[Halld-tagger] Fwd: [EXTERNAL] Fwd: Accidental subtraction

[Richard Jones]

Matt,

It seems you subject the true beam photon to your "pseudo-tagging
> simulation" which is probably just a check on the energy to be sure it will
> hit a tagger element, but the extra beam photons come from data and
> incorporate all real detector effects.
>

Yes, but the efficiency that a given beam photon in the tagged energy
window is actually psuedo-tagged is around 97%. whereas in the actual
detector the tagging probability varies considerably from counter to
counter, and averages around 70% at the beam intensities we are running at
right now.

Here is an interesting study. Repeat the signal MC simulation but randomly
delete some fraction of the true psueo-tags before mcsmear. Then repeat the
analysis as before, and rescale the acceptance by the fraction of preserved
true tags. For example, if I drop 50% of the true tags from the simulation
before mcsmear, and my acceptance comes out just rescaled by this factor of
50% relative to what it was before then my analysis is completely
insensitive to the tagging scheme I am using, at least with regard to the
differential cross section. For a polarization-sensitive analysis,a similar
comparison of angular distributions also probes my sensitivity to
systematic variation in the polarization spectrum coming from accidentals
remaining in my final sample. That would be a good place to start, and help
us to bracket the size of this systematics issue in any analysis.

-Richard Jones

On Thu, Feb 23, 2023 at 9:20 AM Shepherd, Matthew <mashephe at indiana.edu>
wrote:

> *Message sent from a system outside of UConn.*
>
>
>
>
> ---------- Forwarded message ----------
> From: "Shepherd, Matthew" <mashephe at indiana.edu>
> To: Richard Jones <richard.t.jones at uconn.edu>
> Cc: Hall D beam working group <halld-tagger at jlab.org>
> Bcc:
> Date: Thu, 23 Feb 2023 14:20:19 +0000
> Subject: Re: [Halld-tagger] [EXTERNAL] Fwd: Accidental subtraction
>
>
> > On Feb 23, 2023, at 7:43 AM, Richard Jones <richard.t.jones at uconn.edu>
> wrote:
> >
> > Yes, but at nominal GlueX intensity, this cannot be as low as 1% can it?
> Can you look into that and see if that might be based on hdgeant4 Monte
> Carlo where there is no tagger simulation present?
>
> Your calculation seems logical, and yes, I think we can look at this and
> provide some firm numbers.
>
> > All of this can be modeled with a simulation, but the present hdgeant[4]
> simulation is not set up for this. It was designed under the assumption
> that accidentals subtraction would be used.
> >
> > To get quantitative estimates of the systematic errors on our physics
> results from these effects, we need a simulation that includes the real
> tagger and not what is in there now, which I call a "pseudo-tagging
> simulation". This sounds like a lot of work to do correctly, with the
> efficiency of each individual fiber properly accounted for, column overlaps
> which vary at the 10% level along the microscope, etc. We might start off
> by doing bits of it and seeing how dependent the results are on the
> detailed inputs, or whether they are governed largely by just one or a few
> effective parameters.
>
> Can you elaborate on why more detailed simulation is needed?  It is clear
> the true efficiency of the tagger must have complex effects and variation
> between elements, etc...
>
> This is where my knowledge of the exact details of how the simulation is
> implemented gets foggy... and I may completely misunderstand what is being
> done.
>
> My understanding is that we inject extra beam photons into the analysis
> and these photons come from out of time data.  Therefore, these extra
> photons have all the rate dependence, true tagger inefficiencies, etc.
> already built in.  So the probability and energy spectrum of extra beam
> photon hits in the tagger matches reality by construction (also in a rate
> dependent way because of how we manage random noise injection).
>
> It seems you subject the true beam photon to your "pseudo-tagging
> simulation" which is probably just a check on the energy to be sure it will
> hit a tagger element, but the extra beam photons come from data and
> incorporate all real detector effects.
>
> At the end of the analysis (after one choses an algorithm to manage
> accidentals) it seems like what you have is an efficiency corrected event
> yield times the average probability that the true beam photon lands in the
> tagger acceptance -- call this pseudo-tagging efficiency.  I think this
> outcome is is generally independent of the subtraction or best chi^2
> algorithm but there may be differing levels of statistical or systematic
> error in each case (and previously mentioned complications with smearing
> the flux spectrum at high rates).
>
> The key step to go to the cross section is then to use the tagged flux but
> in measuring this flux one also uses the same pseudo-tagging algorithm
> which has the same pseudo-tagging efficiency that is applied to the beam
> photon in the simulation.  This results in a cancellation that yields the
> true cross section.
>
> Do I have an accurate picture of how the simulation is implemented?
>
> Thanks,
>
> Matt
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mailman.jlab.org/pipermail/halld-tagger/attachments/20230223/94dd6e5e/attachment-0001.html>