LCD Sim Weekly
6/28/05

---

NG: TB working on WHIZARD events.  Pass responsibility to JM once the output is verified.

Events at ab^-1.

For ppl doing FastMC analysis.  Also for those with sophisticated analysis --> apply filter to events as "trigger".

NG: DF added for FastMC.

DF: Could add more realism to FastMC, including some from TESLA TDR.  

NG: Talking with RC on cluster centroids/directions.  Put in some modifications last night to do cluster centroids.  Will send modified example later this afternoon.  Can start doing angular pulls/residuals.

GW: Planning to do studies of photon ID and recon on photon 4-vectors.  

Students working on EM id codes.

NG: SK using example from RC.  Pretty good id, because old detector that it "trained on" is fairly similar.

NG: JS on Zvtop?

JS: Fixed test failures.  Now debugging.  Sometimes, no detector is retrieved from the event.

NG: Files?

JS: At Oregon have a summer student trying to bring up-to-speed.  Jim would also like to have SLIC installed.  Hopefully, student can help with testing.

News from Su Dong on validation files?

Met with LCFI group --> he would generate some files using svTop FORTRAN.

NG: No progress on that.

MC: NTM also working on this --> shape and progression of showers through calorimeters.  Looking at charged pions, photons, neutrons.  In SDFeb05 with segmented Si Ecal and RPC Hcal --> shower containment is good in Hcal.  Comprable to that in Ecal.  Much easier time finding the fragments, main shape of core compared to Ecal.  Would like to see what happens with scintillator Hcal --> understand that behavior is very different.  Events through MSTClusterBuilder --> how much of cluster is reconstructed?  In cases where fragments, e.g. small collections of hits that make small clusters, can they be "picked-up" by associating to nearby clusters.  Looks like it works well in Ecal with tight clustering criteria: ~3 cm.  Can go to distance to core of 15-20 cm.  This looks like it will work reasonably well for fragment stage, not core stage.    

Would like to generalize interface to add sub-clusters to a cluster.  Will hold off, because clustering interface is changing in org.lcsim .

SM: Take track and find MIP cluster, then find shower associated with that track.  Extending to endcap to use whole calorimeter.  Clusters defined by barrel Ecal, endap Ecal, barrel Hcal, endcap Hcal.  Use RefinedCluster to combine the endcap and barrel clusters.  Using example from NearestNeighbourClusterer from NG.  Hopefully, it will be generic enough to work with events generated using SLIC.  Discussing whether HC are uniquely assigned to a subdet.  Seems much easier if they are.  Would like to do this with all collections in calorimeter right now.  Once have RefinedCluster, should have shower start point.  Can add shower association onto this.  

MC: Thinking about coming from "opposite direction".  Use WM's MIP finder.

SM: MIP it -> hit that matches track.  Look for big/small energy.  Could find interaction point from this.  Difficult, because need to have sampling fraction, calibration.

NG: Need to know what a MIP deposition is.

SM: Need to fit with Landau first.  MIP = most probably value + signal .  Simplify by using density.  Define a window around the hits and see what the density of that hit is by density of the window.  Get back 1,2,3,etc. for density measurement.  When != 1, should still call MIP?

MC: WM came up with set of criteria for defining a MIP.  Probably in talk from March.  (LCWS)

SM: MIP = "not too many extra hits" until end.  Interaction layer vs. # of hits.

Want to keep the window small.

Start shower earlier and find with cone.

Hope will work with any detector generated by SLIC.

NG: MR?

MR: Experimenting with org.lcsim .  Experimenting with cluster cheater.

NG: GL w/ DigiSim updates?

GL: Improving implementation of x-talk and noise: exponential and gaussian noise.  Moving to NP geometry.  Porting VZ to lcsim.

NG: Encourage to writeup DigiSim tutorial on Confluence.

GL: On ToDo list.

NG: Start simple with "Getting Started".  

Trying to refine cluster classes.

MC: Stuck in hep.lcd framework.  Which correpsonds to AbstractCluster?

NG: All of them!  Interface not well-defined.  

GW: Cal cluster class will change?

NG: Context of LCIO = Cluster.  Persistence probably won't change.  Interface to end user will probably change.