From: Giuseppe Battistoni (Giuseppe.Battistoni@mi.infn.it)
Date: Mon Sep 12 2005 - 17:50:33 CEST
Hoping to help people to clarify the "lattice" capabilities,
I prepared an example starting exactly from the case that you
were mentioning: a Pb-fiber calorimeter.
I made the exercise to simulate a small portion of the
e.m. calorimeter of the Kloe experiment in frascati.
1) See kloe.geo:
I used as basic cell a layer 1.335 cm wide, 0.12 cm high
and 1 m long, where 10 fibers (0.050 cm of radius) are inserted.
Fibers are contained in cylindrical grooves having radius 0.055 cm.
2) This basic cell has been repeated just 10 times, translating
in the z direction so to reproduce 10 layes of the calorimeter.
Every each other layer the cell was shifted in X by a step equivalent to a
groove radius, so to reproduce the staggering of fibers of the actual
calorimeter. These shifts are implemented in the routine "lattic.f"
See the lattic.f here in attachment.
3) the resulting geometry is shown kloegeo.ps
4) Scoring: I used the USRBIN scoring to define a cartesian mesh
that in the x-y plane has a size smaller than a single fiber:
0.032 x 0.032 cm
5) the energy deposition resulting from an e.m. shower of 100 MeV is
visible in the attached shower.ps
You can see that from the point of view of results, replicated
geoemtries behave exactly as if you had explicitly (and tediously)
written the whole geometry. In other words, energy as the geometry
is full: if for instance you are using the actual physical coordinates to
define a volume, you can forget the fact that you built the geometry
using a lattice.
For a region binning you should instead make use of
region number and cell number. In my opinion, in these lattice case it is
instead better to make use of real coordinates and directions.
On Fri, 9 Sep 2005, Anna Ferrari wrote:
> Date: Fri, 9 Sep 2005 18:34:12 +0200 (MET DST)
> From: Anna Ferrari <Anna.Ferrari@lnf.infn.it>
> To: firstname.lastname@example.org
> Subject: some general questions about lattice routine
> Hi All!
> I have some general questions about the way to build repeated
> Suppose to simulate a calorimeter consisting of
> Pb/scintillating fibers.
> In a first moment I thought the best way was to build a basic cell,
> consisting roughly of a fiber surrounded by Pb, and to replicate this
> cell by using the lattice routine.
> Technical details/difficulties apart, the following sentence in
> the manual made me thinking:
> "...a region contained in the prototype cell and all those "mirrored"
> to it inside lattice cells are treated by the program as if they
> were connected with "non-overlapping ORs" into a single region.
> Therefore, any region-based scoring (options SCORE, USRTRACK,etc.)
> can only provide quantities averaged over the whole structure."
> If I well understand, under the point of view of the scored quantities
> all the cells that are built from a single prototype are viewed into a
> single region, even if with the card LATTICE I assign different region
> numbers... Is it right?
> Then (also considering the fact that using lattice is more time
> consuming) I thought to the possibility to use the lattice routine to
> build 'sub detectors' inside the calorimeter, corresponding
> for example to the 'granularity' that is requested for
> the needed informations.
> The subdetectors would be identical in the structure but
> explicitly defined in different regions, with the help of the lattice
> routine. Is this a reasonable compromise? In other words, does it make
> sense or there is some important point that I don't see?
> Thanks in advance for the suggestions....
> Anna Ferrari
This archive was generated by hypermail 2.1.6 : Mon Sep 12 2005 - 18:05:27 CEST