[Jef] Dave comments on Bishnu's slides

[Dave Mack]

Justin,

In light of Simon's data results today, this is an updated version of 
some suggestions I sent out last week to you and a few others.

1. The slides need to be numbered for easy reference.

2. We need an intro slide from Justin overviewing the simulation 
ingredients

3. Cuts:

    i. Add a new fiducial cut P_proton > 300 GeV/c.
(Anything below that is dubious based on Simon's Figure 37 in the JEF 
proposal. This should remove some funny non-gaussian outliers in the 
Phi  and MM2 distributions for True events.)

     ii. Change the too-tight energy cut to +-0.36 GeV.
(Let's aim for +-2sigma cuts - naively pretending they are independent - 
until the kin fit is working.)

    iii. Tighten the MM2 cut from [-0.05,+0.025] to something more 
consistent with a 2sigma cut (though possibly still asymmetric).
(Plot the MM2 on a scale that lets us judge how to set this cut.  
Simon's data plots naively suggest the cut could be tightened to a MUCH 
more restrictive +-0.005.  But the average beam energy in his data is 
lower than in the simulation, and some of what I think is background 
might be Lorentzian tails on the signal. So simulation should guide us 
here.)

   iv. Cuts for discussion: Do we need a fiducial cut on target vertex 
to remove window backgrounds?  Is M_2gamma reconstructed using the 
inferred Z location of the event in the target, or simply the target 
center?

4. Plots

i. Change the M_2gamma range to say 1.5 GeV.
(We don't want to miss anything in the higher mass regions. And we need 
this to get the whole story on the eta'/phi region.)

ii. We can't easily see an extra pion on the MM2 plots.  We might 
consider zooming the range of  MM2 [-0.3,+0.3].
(In units of the pion mass squared, that's already +-15 so plenty.)

iii. Since the fiducial cut on beam energy, plus the cut on unused 
energy, had such a big effect, we can't see how the remaining cuts 
reduce backgrounds in the "Yield plot after various cuts". Can you make 
two versions, one on a linear scale, and one on a log scale?
(This would also be a good time to fix the spelling of hyeildfalse and 
hyeildtrue.)


Next steps:

1. Get more quantitative:

i. to fulfill Justin's initiative on an early physics measurement: we 
need to estimate the reconstruction efficiency, estimate counts per day 
per GeV wide beam energy bin, make sure computer and electronic deadtime 
are reliably monitored, verify tracking efficiency for recoil proton, 
what happens to this analysis if a lower calorimeter threshold is used, 
etc?

ii. to make sure we can prioritize our backgrounds: make a table of the 
5 largest backgrounds in the pi0, eta, and eta' regions, listing the 
percentage for each.

2. What do we have to fix to do kinematic fits? GlueX has mediocre 
resolution and a high energy beam. This makes the isolation of exclusive 
processes a somewhat ill-posed problem without help. Bishnu's plots 
already suggest the omega background in the eta mass region is resistant 
to cuts. With a kinematic fit, we might be able to dramatically reduce 
that background.

Dave

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mailman.jlab.org/pipermail/jef/attachments/20150914/1c7a9076/attachment.html>