(present: AA, MF, TH, TI, ML)

News from the LArg week
Michel told the group that he received a very nice summary from Rob on topics from the LArg week. He encouraged people to look at the slides from the LArg week, which are on the LArg week agenda web pages In particular the slides from the following meetings:
combined testbeam software workshop
combined testbeam barrel & EC
combined testbeam EC2
combined testbeam EC1
software and performance
In all of these meetings, big contributions were made by Rob, Naoko and Michel. Michel's geometry calculations, which take care of the complicated shapes of the calorimeter cells, will be adopted and used in other packages of ATLAS software.

Status Reports

Tayfun Reported that he now looked at smaller cluster size - choosing 3x3 cells in each layer, and finds no difference compared to results he showed previously. (Michel noted as a concern that this cluster choice covers vastly different areas in the different layers.)
He also showed some more beam chamber plots - by requiring that the trajectories pass through F1/F2, he finds that in some runs the beam profile gets severely cut, while in others it remains more or less intact. After some discussion, it appeared that the problem seems to be that when he calculates the trajectories, he inserts the table-y and cryostat-x back into the calculation, even though they have already been corrected for in Naoko's beam-chamber fit.

Margret also showed some slides on trajectories she investigated using the beam chamber information only. She plotted the slope(y) vs y at z=0 (and also at some other large z) and finds some structure (bands)for some of the runs. This structure leads to intercepts of trajectories at several z-position. The run she investigated had 3 intercepts, one between MWPC3 and MWPC4, another around one end of the Fe beam dump and a third around Bend9. As far as she can tell, the energy of the trajectories of the different bands appear to populate the total energy peak in much the same way, so at this point no corrections appear to be required. (But it might be worth keeping an eye on this.)

Tamara showed some nice plots of her cell-occupancy studies. She requires an occupancy of 10% with |E|>2sigma(noise). Then, starting from the centre of the occupancy distribution, she only keeps cells that are connected together. This topology cut somewhat reduces the number of non-gaussian noisy cells (high occupancy) not really part of the list of cells which used to get included before. She will now look at resolution and response for this 10% requirement and then also look at different occupancy percentages.