The new code has been integrated into the hitfinding analysis module, and may be selected through a switch in the hit_sw table. It has several modes and switches of its own. Click here for a list of switches currently in use.
The new hitfinder is currently in "beta" status-- it is being tested by various collaborators, and we hope to make a decision about whether to use it as default based upon evaluations and comparisons of the performance of the new and old algorithms with real data.
Please help as well in the evaluations.
Here are the instructions for getting the files:
You may get the files by logging into vdgus1.mps.ohio-state.edu as e895 (password is our canonical password-- call me at 614-292-4775 if you don't know it).
Go to the hitfinder/ subdirectory.
There, you will find the file E895hitfinder.f (all the routines are in the file-- I do not fsplit!), and the macro file hitfinder_switches.kumac.
Then, you should be able to copy the fortran code to your machine and make the eosgx executable (Dieter has some nice make files for this).
Finally, you will see another subdirectory called evaluation/ (in the hitfinder/ subdirectory). If you would like to do evaluation of the hitfinder (PLEASE DO!) with various settings (e.g. old algorithm vs new), then make an evaluation/ subdirectory in the directory where you will run eosgx.
Then in eosgx, set the hit switches (and track switches) to whatever you want, turn on the camac, hit, and trk modules, and open a data file for processing "tas/data/input ..." Then, "exec do_evaluation *nevents* *setting number*", where *nevents* is the number of events to process for the evaluation (like 20-100), and *setting number* is just some reference number you give to the switch configuration you've set up.
What happens then is that various vectors and histograms are filled, and 4 files are made:
the switch settings you've set.
some histograms useful for evaluating the code performance (track rigidity distribution, phi distribution...)
some vector quantities useful for evaluating performance includes cpu time/event, ntracks found/event ...
postscript file containing plots of some of the histos and vectors mentioned above
(where NN is the setting number you've selected)
I have done some evaluation with these macros, Some are on the web for inspection. *** I would really appreciate it if others would also try some evaluation ***. If you would send the results to me, I will sort of keep all the evaluations together.