Dear Claus,
Julia Thompson of Pittsburgh used
the following optical algorithm to find the
center of cerenkov rings. In her day she was
looking for particles of fixed momenta and so
she could use a dispersive lens rather than a
computer.
All points on a ring are equal distance from the
center. So if you have a track of momentum P you
want to look for rings of radii Rpi(P) or Rk(P)
with centers near the expected intercept with
the detector. So to look for pions take all pixels
within a reasonable area of the intercept and
look for hits. When you find a hit "disperse" the
signal from that hit to all pixels on a circle
of radiu Rpi(P) from that one. (Of course on the
computer you do this with an array that holds
the dispersed signal). Repeat this for all N hits
on the grid. If the particle was a pion then
there will be a pixel where N rings will overlap.
This will show up as a maximium in the array.
However if the particle was a kaon the N rings
will never overlap unless you repeat the search
using the radius Rk(P).
This method has the advantage that it give the
location of the center of the ring independently
of the track.
Of course you can also use the magnitude of
the signal.
Nphoton ~= K*(1-(Pthreshold/P)^2)
For kaons (Pthreshold/P)^2 = (mK/mpi)^2 = 12.5
times larger than for pions.
Yours Michael
Quoting "Claus O. E. Jorgensen" <ekman@nbi.dk>:
>
> Dear Collaborators,
>
> Thanks to Pawels new Bfs tracking and Djamels calculation
> of reflections on the Rich mirror I've been able to do the
> first rough pid with the Rich.
>
> You can find plots on the web:
>
> http://www.nbi.dk/~ekman/rich/RICH.html
>
> I can think of (at least) 4 things that can improve the RICH pid:
>
> 1) Align the geometry of the detector. For now I just use an
> old geo file. I don't know how to tune the geometry. The
> calculated ring centers (reflection of the track lines)
> seems to hit in the center of the rings, but since I find
> the rings from the center it's difficult to check the
> correlation.
>
> 2) Implement the correct geometry of the image plane. The tube
> positions is given by BrChkvParameters, and they are only
> approximate - there are some offsets missing - who knows the
> exact geometry?
>
> 3) Make gain calibrations. Maybe we can find a single photon
> peak in the raw data (or in events with only one track),
> and calibrate to that. I'll look at that.
>
> 4) Improve the ring-finding method. For know I just make a
> histogram of signals per dR (R=distance from center) and
> normalize to the area of the dR rings - this histogram
> is then fitted with a gaussian, and in most cases a nice
> peak is found. The ring radius is then set to mean of the
> peak. Maybe we can use some more sophisticated recognition
> routines (any ideas?)
>
> Any comments or suggestions are welcome!
>
> Have a nice weekend,
>
> Claus
>
> +-------------------------------------------------------------+
> | Claus Jørgensen |
> | Cand. Scient. Phone : (+45) 33 32 49 49 |
> | Cell : (+45) 27 28 49 49 |
> | Niels Bohr Institute, Ta-2, Office : (+45) 35 32 53 07 |
> | Blegdamsvej 17, DK-2100, E-mail : ekman@nbi.dk |
> | University of Copenhagen Home : www.nbi.dk/~ekman/ |
> +-------------------------------------------------------------+
>
Michael Murray, Cyclotron TAMU, 979 845 1411 x 273, Fax 1899
This archive was generated by hypermail 2b30 : Fri Sep 07 2001 - 13:06:15 EDT