Discussion of Geant3 Parameters
Peter Gumplinger, 03.05.29

Original forum post is here:
https://twist.phys.ualberta.ca/forum/view.php?bn=twist_montecarlo&key=1054237800

Posted by Peter Gumplinger (gum@triumf.ca) , Thursday May 29, 2003 12:52

Ahead of the meeting this afternoon, and for future reference, I
summarize again the available parameters in Geant3 which may (or may
not) affect the exact results of the comparison G3vsG4 and/or
G3vsdata.

The e614.ffcards contain may switches available to you, which together
allow for an almost infinite number of possible combinations. The
default cards are only my 'best educated guess' and I have always said
that some of the parameters may need 'tuning'. Such tuning can of
course only be done against data and not another MC and that data is
better not part of the stream used to determine the MPs.

The fist and most global switch is:

C
C ========= AUTO: IGAUTO =========
C
C          1 = Automatic computation of STMIN,STEMAX,DEEMAX,TMAXFD (default)
C                                    (except EPSIL)
C          0 = Tracking media parameters taken from the argument list of GSTMED
C                              (unless parameters are < 0 )
C
AUTO 0

The distributed G3 default is IGAUTO=1. However we are running with
IGAUTO=0 as E614GEANT default. On the other hand, via the card:

C
C ========== STPL: ISTEP STEPLIM(9) ==> Step length limiting parameters
C
C   1st entry: defines which material is linked with 'TARGET' tracking media
C
C                      9 = High Purity Aluminium
C                     21 = Mylar (default)
C                     17 = Gold
C                      6 = Graphite
C                     10 = Iron
C                     22 = Aluminized Mylar
C
C   2nd- 6th: tmaxfd, stemax, deemax, epsil and  stmin for ALL materials
C              [-1.]  [-0.5]  [-0.25] [.0001]    [-.1]
C   7th-10th: stemax, deemax, epsil and stmin      for 'TARGET'  material
C             [.001]  [-0.25] [.0001]   [-.1]
C
STPL 21 -1.0 -0.5 -0.25 0.0001 -0.1 -0.1 -0.0001 0.0001 0.0001

we are setting many of the tracking parameters negative, meaning that
we are in fact running with automatic computation of the tracking
parameters for all materials except the 'TARGET' material. It's your
choice to maybe use a different set of parameters than those
automatically calculated.

The parameter mentioned by Rene Brun is tmaxfd, i.e. the maximum
turning angle allowed in a magnetic field in one step. This angle is
20 deg by GEANT default. In older versions of E614GEANT (before 1.4)
this stepping parameter was as small as 0.1 deg. However, this was
then done as an indirect way of limiting the step size inside the
chamber cells, for the reason of finding accurately the minimum drift
time. In between chamber planes, where we have an He_N_ADMIXTURE, we
are presently using 20 deg and I have verified yesterday that it is
this tracking parameter which in fact limits the step size in the gas
between the chamber planes.

It is known that tmaxfd needs to be much smaller than 20 deg if G3 is
to navigate helices accurately in complicated geometries, for it not
to miss corners. This can happend if the straight line from 'prestep'
to 'poststep' point misses a volume corner, while the helix would
intersect it. With other words, the tracking precision is only as
accurate as the 'miss distance' between helix and the cord. However,
in the TWIST sensitive volume region, volumes are delimited by
virtually infinite straight planes (for the tracks of primary
interest).

I should be surprised if the accurate energy loss and multiple
scattering simulation in this He/N admixture would depend on 'tmaxfd',
for this would then mean that it is also wrong for simulations when
the B-field is off. On the other hand, it might just be interesting to
change it from 20 deg to 10deg or even down to 1deg. A short test
yesterday showed that the penality in performance (CPU at 1deg) is of
the order of 20% for a typical TWIST event.

Since delta rays do leave hits in the TWIST detector, we have no other
option but simulate with:

C ========== DRAY: IDRAY ==========
C
C          0 = no delta rays effect
C          1 = delta rays with generation of secondaries (default)
C          2 = same without generation of secondaries
C          Note: DRAY 1 is only possible for reduced Landau fluctuations
C                (LOSS 1). When full Landau fluctuations (LOSS 2) then
C                IDRAY = 0 regardless of the setting here.
C
C ========== LOSS: ILOSS ==========
C
C          0 = no energy loss effect
C          1 = delta ray and reduced Landau fluctuations
C          2 = full Landau fluctuations and no delta rays (default)
C          3 = same as 1
C          4 = average Energy loss and no fluctuations
C
LOSS 1

i.e. we are by default running with IDRAY=0 and ILOSS=1 (delta ray and
reduced Landau fluctuations).

We are also running with:

C ========== MULS: IMULS ==========
C
C          0 = no multiple scattering
C          1 = Moliere or single Coulomb scattering (default)
C          2 = same as 1
C          3 = Gaussian scattering with Rossi formula
C

i.e. IMULS=1 (Moliere or single Coulomb scattering).

We are running with the Urban model for energy loss for thin layers:

C ========== STRA: ISTRA ==========
C
C          0 = Urban model for energy loss for thin layer (default)
C          1 = PAI model    "    "     "    "    "    "
C          2 = ASHO model for 1<delta<=30
C

You could play and use one of the other two models (PAI or ASHO) and
see if the results change in any way.  Moreover, you can even change
which thin layer option you want to use for each of up to 12 tracking
media via the card:

C
C ========== FSTR: NSTRA M_STRA V_STRA
C
C     NSTRA (max.12!) Number of media with specified non-default thin
C                     layer option STRA
C     M_STRA(12) - List of tracking media numbers for thin layer option
C     V_STRA(12) - Value of thin layer option (must be REAL!);
C
C     i.e. v_stra(i) goes with m_stra(i)
C     in the CALL GSTPAR(m_stra(i),'STRA',v_stra(i))
C
C     NOTE: What's required is the TRACKING MEDIA number (not the MATERIAL
C           number). So, for example, HE-GAS (material) is 30, but
C           'HE IN MAGNET' is tracking medium 16. It is 16 that needs to
C           go into M_STRA. HOWEVER (!!!! NOTE !!!!) if you choose a
C           TRACKING MEDIA which refers to the same MATERIAL as another
C           TRACKING MEDIA, you MUST list all of them below, or else the
C           program will bomb!!!
C
C FSTR 7 30 31 33 36 15 16 19 0 0 0 0 0 1. 1. 1. 1. 1. 1. 1. 0. 0. 0. 0. 0.
C

If used, seven media would have specified options: #30, 31, 33, 36,
15, 16, 19. They would then all be simulated with the PAI model, while
all other media would use the Urban thin layer option.

TWIST uses a narrow energy range to tabulate the energy loss,
mult. scattering angles, with the most available (and convenient bins)
from 10keV to 100MeV for interpolation in the table, via:

C
C *** The ENERGY RANGE of the cross section and energy loss tables can
C     be fixed by the user with the data card :
C                   'ERAN'   EKMIN  EKMAX    NEKBIN
C     which defines nkbin bins from Ekmin to Ekmax in a logarithmic scale
C     The default is, 196 bins from 10 KeV to 100 MeV kinetic energy but
C     in logarithmic scale. NEKBIN must be 50<NEKBIN<200
C
ERAN 0.00001 0.1 196

and we use these lower tracking cuts:

C
C *** Low energy cutoffs - no tracking below these values [GeV] ***
C
C     CUTGAM   Kinetic energy cut for gammas [500 keV]
C     CUTELE   Kinetic energy cut for electrons [20 keV]
C     CUTHAD   Kinetic energy cut for hadrons [50 keV]
C     CUTNEU   Kinetic energy cut for neutral hadrons [50 keV]
C     CUTMUO   Kinetic energy cut for muons [10 keV]
C
CUTS 0.0005 0.00002 0.0005 0.0005 0.00001

i.e. 20keV for positrons and 10 kev (the lowest allowed for G3) for muons.

The actual tracking parameters for each material are listed in the
log-file, when you use:

C ========== PRIN: LPRIN
C == LPRIN ==   NPRIN names of GEANT data structures to be printed
C
PRIN 'TMED'