GSI Forum
GSI Helmholtzzentrum für Schwerionenforschung

Home » PANDA » PandaRoot » Tracking » Geane interface in PandaRoot
Geane interface in PandaRoot [message #4516] Wed, 20 June 2007 11:49 Go to next message
Mohammad Al-Turany is currently offline  Mohammad Al-Turany
Messages: 518
Registered: April 2004
Location: GSI, Germany
first-grade participant
From: *gsi.de
Hi,

in SVN 932 two new directories are added

1. trackbase: Helix and parabola track parameter classes

2. geane : Interface to FORTRAN geane package

For now only propagation to PLANE and Volume are implemented, for propagate to volume the helix rep. is used and propagate to plane uses the parabola (due to limitations in Geane original code a work around is in preparation!)

After discussing with Andrea Fontana we agreed not to put in SVN the example I have for geane (Muon chamber of CBM), Andrea will deliver within the next few days an example using the Panda detector. Please be aware that this release is still very preliminary and has some errors which we hope to correct in the near future


regards

Mohammad

Re: Geane interface in PandaRoot [message #4527 is a reply to message #4516] Thu, 21 June 2007 10:17 Go to previous messageGo to next message
Anonymous Poster From: *188-82-b.business.telecomitalia.it
Hi,

I am in the summer school in Varenna right now, so I wont have a chance to test this before the Dubna meeting. It is great to see progress being made and I am looking forward to see the code in action.

CU, Christian
Re: Geane interface in PandaRoot [message #4530 is a reply to message #4516] Thu, 21 June 2007 17:22 Go to previous messageGo to next message
Sebastian Neubert is currently offline  Sebastian Neubert
Messages: 282
Registered: March 2006
Location: Munich
first-grade participant

From: *e18.physik.tu-muenchen.de
Hi!

I am currently working on the implementation of the geane stuff into genfit.
Question: In the helix parameters it says (x,y,lambda,phi,q/p) but in the whole geane-papers x_sc is along the track, so it should always be 0. Is there a confusion here? Or do I misunderstand something?

Moreover which parameters are the covariances refering to and how are they organized, being represented as a double[15]?

In order to be able to write a decent constrcutor GEANETrackRep(const TrackParH& par) the accessors in the TrackPar-class should be declared const.

Cheers! Sebastian.



Sebastian Neubert
Technische Universität München
Department Physik E18
sneubert@e18.physik.tu-muenchen.de
tel: +49-8928912592
Re: Geane interface in PandaRoot [message #4532 is a reply to message #4530] Thu, 21 June 2007 21:35 Go to previous messageGo to next message
Andrea Fontana is currently offline  Andrea Fontana
Messages: 32
Registered: May 2007
continuous participant
From: *cern.ch
Dear All,
we have received the interface from Mohammad yesterday and
are now working to integrate it with the new features that
we developed with the old interface. So for now you will only
find the standard Geane in svn, not yet, for instance, the propagation to the point of closest approach.
So, please, be aware of this!

I will update the files as soon as we are happy with the results
that we now have with the old interface: only at this point
I would expect meaningful results from the integration in genfit.

Moreover I will also prepare an example in the form of tutorial
to show how to use extrapolate to closest in case of STT.

Coming to the questions from Sebastian:

- it is correct that x_sc in the SC system is zero, but also
y_sc and z_sc are zero (see eq. 38 of our report and the whole
section 5). What matters is that in the SC system the
variations different from zero are: delta_y_perp, delta_z_perp
and delta_l_perp (l or s is the track lenght). In geane x
is always along the momentum, but recently also Wittek
has changed convention(see NIMA 566(2006)687):
this is however only a convention without any consequence since
only errors are treated. In the interface we will always
maintain the old convention (along x).
Looking at the class CbmTrackParH, there is some redundancy
as a few variables are useless: we plan to clean it up.

- the covariance matrix is a symmetric triangular 5x5 matrix,
so it only has 15 independent components: the correspondance
is again on our report, page 35. I send you in attachment
2 functions written by Alberto for the conversion.

- I agree on the const!

As we decided in Ferrara, we now are working to release as
soon as possible the new interface to the Collaboration.

Best regards to all,
Andrea

Re: Geane interface in PandaRoot [message #4536 is a reply to message #4532] Fri, 22 June 2007 13:36 Go to previous messageGo to next message
Sebastian Neubert is currently offline  Sebastian Neubert
Messages: 282
Registered: March 2006
Location: Munich
first-grade participant

From: *e18.physik.tu-muenchen.de
Hi Andrea!

Thanks for your reply.
I would suggest, that we use the ROOT-Matrices for the covariances, to take advantage of the nice features of these.
The paper by Wittek and Strandlie explicetely states that tracking becomes unstable when performed with single (=32bit) precision. So I do not understand why in the interface the covs are declared as double32_t.
Finally there is the issue that we have discussed under the term of "virtual detector planes". The corresponding representation would be in the SD-system. Can you help me to find this in the interface, please!?

I will wait until you have cleaned up and checked in the new features befor proceeding in genfit.

Have a nice weekend!
Cheers! Sebastian.


Sebastian Neubert
Technische Universität München
Department Physik E18
sneubert@e18.physik.tu-muenchen.de
tel: +49-8928912592
Re: Geane interface in PandaRoot [message #4538 is a reply to message #4536] Fri, 22 June 2007 18:58 Go to previous messageGo to next message
Mohammad Al-Turany is currently offline  Mohammad Al-Turany
Messages: 518
Registered: April 2004
Location: GSI, Germany
first-grade participant
From: *dip.t-dialin.net
Hi,

Double32_t means 8 bytes in memory, written as a 4 bytes float in file, so this will not have any effect on the stability of the calculations!! and speaking about stability all calculations are done in the FORTRAN part which use float! Any way the Double_32 was only in the CbmTrackParP.h I simply forget to remove it there. In SVN 981 now there is no Double32_t any more, hope that this will make the code more stable!
Re: Geane interface in PandaRoot [message #4542 is a reply to message #4516] Sat, 23 June 2007 10:34 Go to previous messageGo to next message
Anonymous Poster From: *188-82-b.business.telecomitalia.it
Hi everybody,

I think is important to note in the discussion about the float precission, that what Wittek published is not directly related to GEANE! He might be the original author of some routines which are used in GEANE, but any developement of his stuff (e.g. using double precision now) may very well not be included in the FORTRAN routines of GEANE.
I think in the moment this is probably not so important, but it will have to adressed at some time.

Regards, Christian
Re: Geane interface in PandaRoot [message #4545 is a reply to message #4516] Mon, 25 June 2007 09:15 Go to previous message
Andrea Fontana is currently offline  Andrea Fontana
Messages: 32
Registered: May 2007
continuous participant
From: *PV.INFN.IT
Hello Sebastian,
thank you for waiting, I hope it will not take long: this week
Alberto, Pablo and myself will work almost full time on this and
I made some progress in the weekend already! I will keep you
up to date when there are news.

Regarding the virtual detector planes there was a long and very
useful discussion among Mohammad, Christian and myself two VRVS
meetings ago (I seem to remember you were not present that day):
anyway, as I understand it, to use virtual plane with the
extrapolate to the point of closest approach everything can be
done in the SC. In fact we extrapolate to a given length (and
this is in the SC, i.e. with errors projected in the plane
orthogonal to the track) and the virtual plane is by definition
orthogonal to the track. So the conclusion to this was that
the first interpretation (change from SC to SD) as written in
your note is perhaps not necessary. I have to try it to become
sure about this, but I think it might be correct: to be safe
we have also recoded in double precision the Tr.... routines
in case we need them one day.

Concerning the precision issue, I agree with Christian: this
might be a problem, but this depends on the material budget and
it is difficult to predict it now. I am in touch with some
people in CMS that are studying this effect with the native
GEANE and they will let me know if there are problems.
But we can pospone this if it will become an issue...

Talk to you soon...

Ciao,
Andrea

Previous Topic: Monte Carlo Access
Next Topic: genfit & recotsks library loading order
Goto Forum:
  


Current Time: Mon Nov 25 11:12:42 CET 2024

Total time taken to generate the page: 0.00798 seconds