GSIM- New Features

Introduction

Changes, Improvements, New Features:

SAVE - The option to save secondary tracks/vertexes.  Franz Klein.

I defined MCTK (#1) and MCVX (#1) banks that contain secondary
particles/vertices.

It took a while because I couldn't simply copy the ZEBRA banks
into BOS (if you use the electron library, not all input tracks
were copied into ZEBRA banks ...).
The other problem was the inconsistent creation of a KINE subbank
in step_ec.F.

The way how I implemented the code is the following:
if a daughter particle is produced during the tracking,
the corresponding generating process, medium and volume numbers,
and parent track are stored in a KINE subbank.

This info is read out at the end of a event (bos_mcev.F)
and copied into the corresponding BOS banks:
MCTK (#1) and MCVX (#1)     (<-record no.=1)

You can specify which particles you want to save via the
geant lsave list:
FFread card: SAVE  (or interactively: > SAVE ).
To get any secondaries the SAVE card has to be set:

SAVE 'ALL '                   <-save all secondaries
SAVE 'DCAY' 'ALLP'            <-save all secondaries from
                                decay/absorption processes
SAVE 'DCAY' 'HADR'            <-save all hadrons from
                                decay/absorption processes
SAVE 'HADR' 0.01              <-save all hadrons with energy>0.01 GeV
SAVE 'ALL ' 'LEVL' 10         <-save all secondaries up to cascade
                                level 10
SAVE 'ELMG' 'MUON' 'DCAY'     <-save all elmg.parts (photon,e^-,e^+)
                                and muons from decay/abs. processes
etc.

For 'HADR' 'ELMG' 'MUON'   one can specify the min. energy (as above)
default values: Emin_hadr=0.1 GeV  (i.e. all hadrons)
                Emin_muon=0.1 GeV  (i.e. all muons)
                Emin_elmg=0.05 GeV
                max. cascade level=100

The MCTK (#1) bank entries contain information about generating
and ending vertices as well as parent tracks
(if parent_track<0 : track_no from MCTK (#0) i.e. primary track)
The flag entry in MCTK (#1) reflects the process that causes
this track (the number corresponds to the geant NAMEC names).
The flag entry in MCVX (#1) shows the (unique) volume no. so that
one can retrieve those geometry informations.

Good luck

Franz

New Target for E2               Natalya Dashyan <natasha@jlab.org>

I made modifications in some of GSIM routines and added few other routines in order to
include CLAS target cells and physical targets used during E2 running period (there were
4 target cells for 3He and 4He and 3 solid targets used).

Changes had been made in the following files:

gsim//clas_ffky.F
gsim//clas_geom.F
gsim//clmate.inc
gsim//gsimcontrol.inc
gsim//cltmed.inc
gsim//gsimpar.inc
gsim//init_mate.F
gsim//init_tmed.F
 

The following files were added:

gsim//ugeom_12_15.F
gsim//ugeom_15_27.F
gsim//ugeom_8_12.F
gsim//ugeom_stg.F
gsim//ugeom_tg_e2.F
gsim//ugeom_ur.F

There are 2 parameters that control cell/target settings (should be defined in ffread
file):

1. TGE2 - logical variable to turn new stuff on and off, default .FALSE. - GSIM runs
with old target definitions.

2. TGTP - integer area of 3 elements
          TGTP(1) - cell type, 0=No cell, 1=all Al cell (as for E1 and beginning of E2),
2=E2 second cell, 3=E2 third cell, 4=E2 last cell.
          TGTP(2) - liquid target type 0=empty, 1=H2, 2=D2, 3=3He, 4=4He.
          TGTP(3) - solid target type 2=CH2, 12=12C, 56=56Fe, nothing otherwise.

Example of ffread with new target:

TGE2 TRUE
TGTP 4 3

Liquid 3He in the last cell of E2 run period.

TGE2 TRUE
TGTP 1 1

Liquid hydrogen in the aluminum cell (E1 setting).
 

Please let me know if you will fiend anything wrong with new stuff or if you will have
any difficulties to run with.
 
 
 

New BOS Bank GPAR

Franz Klein

A new BOS bank was added to the GSIM output which contains all the FFREAD keys, the user name of the person executing the code, the name of the input file and the unix time. The new bank, GPAR, contains an integer, float and 16 character key name. For encoding details see gsimpar_2_bos.F and gpar.ddl

The basic behavior for the FFREAD flag is changed with the new version of clas_ffgo.F. Now when RNDM is NOT specified, or set to 0 0, the random seed will be derived from the unix time as:
seed1 = unix_time/100000+mod(unix_time,10)
seed2 = mod(unix_time/10,10000)
This ensures that both seeds change from second to second, and will be usefull when farming out long runs.

This is a major revision and rewrite of the clas_step.F routine. It is converted to c, and the internal logic is much improved and new features are enabled.
New features include:
1. All decay products are now always tracked regardless of the NOSEC setting. contributed by Cole Smith.
2. Of the decay products, neutrinos are skipped. contributed by Cole Smith.
3. The NOSEC switch now works properly for MINI and TORU and FOIL.
4. Tracks are now stopped right after the EC, we no longer step through meters of empty space.
5. The track debugging features are much improved (see below Track Debugger).
6. The LAC  and ST stepping function in now properly called.
Decays: In GEANT when a particle decays the resulting new particles are normally treated exactly like the secondary particles from other processes. However, to increase the speed of GSIM we usually turn of the generation of secondaries in inert volumes. As a result, the decay products of a pi0 or Kaon would be lost. The new step routine will first check whether a particle decayed or not. If it did decay it will store all the secondaries, except for neutrinos, for further tracking. If it did not decay, it will check the setting of the NOSEC flags to decide whether secondaries are tracked or not.

Debugging the tracks that GEANT produces poses some problems. Drawing the track on the screen with "swit 1 2" is great, but does not always provide sufficient information. Printing out the track to the console with "swit 1 1" provides screen after screen of step information that is difficult to sort through, and still does not give all the info you may want. Thus a reason for the new track debugger.

This new feature for debugging is controlled with the "swit 3 #" command line switch ( which performs the same action as the ISWIT ffread flag).  Currently two settings are implemented:

1. Every step is printed to the screen, but with different information than swit 1 1
2. A line is printed every time the track leaves a volume.
The second setting is most usefull, allowing you to see instantly how many steps were taken to cross a particular volume and how much energy was lost. Each line contains the following elements:
[Track #, Stack #, Particle Name] Name of Volumes in tree
   (#of steps in this Volume, Tot # of steps for track)
      Kinetic Energy of track,
        List of Physics processes that affected this step.
An example printout of the first 4 lines with swit 3 2 and kine 3 1 1. 1. 30. 30. 0. 0. ( 1 Gev electron at 30 degrees.):
Trk# partname   ]   Volumes                          (#step #step_tot) Kin_e  Processes.
[ 1  0 ELECTRON    ]: CLAS FOIL GAS2                     (   8      8) E: 999.486 MeV NEXT MULS LOSS FIEL
[ 1  0 ELECTRON    ]: CLAS FOIL FOI1 FOI2                (   3     11) E: 999.434 MeV NEXT MULS LOSS FIEL
[ 1  0 ELECTRON    ]: CLAS FOIL FOI1 FOI2 WIND           (   2     13) E: 999.396 MeV NEXT MULS LOSS FIEL
[ 1  0 ELECTRON    ]: CLAS FOIL FOI1 FOI2                (  48     61) E: 999.291 MeV NEXT MULS LOSS FIEL
One can see that 48 steps were taken in the volume FOI2 which is a daughter of FOI1 which is a daughter of FOIL in CLAS. The particle underwent MULS LOSS and FIEL = multiple scattering, energy loss, and bending is a field.
An example of the decay of K0-short kine 3 16 1. 1. 30. 30. 0. 0.:

[Trk# partname   ]   Volumes                          (#step #step_tot) Kin_e  Processes.
GUSER_STEP I: *** Default version of guser_step called ***
[ 1  0 KAON 0 SHORT]: CLAS FOIL GAS2                     (   2      2) E: 619.323 MeV NEXT

Track   1 total # of steps:    2:   0
Secondaries [2] = PION PION

[Trk# partname   ]   Volumes                          (#step #step_tot) Kin_e  Processes.
[ 3  0 PION 0      ]: CLAS FOIL GAS2                     (   2      2) E: 220.271 MeV NEXT

Track   3 total # of steps:    2:   0
Secondaries [2] = GAMMA GAMMA

[Trk# partname   ]   Volumes                          (#step #step_tot) Kin_e  Processes.
[ 5  0 GAMMA       ]: CLAS FOIL GAS2                     (   2      2) E: 126.588 MeV NEXT
[ 5  0 GAMMA       ]: CLAS FOIL FOI1 FOI2                (   2      4) E: 126.588 MeV NEXT
[ 5  0 GAMMA       ]: CLAS FOIL FOI1 FOI2 WIND           (   2      6) E: 126.588 MeV NEXT

This shows that the kaon (with 600 MeV kinetic energy) decays into two pions immediately, and the pion-0 decays into 2 gamma's immediately. The gamma is then stepping through the detector, relatively unaffected by any physics processes.

A new BOS bank, the DOCA bank, stores information from the Drift Chamber (DC) hits, which allows the gsimko code to post-process DC wire information. GSIMKO is now able to add resolution smearing and distance to time conversions.

A new commandline switch which allows GSIM to skip the first # events of the BOS input file.

A new FFREAD card definition was added. The "FAST" card is now recognized in the FFREAD input file. Alternatively you can use the command line switch "-fast" which is equivalent to the FFREAD statement: "FAST 'ALL'" . This switch is intended to allow you to switch between different levels of detail in the code. The "fast" version will include less detail, and therefore be less accurate, but execute faster. So far fast code has been implemented for the SC system (30% speed improvement.)

Coding Using the new "FAST" switch:

To add code that makes use of the FAST switch, you need to include gsimcontrol.inc (FORTRAN) of gsim_c_common.h (C code). For C code the statements would then be:
#include "gsim_c_common.h"
extern GSIMCONTROL_t gsimcontrol_;
if( !gsimcontrol_.fast_code[SC] ){
/* Your SLOW code goes here. */
}else{
/* Your FAST code goes here. */
}
For Fortran:
#include "gsimcontrol.inc"
if(.not.fast_code(SC))then
! Slow code goes here.
else
! Fast code goes here.
endif
Please try to make use of this new feature !
 

The SC code dealing with geometry was completely re-written in C and now makes use of 3-vectors to calculate proper geometry. All 6 sectors are now independent and the geometry definition comes from the BOS bank: sc_geom.bfp with the translations of the forward carriage and clam shells applied for run #1 through a call to make_SCG_banks(1). This ensures that the geometry for GSIM is identical the to geometry for RECSIS/A1.  If compiled with -DUSE_NOMINAL_BANK the file sc_geom.bfp.old_nominal.3.5.98 is used instead and the translations from the MAP are not applied.
The code has two modes: SLOW and FAST. The slow code includes the lead wrapper around the SC paddle and the foam backing and steel support structure, the fast code leaves these features out. A speed gain of 30% is achieved when using the FAST option.

The geometry of the mini torus has been updated by William Dale Conwell from the University of Virginia Physics Dept. More information and pictures of this update are available at UVA: "New Mini Torus Geometry"
Unfortunately this code still needs a MAJOR revision for a code cleanup.


---

I have updated the geometry of the main CLAS Torus to correspond to the outer limits of the vacuum case, according to the ACAD drawings provided by Oxford (as dxf files.) Drawings used are: 0100-1.dxf, 0200-1.dxf,0350-1.dxf. This is somewhat larger than the previous torus, which was implemented as the coil windings. The old configuration can still be used by specifying -DUSE_OLD_TORUS when compiling geom_mg.F of the gsim library. The file clas_init.F was only changed to increase the half thickness of the coil from 7. to 7.2 cm.

I have also added the three support rings as a set of four tube shapes (tubs) in the mother volume LSS-. I used this mother volume rather than MG because the support rings would extrude from MG. Drawings used are 0100-1.dxf ,0426-1.dxf and 66100-E002499. The material chosen for the rings is clas_med_iron. The drawings below show cutouts of the 3 rings shown in red.

Main shortcomming: The geometries are still hard coded in a "data" statement instead of read from a BOS bank.

The new Makefile is very close to the old one, with the added feature that all the *.F and *.c files in the gsim_int or gsim_bat directory are included in the compilation, and will be included in the linking before any of the libraries. (Note that they are also included in the libgsim_int.a or libgsim_bat.a library, but the standard make-scheme does not ensure that those will be the routines used when linking.) Caveat Empor: This will most likely FAIL on the Sun's with the new OS, since the (stupid?) linker can't figure out which object to use: the one from the library or the one from the command line. Solution: use Linux.

The big advantage of this new Makefile is that you can now edit only one of the files from the gsim library and recompile gsim_int or gsim_bat without having to checkout and recompile the entire libgsim.a The following example assumes that you have a "packages" directory, we checkout the gsim_int package, and only some of the guser*.F files from the gsim package, next we move those files to gsim_int, edit them, and recompile:

packages>cvs checkout gsim_int
packages>cvs checkout gsim/guser_book.F gsim/guser_hist.F gsim/guser_last.F
packages>mv gsim/guser_book.F gsim/guser_hist.F gsim/guser_last.F gsim_int
packages>cd gsim_int
gsim_int>sed -e 's/Default/Personal/' < guser_book.F > tmp.tmp && mv tmp.tmp guser_book.F
gsim_int>sed -e 's/Default/Personal/' < guser_hist.F > tmp.tmp && mv tmp.tmp guser_hist.F
gsim_int>sed -e 's/Default/Personal/' < guser_last.F > tmp.tmp && mv tmp.tmp guser_last.F
gsim_int>make exe

The resulting gsim_int (which will be in the usual place, most likely ~/bin/SunOS ) will report with "*** Personal version of guser_hist called ***" after the first trigger.

Now you can edit the guser*.F routines (and any other routine you checkout and copy to the gsim_int directory) and modify them as you like.

REMEMBER: Never "commit" the personal modifications you made to guser*.F routines. Never "commit" modified routines from the gsim_int or gsim_bat directories !

The BOS input file format now also supports the PART bank. This new bank is one of the output banks of RECSIS. Thus it is now possible to read a RECSIS output file (once the RECSIS modifications are finished) directly back into GSIM. This completes the loop: GSIM->RECSIS->GSIM which is usefull for testing the consistency of the data analysis.
A small utility "gsim_bos_io" is now available in packages/utilities/bosanal which reads, writes and converts GSIM BOS files. To get it use "cvs checkout packages/utilities/bosanal" and compile with "make gsim_bos_io". To get help on it's use type: "gsim_bos_io -help". This code is also intended as an example event-generator code. See "Example Generator"

See also: http://aiacehp.ge.infn.ir/~biselli

I've added the polarized target field map. I created this map by a Volker Burkert program and I've modified a set of subroutines to use this map. I created two maps for polarized target field: the first is defined in a small region around the target where the target field has an high gradient and it has a finer step (bgrid_ptg.fpk ) the second is defined in an external region and it has the same parameters of the existing torus and mini map. This map is added to the other two in the file bgrid.fpk

New behaviour:

I defined a new value for PTGIFIELD. GSIM read the map if PTGIFIELD=3 I've summarised in the next table all possible magnetic field configuration
 

MAGTYPE	PTG	PTGIFIELD	action
0	F	-	no magnetic field
0	T	1,2	only ptg field analytically evaluated
0	T	3	only ptg field from map
1	F	-	only torus field analytically evaluated in a simple way.
1	T	1,2	torus field analytically evaluated in a simple way +ptg field analytically evaluated
1	T	3	only torus field analytically evaluated in a simple way ptg field from map
2	F	-	only torus field from map
2	T	1,2	torus field from map ptg field analytically evaluated
2	T	3	torus field from map ptg field from map
3	F	-	torus field from map+mini field from map
4	F	-	only mini field from map
3,4	T	-	error mini and ptg are incompatible
-	F	3	error ptg field without ptg geometry

The default name of the ptg map is bgrid_ptg.fpk and the default directory is $CLAS_PARMS . To change the name and the path I defined a new command line flag and a new environment variable command line flag : -bptg or environment variable : GSIM_BGRIDPTG

Modified files:

bgrid.inc : Contains the definitions of the polarized grid parameters

gsimpar.inc : Contains the definitions of the command line flag "bptg", of the environment variable "GSIM_BGRIDPTG" and of the fpack file key "PTG"

init_mg.F : reads , if requested, the polarized target maps

clas_fld.F : evaluates the field in a point using the polarized target maps if requested

ptg_field.F : evaluates the polarized target field in a point by interpolating from the polarized target map like snake.F

init_flag.F : now, if MAGTYPE=0 it's possible to have only the polarized target field

A copy of the polarized target field is in $CLAS_PARMS with the name bgrid_ptgnew.fpk. If you want to use the new features you have to set the name of the map with the environment variable GSIM_BGRIDPTG or with the command line flag -bptg.
Example: -bptg $CLAS_PARMS/bgrid_ptgnew.fpk

For questions or comments: biselli@infnge.ge.infn.it

Two screen shots of gsim_int with background simulation turned on. The picture on the left has no mini-toroid field, the picture on the right does. Note that the electron tracks (red tracks) for low energy electrons are truncated because gsim does not do more than 1000 steps per particle.

I have implemented a new set of subroutines for the gsim package that simulate target related background events. This code was directly derived from Volker Burkert's background routine in the old geant simulation. These routines produce background event distributions that are based on an EGS simulation of the target. These distributions are stored in an HBOOK file and sampled randomly. The code will generate electron, photon (including bremstrahlung) and positron background events.

New behavior:

The kine switch ( ffread key or gsim command-line) now accept new values which are equal to the original kine value +10 ( -10 for negative kine values). This turns on the background event generator. Thus, kine=1 reads events from mcin, kine=11 reads events from mcin, plus generated background events. A new value, kine=4, does nothing so that kine=14 generates only background events.

New ffread keys:

NTARG: Select target type: (defaults to 1)
 

NTARG = 1	Hydrogen gas dens. 0.007g/cm3, length 76mm reads file $CLAS_PARMS/h_76mm_g.hst
NTARG = 2	Hydrogen liquid dens. 0.07g/cm3 , length 76mm reads file $CLAS_PARMS/h_76mm_l.hst
NTARG = 3	NH3 solid dens. 1g/cm3 , lenght 15mm reads file $CLAS_PARMS/nh3_15mm.hst
NTARG = 4	Cu solid dens. 8.9 g/cm3 , lenght 10microm reads file $CLAS_PARMS/cu_10um.hst
NTARG >= 5	User supplies background information, set ZTAR,ATAR,RHOT,TLENG and supply a file bckgr.hbook.

LUMEN - Luminocity, (defaults to 1E-32)
TIMDC - Time that one event is sensitive to background. (defaults to 250E-9 sec)
NINCE - Number of events used to generate background file. (defaults to 100000000)
ZELEM - Z of target. ATOM - A of target. RHOT - density of target (g/cm3)
TLENG - lenght of target (cm)

New files: (in gsim package)

clas_bckgr.F - Contains the routines to calculate background events.
ffky_bckgr.F - Handles the new keys (ffread) for target description.
ffpar_bckgr.inc - Contains common block that controls clas_bckgr routine.

Modified files:

clas_kine.F - Now calls clas_bckgr when the flag ikine (kine command on gsim command line) is set to previous kine value +10 - A new "event generator", kine=4, was defined which does nothing. Thus kine=4 does nothing, kine=14 gives only background events.
clas_ffky.F - Now also calls ffky_bckgr

This page has been viewed [an error occurred while processing this directive] times