[Clas12_verystrange] Fwd: FT Timing and PiD taskforce!

[Eugene Pasyuk]

-Eugene

----- Forwarded Message -----
| From: "Raffaella De Vita" <devita at ge.infn.it>
| To: "Derek Glazier" <Derek.Glazier at glasgow.ac.uk>
| Cc: "Lei Guo" <lguo at jlab.org>, "Eugene Pasyuk" <pasyuk at jlab.org>, "Achyut Khanal" <akhanal at jlab.org>, "Raffaella De
| Vita" <Raffaella.Devita at ge.infn.it>
| Sent: Wednesday, June 20, 2018 7:42:59 PM
| Subject: Re: FT Timing and PiD taskforce!
| 
| Hi Derek,
| thanks for compiling the list. I included some comments below.
| Cheers,
| Raffaella
| 
| 
| > **  FT time **
| >
| > - Need to include full swimming to get correct path
| >
| >  (  --related to this but not specific to time, the e- in the FT
| >  have
| > a phi offset due to the solenoid. Perhaps this can be corrected at
| > the
| > same time as the path. Perhaps this also need FTTrack info?)
| This is already done. Angles, both theta and phi, are already
| corrected.
| For now, the cluster position from the calorimeter is used but, when
| tracking reconstruction will be implemented, we should certainly use
| the
| fttrk position.
| >
| > -- Should be different for charged and neutrals
| It is
| >
| > - Need to account for shower profile/depth
| >
| > --Probably this can just be a constant offset (~1cm) which can be
| > determined once the path is corrected.
| Shower depth is already implemented. The depth was tuned based on
| simulations.
| 
| > ** FT time for Start Time**
| >
| >
| > - Should the FT time be used to detemine the start time for
| > time-based
| > tracking?
| >
| > --This is distinct from should the FT time be used to determine the
| > start time for PID. In general it should be for PID as not all
| > reactions
| > will have a start time from the FD and those that do will not be as
| > reliable as the FT if the resolution we have seen for the FT holds
| > up.
| I agree
| 
| >
| >
| > -- For time based tracking,
| >
| > case e- in FD : use FD based start time
| >
| > case pion in FD and FT hit : I attach a plot of the vertex time- RF
| > time
| > , for the FT and for different PID in the FD. The FT time looks
| > better
| > than pion times. Although as I mention the FT time resolution is
| > not so
| > good for run 3432.
| >
| >  --- How often is the FT timing alignment performed, should this be
| > done for every run/week/run period? Probably someone should monitor
| > the
| > resolution for the cooked data files, is this done in CLAS12 mon
| Mon12 can only display raw data while this kind of check requires
| reconstruction. This is part in fact of the calibration suite. So
| far,
| calibrations were performed for 3-4 runs and no big changes were
| seen.
| Once pass0 will be done, we can do monitoring plots for all the runs
| and
| see when we need to recalibrate.
| 
| >  --- In general an algorithm could determine the RF based start
| >  time
| > for FD and FT, if these are the same then everything is fine and
| > the
| > tracking can proceed. If these are different then a choice needs to
| > be made.
| >
| > --- I also attach a plot of my own RF based starttime - EB
| > starttime
| > for background subtracted 2 pion events. I sent this to Raffaella
| > yesterday and include the explanation below +++. But basically it
| > looks
| > like for good pion events the 2 agree 99.9% of the time.
| >
| > --- If I understand this should be equivalent to the plot Achyut
| > showed
| > yesterday, but I did not understand why that plot did not have
| > spikes
| > but rather Gaussian distribution.
| >
| > --- One complication may be random background events which I have
| > subtracted. It may be that a different beam bucket track is
| > selected to
| > give the start time for all tracks if it has the highest momentum.
| >
| > ---- Is this an actual issue? In general the event multiplicity is
| > very
| > high (I think mostly from tracking artefacts rather then real
| > random events)
| >
| >
| > ** PID **
| >
| > --Event Builder already provides a PID, why do we need something
| > else?
| >
| > --- PID not 100% particularly pi/K > 2GeV . For exclusive reactions
| > we
| > can take both hypothesis and use energy and momentum conservation
| > to
| > decide. This is essentially what GLUEX do with their kinematic fit.
| >
| > --- Can try additional information (e.g. DeltaE)  and use
| > correlated
| > information between different tracks, rather than 1 track on its
| > own.
| >
| > --- Can try additional methods like machine learning (I have a
| > couple of
| > summer students looking into this)
| >
| > -- I think we still need to monitor/control what PID is being
| > developed
| > and have (strict) guidlines for validating it before the working
| > group
| > can accept it.
| >
| > Cheers
| >
| > Derek
| >
| > +++ For events which have a start time (~28% do not) I find very
| > little
| > difference between the EB starttime and the one that I find (using
| > FTCAL::clusters::time, and I think this is the time that should go
| > in
| > the EB for the FT rather than FT::particles::time; or
| > REC::scintillator::time for the FD e-). This is when I select
| > exclusive
| > e,p,pi+,pi- topology and perform a background subtraction i.e. it
| > should
| > be pure signal events. The plot is attached which shows the
| > startime
| > difference between my code and the EB for e- in the FT and e- in
| > FD.
| > At the moment I am only looking at events where all particles have
| > a
| > measured time, so I think that ensures there is a valid start time
| > too.
| >
| >
| >
| > [University of Glasgow: The Times Scottish University of the Year
| > 2018]
| > <RFVertexTime3050.pdf>
| > <ExclusiveStartTimeDifference.pdf>
|