate this forum (XML) ] [

] [

]

Current Time: Thu Nov 28 03:17:43 CET 2024

Total time taken to generate the page: 0.01030 seconds