LCD Weekly: Simu and Reco 5/3/05 SLAC NG (Norman) TJ (Tony) JM (Jeremy) SM (Steve) WM (Wolfgang) JS (Jan Strube) WL (Willy) NTM (Niels) NS (Nik Sinev) MA (Makoto Asai) TK (Tatsumi Koi) CJ (Chen Jin) TB (Tim Barklow) TN (Tim Nelson) NICADD GL (Guilherme Lima) ANL SK (Steve Kuhlman) LX (Lei Xia) -- NG: SK to continue short-term with current analysis package for Snowmass. SK: Since January, improve the algorithm to get 40% / sqrt(E). Don't care which software. Take unused hits and put in package and gives quality for neutron or fragment. Since we already have [something like this], then try this package. Wait until I can try something that will give neutron quality. SM: Package doesn't work -- cluster ID example. Has a lot of histograms. NG/TJ: Trivial to fix. SM: In principle, if the package works on any cluster, then should be able to use the fixed cone cluster and feed into it. NG: See if neural net is sensitive to what's left over. SM: Never trained on fragments of tracks. Only trained on photons and tracks. SK: Fragment a true neutron, then not too difficult to distinguish. 30 neutron cluster candidates / event. Only 2 true neutrons / event. SM: Change cut, cone. SK: Mostly "junk". NG: Almost all = fragments. SM: No fragments from photon hits because of way we associate hits. Only comes from tracks. Fragments of neutrons and "real" charged hadrons. SK: Almost all tracks. SM: Not sure why? SK: 30 tracks / event and a few neutrons. SM: Only thing left. SK: Talk at Thursday meeting. Lei and I didn't know about the meeting. Somehow ropped off the list. NG: Will send around again. Get a survey of ppl on list and send out reminder. SK: LX went back and looked over everything as "new user starting from scratch". Let him tell. LX: Working on developing clustering algorithm. Tried to get a clustering algorithm out of density function. Seems to work okay. Still testing results of clustering. Starting from there, try to get full PFA. NG: Starting with assumption of digital hcal -- density = locality of nearby hits? LX: Yes. SM: Contact Vishnu @ NICADD because done the same thing. SM: NICADD has clustering w/ density? GL: Density function that Vishnu has already implemented. SM: Neutrons and Klongs email? Physics list has different program for fragmenting neutrons and klongs. Neutrons, protons and pions all use Bertini cascade. Klongs use old model -- low energy parameterised. Now more data so change klongs to use Bertini. LX: Neutron results at low energy = true? Poor Eres on low energy neutrons? SM: Worsening of energy from Klong to neutrons. Should get DW's paper from MC workshop. Big differences between Bertini and parameterised. New G4 datasets there should be change to Klongs. SK: Lei found some events w/ ~200 MeV neutron left nothing in calorimeter. Is that expected? NG: I guess it could happen. sM: Looked at Eres of neutrons and Klongs --> different. Neutrons at any energy -- all had less energy in calorimeter. NG: No time cut. Neutral hadrons will change, because going to new physics list. SK: Is an SiD "default detector" now? Regen w/ new G4 physics list. NG: New detector = "SiD" rather than "SD" NG: GL bring up to date re NICADD and DigiSim for Snowmass? GL: Meeting yesterday but didn't come up with final list for Snowmass. Nothing final yet. Physics results with preliminary algorithms + DigiSim. Going to finalize by next week. NG: Is anyone moving to org.lcsim? Or mostly working with hep.lcd? GL: Working on DigiSim in org.lcsim -- currently working. Clusters interface implemented in org.lcsim "pretty soon". If it doesn't take a long time, the goal is to use org.lcsim. TJ: Need anything else after bugfixes? GL: Finding missing codes as go along. TJ: SDJan03 same between LCDG4 and SLIC? GL: Different barrel and endcap hits need to be in different collections. Need to change in LCDG4. TJ: org.lcsim could handle 1 collection for barrel and endcaps but more convenient to separate. GL: Ecal and Hcal barrel/endcap hits into different collections. Tried to load w/ JAS but unable to make a list of hits. Send file to SLAC for debug. Problem could be def of theta/phi bits. TJ: org.lcsim plugin? GL: Yes. TJ: Requires data and description must agree with each other. Send file to me. NG: Progress on Geant4 Eres? MA: No final conclusion. Smallest range cut probably correct. Limit step length in absorber -> recover > 2/3 of missing energy. Depends on limitation of the step size. ~2/3 is from backscattering of "tiny electrons". Smallest range cut gives correct results. Introduce a smaller box inside the absorber box w/ a different range cut. NG: "Thin skin" on outside. MA: Skin thickness should be a parameter. Working with Michelle. NG: Reproduce w/ EGS? MA: EGS/G3/G4 all have same issue of production threshhold close to the boundary. Smallest range cut gives the best answer. Limiting the step length is not very efficient. Step limits different in "skin" or main volume may give similar results. TJ: Need skin on absorber? MA: Backscatter from scintillator <-> absorber, but abosorber -> scintillator dominates. SM: Different w/ neutrons? MA: Ratio should be different. Active layer thickness also affects. Most worried about gas chambers. NG: ~1 mm sensitive layer w/ TPC and almost empty. MA: Limit backscatter from TPC to absorber. NG: Current simulation of RPC response is still in flux. NG: Status report on tracking? TN: Still have outstanding questions and need to master/understand tools. NG: Bruce and MS students working on long-shaping digitization including readout electronics. TN: Talked about it. WM: Bug in clustering. JS: Wanted to merge swimmer into helix [?]. Put into CVS once it is working. Test packages for fitter. Question later on how to integrate with JAS. Tried to run over event samples and had problems. How to tell JAS where to find packages? Gave CLASSPATH but got "package not found". TJ: Hierarchy of class loaders -- UI, extensions, and system. Point CLASSPATH to top of package dir. NG: Builds org.lcsim with mods, how to point to jar file? One way is to copy into extensions dir by hand. ID maven target to distribute it. TN: Analysis job to make some plots. One end to other. Couldn't find a good example of analysis jobs for lcsim. NG: [show some examples from JAS3 browser] TN: org.lcsim mixed in with old examples. JS: Compare old code w/ new code. hep.lcd swimmer and my swimmer and see results. TJ: Same event as LCIO/SIO and feed into old and new frameworks. Start from different points. NG: LCDG4 output will give both SIO/LCIO. JS: Doesn't help if it is in different packages. JS: Track parameters may change? TJ: Hope agree on track parameters f/ BaBar. cm -> mm. JS: Helix Swim -- lcd uses different parameter f/ curvature. NG: No longer valid. TJ: Make sure that HelixSwimmer is not swimming in the wrong direction. NG: Updates from NS? NS: Working on CCD-based reco w/ disks. Efficiency is ~85% for far forward region. TJ: New version of LCIO next week. NG: Changes for TPC hits. At least 1 additional version before Snowmass. Also bugfixes for double/float bug f/ RC. JM: Get the track momentum in SimTrackerHit? NG: No, but should. JM: SimTPCHits different than SimTrackerHit? NG: Yes. WM: CD to include org.lcsim and hep.lcd? NG: Would like to deprecate hep.lcd but may include it.