Reconstructing recent cosmic runs:
:-
For these tests, I've run the online event display in the playback mode with one of Monday runs' data (svtmon990105-1836-C.xtc), using release 7.7.2. I've used the IR-2 geometry, and the new bonding scheme. However, the question of putting in the correct layer 1 channel offset still requires some clarification.
In order to be able to reconstruct the cosmics, I've modified the current track finder, SvtTrackFinder, largerly following Gerry Lynch's recommendations on how this can be done. SvtTrackFinder needs at least three space points to find a track, so with only two layers it must be tricked into treating the top and bottom of one layer as two different layers. This worked quite well. Out of 105 events this tape contains, tracks were found in 16 of them.
Here is a typical picture of a reconstructed cosmic ray (the circles are space points, the black dots are hits on the track):
More pictures of cosmic rays are available here, here, here, and here.
Just for fun, here is a plot of Phi occupancies from that run, and here, a plot of Z occupancies. This plot shows the time stamp for the digis which went into the reconstructed hits.
L3 sequence:
:-
I am currently using the time of event obtained from the L3 trigger sequence in SvtHitReco to pre-select digis that are likely to have come from BBbar (as opposed to background) for further processing (see this note for a detailed discussion). This is done in SvtModuleBuilder::loadStripsFromDigis which creates root clusters from digis (in agreement with John Walsh's suggestion).
Anders Ryd's L3 sequence works well in 7.8.2. However, there are problem with testing the code with the new (supposedly much more realistic) background frames which I would like to use to tune the cuts. For example, SimApp does not seem to be able to store them in event collections. These problems might be resolved next week.