Testing changes to cvs geometry file
This page contains work I've done looking at cvs vs. v23 with the extra SMD layers turned off. Results are presented in reverse chronological order.
Tests with PbAlloy Mixture - 4 November 2009
I got the idea to look into the PbAlloy mixture more, since it's really the only change between cvs and v23. v23 of Jason's geometry defines the PbAlloy Mixture as:
c--
c-- Lead alloy used in the radiators
c--
Component Sn A=118.710 Z=50 W=0.014
Component Ca A=40.0780 Z=20 W=0.00075
Component Al A=26.9815 Z=13 W=0.0003
Component Pb A=207.190 Z=82 W=0.98495
Mixture PbAlloy DENS=11.35
in other words, it is almost 98.5% lead. The ELED block looks like:
c----------------------------------------------------------------- Block ELED --
c--
Block ELED is a lead absorber plate
c--
c--
Material PbAlloy
Medium Ecal_lead
Attribute ELED seen=1 colo=4 fill=1 lsty=1
c--
Shape TUBS dz=emcs_pbplate/2,
rmin=(current)*tan_low,
rmax=(current+emcs_pbplate)*tan_upp,
c--
Call ecal_set_cuts( ag_imed, 'radiator' )
c--
EndBlock
c--
c-----------------------------------------------------------------------------
Note: I am unsure of the purpose of defining Material PbAlloy and then Medium Ecal_lead.
I redefined the PbAlloy so it was 100 % lead to make the ELED block look like:
* - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
Block ELED is lead absorber Plate
*
Component Pb A=207.19 Z=82 W=1.0
Mixture pballoy Dens=11.35
Medium Ecal_lead
Attribute ELED seen=1 colo=4 fill=1
Shape TUBS dz=emcs_Pbplate/2,
rmin=(current)*Tan_Low,
rmax=(current+emcs_Pbplate)*Tan_Upp,
Call GSTPAR (ag_imed,'CUTGAM',0.00008)
Call GSTPAR (ag_imed,'CUTELE',0.001)
Call GSTPAR (ag_imed,'BCUTE',0.0001)
Call GSTPAR (ag_imed,'CUTNEU',0.001)
Call GSTPAR (ag_imed,'CUTHAD',0.001)
Call GSTPAR (ag_imed,'CUTMUO',0.001)
endblock
* ----------------------------------------------------------------------------
Note: calling ecal_set_cuts should be equivalent to all the GSTPAR statements I have.
I would expect that this alloy has identical properties to lead. However, if I load this geometry and do a gprint mate I see:
MATERIAL A Z DENSITY RADIAT L ABSORP L
ECAL_PBALLOY 207.190 82.000 11.350 0.562E+00 0.177E+02
compared with pure lead:
LEAD 207.190 82.000 11.350 0.560E+00 0.185E+02
and the original alloy:
v23 205.772 81.485 11.350 0.564E+00 0.176E+02
I made more simulations of single photons thrown over the entire endcap with this 100% Lead mixture. I run with (solid lines) and without (dashed lines) the LOW_EM option and with (pink) and without (blue) the "Medium Ecal_lead" line in the ELED block. The plots are shown below.
.gif)
Conclusion: Defining a mixture seems to change the outcome with and without the LOW_EM option. The "Medium Ecal_lead" line seems to have no effect.
Another comment: you might then worry about defining the cladding as a steel mixture, rather than pure iron. I tried the same trick as above but made the steel mixture 100% iron, and I see the following:
IRON 55.850 26.000 7.870 0.176E+01 0.171E+02
ECAL_STEEL 55.850 26.000 7.870 0.176E+01 0.168E+02
For whatever reason, the radiation length doesn't change for the iron and steel combination. Perhaps the way GEANT handles mixtures has to do with how heavy they are? This is probably why we see the ELED change account for almost all of the change.
More tests with LOW_EM option - 31 October 2009
I ran over the same samples as 30 October, with only the full geometry and the LOW_EM option. I added another test sample, where I changed the particle tracking cuts in the cvs file to be identical to what Jason had (labeled "+cuts" in the plots), in case that would be important when running with the LOW_EM option. The plots show reconstructed tower energy in the entire endcap/thrown energy.
.gif)
This confirms what Ilya has seen - that with the ELED block the original and modified cvs files behave very differently with the LOW_EM option.
More Results, running over full endcap - 30 October 2009
I produce more single photon events where I throw over the full endcap in eta and phi and a pt range from 6-10. Below are plots of Tower Energy Reco/Energy Thrown for the various conditions. There are 10,000 events at each point. I do not use the LOW_EM option in these tests.
Changing the ERAD and ELED blocks is cvs version gets us significantly closer the v23. The means and widths of the distributions are give in the table below.
| Geometry Version | EEMC-only mean | Full mean | EEMC-only width | Full Width |
| cvs | 0.94 | 0.96 | 0.05 | 0.05 |
|
cvs+ELED+ERAD blocks changed |
0.94 | 0.93 | 0.05 | 0.05 |
|
v23 with extra SMD layers turned off |
0.94 | 0.92 | 0.05 | 0.05 |
This seems to confirm that the ELED block is the cause of the trouble.
Introduction - 29 October 2009
I have tried to carefully determine the changes made to ecalgeo.g after the initial CAir fix and look at the effect of each one of these changes. I identified some potentially significant changes that could effect the tower sampling fraction:
- Changes to the ERAD block
- Changes to the ELED block
- Changes to particle tracking cuts
- Changes to the tie rod diameter
Note I say "potentially significant" because often more than one change was implemented at a time - I have to test to see if it was the important one. I run the most recent geometry file with the new SMD layers turned off so I've neglected changes that occured when only these new SMD layers were changed.
I have simulated 500 single photon events where I shoot the photons in a very narrow range of energy directly in the middle of a tower in Sector 11, with no LOW_EM option. The thrown energy is 30 GeV. I use StADC2EMaker to get the reconstructed energy (set scale factor to 1.0) in the tower.
Results - 29 October 2009
Below are comparisons of reconstructed tower energy for various changes to cvs version, compared to v23 with additional SMD layers turned off. I keep previous changes each time - i.e. "ELED changed to match v23" also has "ERAD changed to match v23" Typical uncertainties on the mean reconstructed energy are 0.1 GeV.
| Geometry version | EEMC stand-alone | Full STAR geometry |
| cvs | 28.3 | 28.8 |
| ERAD changed to match v23 | 28.2 | 28.7 |
| ELED changed to match v23 | 28.2 | 27.9 |
|
Particle tracking cuts changed to match v23 (pre,post, tiles all the same) |
28.2 | 27.9 |
| Change tie rod diameter to match v23 | 28.2 | 28.0 |
| v23 with SMD layers removed | 28.1 | 27.9 |
For reference, below is the ERAD block in the cvs version:
-------------------------------------------------------------------------------------------
Block ERAD is radiator
*
Material Iron
Attribute ERAD seen=1 colo=6 fill=1 ! violet
Shape CONS dz=radiator/2,
rmn1=(current)*Tan_Low-dd,
rmn2=(current+cell)*Tan_Low-dd,
rmx1=(current)*Tan_Upp+dup,
rmx2=(current+radiator)*Tan_Upp+dup
Create and Position ELED
endblock
-------------------------------------------------------------------------------------------
Below is the ERAD block in the modified version:
* ----------------------------------------------------------------------------
Block ERAD is radiator
*
c--
c--
Component Cr A=51.996 Z=24 W=0.19
Component Ni A=58.6934 Z=28 W=0.09
Component Fe A=55.845 Z=26 W=0.72
Mixture steel Dens=8.03
c--
Attribute ERAD seen=1 colo=6 fill=1 ! violet
Shape CONS dz=radiator/2,
rmn1=(current)*Tan_Low-dd,
rmn2=(current+cell)*Tan_Low-dd,
rmx1=(current)*Tan_Upp+dup,
rmx2=(current+radiator)*Tan_Upp+dup
c--
Create and Position ELED
c--
endblock
--------------------------------------------------------------------------
Below is the ELED block in the cvs version:
* - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
Block ELED is lead absorber Plate
*
Material Lead
Attribute ELED seen=1 colo=4 fill=1
Shape TUBS dz=emcs_Pbplate/2,
rmin=(current)*Tan_Low,
rmax=(current+emcs_Pbplate)*Tan_Upp,
Call GSTPAR (ag_imed,'CUTGAM',0.00008)
Call GSTPAR (ag_imed,'CUTELE',0.001)
Call GSTPAR (ag_imed,'BCUTE',0.0001)
Call GSTPAR (ag_imed,'CUTNEU',0.001)
Call GSTPAR (ag_imed,'CUTHAD',0.001)
Call GSTPAR (ag_imed,'CUTMUO',0.001)
endblock
* ----------------------------------------------------------------------------
Below is the modified ELED block:
* - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
Block ELED is lead absorber Plate
*
Component Sn A=118.710 Z=50 W=0.014
Component Ca A=40.078 Z=20 W=0.00075
Component Al A=26.9815 Z=13 W=0.0003
Component Pb A=207.19 Z=82 W=0.98495
Mixture pballoy Dens=11.35
c--
Attribute ELED seen=1 colo=4 fill=1
c--
Shape TUBS dz=emcs_Pbplate/2,
rmin=(current)*Tan_Low,
rmax=(current+emcs_Pbplate)*Tan_Upp,
rmax=(current+emcs_Pbplate)*Tan_Upp,
Call GSTPAR (ag_imed,'CUTGAM',0.00008)
Call GSTPAR (ag_imed,'CUTELE',0.001)
Call GSTPAR (ag_imed,'BCUTE',0.0001)
Call GSTPAR (ag_imed,'CUTNEU',0.001)
Call GSTPAR (ag_imed,'CUTHAD',0.001)
Call GSTPAR (ag_imed,'CUTMUO',0.001)
endblock
Note: this block has an additional Medium Ecal_Lead statment in the final version
Different material properties from GEANT
For the Lead Absorber
A Z DENSITY RADIAT L ABSORP L
CVS 207.190 82.000 11.350 0.560E+00 0.185E+02
v23 205.772 81.485 11.350 0.564E+00 0.176E+02
For the Radiator
CVS 55.850 26.000 7.870 0.176E+01 0.171E+02
v23 55.370 25.800 8.030 0.173E+01 0.164E+02
- aliceb's blog
- Login or register to post comments