[Halld-offline] choice of tagger hit / beam bucket for analysis

[Paul Mattione]

Sean, Simon, etc. should probably comment on the reconstruction, but it may be that the current reconstruction algorithms aren't doing any kind of calibrations/corrections for the time offset yet.  

Introducing tagger hits won't break anything at the analysis (post-REST) stage.  A few comments: 

1) When selecting the RF beam bunch (DEventRFBunch) (used for event start time: PID confidence levels and selecting beam photons), right now the software always assumes that the true beam photon crossed the target midplane at t = 0.  This needs to be changed to give more realistic results, but adding tagger reconstruction won't break anything here.  To do this correctly, we need a DRFTime object that will contain the measured RF time & variance (prior to bunch selection).  

2) When selecting the correct beam photon for particle combinations, users can place a cut on the RF/beam-photon delta-t in their DReaction. So placing this cut in your plugin should reduce the # of combinations for having additional reconstructed photons.  Recommended cut:

locReaction->Set_MaxPhotonRFDeltaT(2.004*0.5); //2.004 ns between beam bunches

- Paul

On Jul 3, 2014, at 5:53 PM, Richard Jones wrote:

> Dear GlueX Offliners,
> 
> I was about to check in a change to hdgeant, but decided to solicit your input on it first since it has the potential of complicating life for those doing analysis studies. I think this complication is needed in preparation for analyzing real data, but it needs to be anticipated and phased in correctly.
> 
> Right now multiple tagger hits, both microscope and fixed array hodoscope hits, are being recorded by hdgeant and saved in the output hddm event record. The tag times are already locked to the accelerator RF, so the values are integer multiples of 2ns. This means that, in principle, anyone doing reconstruction should be looping over all tags present in the event and computing all times measured in the detector (Start Counter, drift times, Bcal and FCal hits, TOF, etc.) relative to the particular tag of reference. Together with the tag hit time, there is also the tag energy recorded in the hit record as well as the tagger channel number, for easy reference.
> 
> However, I believe that essentially no one is doing this. If you just naively assume that the simulated event is timed against the "correct" RF bucket and let t=0 mark the time that the interacting photon bunch crosses through the target midplane, you do get consistent answers at present. As a result, I think we are not making any use of the tagger in the reconstruction. Therefore I propose that the correct beam bucket for the true tag be smeared in the simulation by assigning the event time reference to a random beam bucket near but not fixed to t=0. I was thinking a RMS of +/- 5 beam buckets would be a good estimate for the time resolution of the level one trigger. We could store/fetch this rms trigger time in the ccdb, to make it keep up with improvements in the online.
> 
> What do you think?
> 
> -Richard Jones
> _______________________________________________
> Halld-offline mailing list
> Halld-offline at jlab.org
> https://mailman.jlab.org/mailman/listinfo/halld-offline