D0 Calorimeter Task Force Meeting Minutes - Aug 20, 2002

Present: Nirmalya Parua, Volker Buescher, Marumi Kado, Laurent Duflot, Suyong Choi, John Womersley, Leslie Groer, Marek Zielinski (scribe)

Video: Dean Schamberger, Serban Protopopescu, Ursula Bassler

Laurent reviewed the structure and implementation aspects of the pileup package. It handles application of calo weights, pile-up of min-bias MC events, simulation of noise and of non-linear effects in the calorimeter; a path for overlaying the zero-bias events from data is being developed.

Of particular interest to this group, the noise is simulated using the difference of two Gaussian random numbers, which is not consistent with current determination of noise RMS values from data. This needs to be fixed (by using only one Gaussian). Also, care needs to be taken that non-linearity corrections and the noise simulation are done in consistent manner. Rather than simple rcp vectors, we'd prefer to use the pedestal&sigma file from calunpdata to simulate the noise. The MC farms should be contacted to make sure that this file is part of their tarball and is known to mc_runjob. There are different noise for the x1 and x8 gains so we will have to determine the gain from the cumulated energy before noise simulation and non-linearity simulation. This also requires adding some functionality to CalDataChunk to keep track of the gain path that was selected so that CalMCtoUnpReco forces this gain path.

Dean made comments that NLC should be applied to noise-induced energy at nominal value (ie no noise) rather than the observed value, and that there are two sources of SCA non-linearity: for the current crossing signals, and for the baseline subtraction (this level of detail appears unnecessary to simulate at this time).

For real data overlay path, we may need to simulate noise only within the zero suppression limits. Dean is developing a way to take non-zero-suppressed zero-bias triggers interspersed within the regular data stream, so we still have to check if the zero-bias data is zero suppressed.

We had a long discussion about the need to clean up the code for calorimeter weights. This would be a longer-term project and it should be discussed some more by a few experts off-line, before we can make a decision. One item related to accuracy of the simulation, rather than the organization of code, was that the L1 tower weights should correspond to the hardware settings (resistors+dial adjustments).

Sike continues studies of jet variables in tau and QCD MC events. She generated 3 sets with of files: with no non-linearities, with simulation of non-linearities but no correction, and with non-linearities in MC and NLC in reco; each set was done in 4 versions, using zero-suppression thresholds at 1, 1.5, 2 and 2.5 sigma. She also looked at tau events with no noise and with double noise. When varying the zero-suppression threshold from 1.5 sigma to 2.5 sigma she noted little effect on the tau width and profile, but significant impact on tau isolation (see the talk for all distributions). She will look next at the QCD files.
Marek and Leslie are contacting MC farmers at Lyon and UTA for information about release verions, storage constraints, and possible difficulties with the intended mode of operation (ie. reprocessing individual steps of the generation chain). Pythia card files and other parameters for MC jobs are being collected with the help from Qizhong and Steve Muanza. Serban agreed to be the cvs czar for the CTF MC cards.