The NLD EM resolution (19.5%/ÖE) is higher than the expected value of 15%. The 40% HAD resolution seems very low when compared to the current SLD value of 65%. If GISMO, based on SLD LAC and LUM configurations, can reproduce established SLD data, the unexpected NLD values might be more convincing. The dampening effect of a threshold value on overall NLD response might also help explain NLD results.
The GISMO NLD calorimeter is broken into two basic geometries: the barrel, and endcap (EC). The SLD LAC can be simulated using a barrel configuration in GISMO, based on current SLD specs (SLD Design Report, 12/85). The SLD LUM can replace the current NLD EC configuration, thereby verifying that component of NLD geometry.
An arbitrary threshold value applied to segment energies at the analysis stage, will restrict energy measurement to segments where ionization is the primary mechanism for energy loss.
The number of layers, their thicknesses, and the layer material type can be modified to simulate the barrel or endcap geometries of any detector. SLD specs were inserted in place of NLD, and particular alloys generated using PEGS. The barrel and endcap dimensions were changed to accomodate SLD layering.
The Ntupler arrays used to insure proper booking of the CW Ntuple, and their maximum dimensions used to calculate and book particle IDs, were edited to accomodate the new specs.
A ROOT macro was used to form energy aggregates, calculate population mean and width, and to filter corrected segment energy through a threshold value.
The NLD EM EC was replaced by a SLD LMSAT geometry:
and the NLD barrel by the SLD barrel LAC:
where the LAC EM component is preceded by 8.455 cm Al simulating the SLD cryostat.
rin(cm) | rout(cm) | Layers | length(cm) | j seg | Dj | tanl seg | Dtanl | |
LUM EC | 15 | 48 | 23 | 15 (inner z=110) | 192 | 0.033 | 70 | 0.022 |
LAC EM | 40.55 | 75.7 | 56 | 125 | 192 | 0.033 | 70 | 0.022 |
LAC HAD | 80 | 126.5 | 52 | 125 | 32 | 0.065 | 96 | 0.016 |
The response and resolution.simulated by the GISMO LUM were 78% and 2.7%/ÖE. Accepted SLD values are 77% (cf. Kevin Pitts) and 3%/ÖE at 50GeV. The LUM data was corrected using a 15% sampling fraction.
The LAC specs ran with a cryostat layer in one instance, and no cryostat in another. The data set that ran without the cryostat accounts for EM showers only. The response and resolution for both configurations are overlayed for easy comparison to one another. The e/m of 76% and 7% resolution should be compared to SLD values of 69% and 13%. A hadronic shower through this geometry produced a p/m of 0.52 and resolution of 55%/ÖE compared to SLD values of 0.45 and 65%/ÖE. Raw LAC data was scaled using an 18% EM sampling fraction and 7% HAD.
The response fit for p exhibits a std. dev. not apparant in NLD data. The resolution chart differs from NLD insomuch as a complete shower was hard to come by. The trendline indicates a resolution between 0.45 and 0.65 for incident energies between 2 and 10GeV. The LAC does not fully contain a shower of more than ~5GeV, where the more energetic shower can be seen punching through in a ROOT histogram. The geometry above contains ~90% of a 10GeV HAD shower.
After tripling the HAD layering to 150, 98.7% of 12GeV showers were contained. The response increased by 4% to 0.56 while resolution improved to 42%/ÖE, a difference of 13%/ÖE when compared to a shower less contained.
A 5MeV threshold on corrected segment energies formed a De/m of ~3% between the 75.5% of uncut and 72.6% cut responses. When filtered through a 15MeV threshold this difference increased to 4% for an EM shower.
There are a couple of differences between SLD LAC and the GISMO simulated SLD configurations. The CRID was excluded as a radiator and it's effect on response was neglected. The WIC is also not used though it is not implemented in SLD either.
Though the LAC specs reproduce SLD data to within 10%, the exceptional hadron resolution is suspicious and raises concerns about the validity of the GISMO barrel algorithm. When a 10GeV hadronic shower was contained more fully, the response increased by only 4%. The resolution improved by nearly 13% leading to a significant difference with accepted SLD data. The increased containment also brings the GISMO LAC resolution quite close to the NLD value of 40%/ÖE.
The GISMO LUM response agrees to within 1% of current SLD LUM data, and 0.3% of SLD resolution. The EC algorthim seems to accommodate a LUM configuration and is consistent with current SLD data.
C.Colgur and R.Dubois
Last Modified: 01/13/04 12:55