I finally finished the full geometry of the SC for gsim, after a long battle
with the BOS banks. BUT I want to issue a warning:
(my excuses for the somewhat disjointed email, but it's late, and I am not
going to be around for a week to answer questions. If you don't want these
"fixes" please don't check them out, you get the old version back with "cvs
checkout -r release-1-16 gsim")
WARNING WARNING WARNING WARNING WARNING WARNING WARNING WARNING WARNING
The updated GSIM package has NOT BEEN FULLY DEBUGGED !!!!!!!!!
WARNING WARNING WARNING WARNING WARNING WARNING WARNING WARNING WARNING
I have tested it on Linux only with 100 events of Will Brooks' 2.4 Gev
sample file.
I am checking this in now because some people would like to see the SC work
properly in gsim, and since most of the long runs planned with GSIM for the
next week are mostly going to be on LINUX.
Besides this warning a known "bug" is that it is not "convenient" to compile
the code, because it relies on changes in packages/include/sc.h and those
have not made it to the DEVELOPMENT distribution. The Makefile for gsim does
not properly use your own copy of this include directory, so instead use:
make -f Makefile2
(This will try to build an executable and fail. Don't worry gsim_int and
gsim_bat will still work.)
A major issue is that the code does not check if the geometry of the SC
paddles is correct in the geometry bank. I checked everything using:
sc_geom.bfp.old_nominal.3.5.98
and no paddles seem to overlap, and all paddles (also those in panels 3 and
4) get hits which are indeed digitized. Code to read this nominal bank is
currently only accessible through a "compile time switch"
-DUSE_NOMINAL_BANK, or you can relink your sc_geom.bfp to point to the
nominal one. At some point soon we should check the new geometry bank and
make sure none of the paddles overlap, which would cause them to become
"invisible" to particles. I checked only a few tracks into panel 4 sector 1
with the full geometry.
New things you'll notice when you start up the new version:
A) The long list of bos banks (you see it in recsis all the time) is now
also to be enjoyed when running GSIM.
B) There will be complaints by GEANT that the rotation matrices are not
orthogonal, and that it "fixes them up". Well, they are orthogonal but GEANT
makes roundoff errors in the FORTRAN. The fixes should be quite small.
C) If you use OPTI, then SC won't be optimized because it has more than 500
volumes. That's just reality: 6 sectors *56 paddles *5 volumes > 500.
Perhpas we should nix the lead and the foam and the steel ? I will in a few
weeks use the code optimization routines to optimize the SC by hand. In the
mean time the code may be slowed down due to this. If you like, compile with
-DNO_EXTRA_STRUCTURE to get rid of foam, lead, steel.
I am welcomming feedback about these modifications (by all means look at the
c code) but won't be around to answer them for another week or so. Updated
web pages will have to wait about 2 weeks.
Enjoy,
Maurik
=================================================================================
For the experts:
Files now obsolete (get rid of them !):
bos_out.F -> gsim_bos.c
init_bos.F -> gsim_bos.c
geom_sc.F -> geom_sc.c
geom_sc_init.F> geom_sc.c
modified: digi_trig.F clas_step.F digi_sc.F sets_sc.F
----------------------------------------------------------------------
Maurik Holtrop web: http://www.physics.unh.edu/Maurik
UNH Physics Department tel: (603) 862-2019 | fax: (603) 862-2998
DeMeritt Hall mailto:maurik.holtrop@unh.edu
Durham, NH 03824 Home: (603) 659-8765
----------------------------------------------------------------------