Minutes of the Testbeam Meeting, held Wednesday May 14, 2003 ================================================================= (Present: ML, NK, TI, MF, TS, RK, IG) Margret gave a short status report on her shower shape analysis The corresponding slides are on our web site. She has now looked at 5 different energies, each with various amounts of X0 in front of the cryostat. This allows a parametrization of the shower shape as a function of energy. The resulting prediction for the depth of the shower maximum is in line with the commonly used parametrization: tmax = ln(E/e) -c TBRootAna: Ian is now making good headway on the updates for TBRootAna. All Ntuple analyses should eventually be done using TBRootAna, which will likely make them more fool-proof and will facilitate contributions of useful packages written by any group members to be used by all. For example, one of the things that will go into TBRootAna is a full geometry service. Analyses that require information not given in the Ntuple will have to be done in athena. The design of TBRootAna will keep the inheritance structure of the early version. Ian is now using the latest versions of gcc and root and he is refining the build system. This involves using "automake" and "autocomp", which automatically takes care of versions and dependencies and _generates_ a makefile automatically. This will make the package a lot more robust to version-changes in the future. Next on the list are the generation of classes to that different types of Ntuples can be read with the same package. Once they are ready, the updated Ntuple ready can be integrated into the program. The plan is to make the HEC and EMEC Ntuple information available at the same time. It should also eventually be possible to analyze multiple runs with TBRootAna. Once Ian has a useful working version of the updated package, he will give a little presentation to teach the group how to use it, how to make user algorithm and how user algorithms can be turned into system algorithms if they turn out to be generally useful. Richard pointed out that once the package is running, it should be made available to other collaborators outside UVic too. Ian estimates that he'll have a bare-bones updated version ready by the end of this month, or earlier. Aspects of the Data that came up during the meeting: The default-processes for the ntuple production are: pedestal subtraction (all pedestal files already exist), digital filtering with weights, and (optional) adc to nA conversion with one of Pavol's calibration files. The pedestals are typically set to be around 1000 adc counts, so they are unlikely to be clipped at the bottom. Manuella and her student are just wrapping up a noise correlation study on the pedestals and word has it that a) the pedestals are very stable and that b) there is very little correlation between different channels in the HEC. Richard mentioned that one of the first things to look at with TBRootAna should be the spill-dependency of the out-of phase events. It was mentioned that at the moment, the LArBookkeepingSvc doesn't work in batch jobs. This is too bad, because Naoko had put a lot of corrections into the bookkeeping for things that were entered wrong in the run-header by the shift crew, but later got corrected in the run logbook (like: beam-energy or cryostat position). Tayfun found in run 12201 (119.1 GeV positron) that about 126 out of 5000 events had zero (exact, 0.) as a signal for all channels. Nobody could explain that off-hand. It needs to be investigated and Naoko agreed to help. A.O.B.: Tamara has been brushing up again on ROOT, athena and reading the LArTBCombNtuple.cxx program to see what it does. Michel thought that Tamara and Tayfun should practice writing a TopAlgorithm in athena, but it was questioned whether that would be a pressing item, since the near-future analyses should be tackled on the Ntuples and with TBRootAna. Notes added after the meeting: Michel had a phone conversation with Sven, and the following was discussed. Out-of-phase events: Sven confirmed that the event header event number and the EMEC event number see always in sync. Also, Sven confirmed that sometimes the HEC and EMEC event numbers do not agree, but the events seem to be in sync according to the monitoring of some observable as a function of event number. Curious HV problem: Sven and Hendrik for a troublesome problem with the second layer (iz = 2) of the HEC module 2 (iphi = 3 and 4). It was known that for run 13061 and up (I think) one (of 4) HV channels was disconnected. Since this affects one subgap out of four in each of the 16 gaps (yes, there are 16 8mm gaps in HEC layer 2), the problem is well distributed longitudinally and we can correct by applying a correction factor (of 4/3 here), and getting a bit more noise. But now it appears that for all runs, the first 2 gaps (that is the first 8 sub gaps) seem to be not working. This is a problem because this cannot be corrected. Sven and Hendrik found this by looking at the signal distribution in phi for a) pion events that have little energy in the first layer (late showers) and b) pion event that have little energy in the 3rd layer (early showers). A phi dip in module 2 is seen only for the early shower events, pointing to a problem in the front of the layer 2 cells. They concluded that only 2/16 gaps were affected by looking at the muon response. It is a mystery as to what kind of hardware fault could cause such a failure. Under investigations. This will impact all pion analysis. A. Kiryunin has been asked about the possibility of simulating dead gaps. Michel, Sven and Naoko also discussed some urgent LArHECTBAna improvements. A list of short term tasks was proposed and is contained in the email below. Dear Sven and Naoko, Many thanks again for spending time with me today discussing various LArHECTBAna improvements. I have tried to summarized here a possible very short term task list that follows what we discussed. LArDigitalFiltering ******************* Naoko will correct the share/LArHECTBAna_jobOptions_combined.txt file, replacing //Digital.t0_cub_epi = 100; with //Digital.t0_cub_epi = 103; as this is the correct global digital filtering time offset for the HEC when using global cubic timing. Also, she will put the options LArNTup.fill_TimeInfo =1; LArNTup.fill_Properties =1; to one, so that the ntuple structure is fixed to the maximum size. This is important for our UVic development of TBRootAna. This is already done! Sven and I both agree that the user should not have to know about the t0_cub_epi nor about the t0_dig properties. They should be for expert only. This information should go in the time offset files. I propose that we allow for some words at the top of each time offset file for global constants. At this time, we only need two, that is t0_cub_epi and t0_dig, but we may need more later. We can put only two for now, as this does not affect the database table which can cope with a variable number of such global constants. Sven will modify LArDigitalFiltering to read this new format of time offsets and put new time offset files (with different date on their name!) on hectbmon. Naoko will modify the database table format of the digital filtering offsets accordingly. Run information *************** Sven will prepare an ascii file with out-of-phase information for run numbers, following his experience in tracking down these problems, and let Naoko know about it. Naoko will input this information in LArBookkeeping (she has the appropriate privileges) in the "comment" field. TDC event number **************** Naoko will improve the Alg_tdc.cxx to include the TDC event number within the converter. Please let me know if this is all agreeable to you! Cheers Michel