Question by A. Morsch:
How you use the TGeant3 Geometry?

Answ: We need the GEANT3 geometry during the geometrical propagation of
the extracted TPC tracks in order to know if, at a given step, the
particle reach the TOF detector and, if yes, the identifier of current
volume (e.g. if it is a TOF sensitive volume or not et cetera).
This is the reason why the TOF reconstruction needs to create an instance
of the ALICE geometry (gAlice->Init() ). Considering also that geometry
has to be known only for TOF detector, usually, in the Config.C, only the frame
and the TOF detector are switched on.

Question by Ivana Hrivnacova:
Concerning this point, what are the GEANT functions are you using?

Answ: The GEANT function which we are using are:
Gmedia
Gcvolu
gcvolu->names[i]
gcvolu->glx[2]

Ivana:It would be usefull to put this method on a more basic level (e.g.
in AliMC).
Answ: Yes, of course, I will send to You the code with some comment
explaining exactly our needs.

 Question by D. Di Bari:
Concerning the probabilistic approach to particle identification, the
contour cuts you are using for PID will be replaced by the amplitude
method?

Answ: Yes, of course, it is only the present status of the code. We will
move to probabilistic phylosophi for PID asap.

Question by A. Dainese:
Fabrizio, this slide is sayng that you can use different contour cuts for
PID.
Answ: Yes, in fact, I said that the AliTOFPID class has only (as data
member) the pointer to external file containing cuts.
A. Dainese: could you provide the std contour cuts in CVS?
Answ: Yes, of course, we will commit asap the std contour cuts used for
Pb-Pb and pp PID.
A. Dainese: But in any case, the AliTOFPID class uses the output of the
reconstruction.
Answ: Yes Andrea, we can provide to You the reconstructed data. In any
case the code for the TOF reconstruction will be committed in few days.

Remark by K.Safarik:
Why You do not use Kalman tracking for the matching, the method for the
Kalman propagation till external detectors is already available?

Answ: This is a good new. If You provide this method, we can move to
Kalman quickly. Please, can we meet?
 
Remark by K.Safarik:
Concerning the event time zero determination, there is not longer need to
use a combinatorial algorithm (which is bad), you can use the time zero
calculated from all identified pions -> In this way, due to the high
statistics, you can get in principle a resolution

(time resolution/sqrt(N_pions))

Answ: Yes, but I'd like to verify this formula. The current combinatorial
algorithm do not use PID (because it can be used also for PID). Please
remember that if You take all identified pions, You get also
identified pions -but matched with a wrong time, due to noise- and this
can to make worse Your resolution for the event time zero determination.
In any case I will try and I will let You know.

Question by P. Hristov:
Could You provide a set of test macros for TOF?

Answ: Yes Peter, I promise ;)