[Deeppwg] Raw RGA DVCS BSA

[mcontalb at jlab.org]

Dear all,
I apologize but my previous submission(s) failed due to an
unknown reason. The comparison we required among Guillaume
and Sangbaek DVCS BSA results is available at the DPANA
docdb web page

https://clas12-docdb.jlab.org/cgi-bin/DocDB/private/DisplayMeeting?conferenceid=9

on 22 Oct release session (attach to Guillaume talk).

As anticipated at the release meeting, preliminary results
are intended to be a work in progress. As so, the
cross-check should not be taken as finalized yet, but
good enough at the level that Guillaume and Sangbaek
are not going to claim any quantitative statement on
their talk.

Bests, Marco.



On Tue, 27 Oct 2020, Guillaume CHRISTIAENS wrote:

>  Dear Marco,
>
>  Attached are the plots that I have produced for the comparison.
>  I have only compared with Sangbaek values for now.
>  I compare the fit amplitude and the uncertainty on the amplitude :
>  -plot 1: histo_comparisonSG.pdf shows the results of the amplitude and the
>  uncertainty on the amplitude for both Sangbaek and me
>  -plot 2: histo_comparisonSG_ratio.pdf shows the result of the ratio of
>  amplitudes (amplitude from Sangbaek analysis divided by amplitude from my
>  analysis) with uncertainties propagated.
>  -plot 3: histo_comparisonSG_ratioErr.pdf shows the result of the ratio of
>  amplitudes uncertainties (amplitude uncertainty from Sangbaek analysis
>  divided by amplitude uncertainty from my analysis).
>  Bin numbers are shown along the x-axis.
>
>  We can see from the plot 1, a very good agreement on about half of the
bins
>  and some discrepancies for the other half.  From plot 2, we can see that
>  about 2/3 bins have discrepancies <20%. For the other 1/3,
discrepancies are
>  within 35% (except for bin 12 and 24).
>  From plot 3, we can see that discrepancies on the uncertainties are within
>  15% in general and up to 25%.
>
>  I think it is worth noting that most of the points that have discrepancies
>  (3,6,9,12,21,24) are high-xB bins (bin numbers multiple of 3). These bins
>  are where the pi0 cross-section is the highest. The discrepancies are
>  probably coming from differences in the contamination: Sangbaek and I do
>  subtract 2-photons pi0 events to compute raw asymmetries, but we do it
with
>  different pi0 selection cuts, which leads to different contamination.
>  Obviously this is only the case for raw BSA, since for pi0-corrected BSA,
>  remaining contamination should be removed at the stage of 1-photon pi0
>  contamination subtraction.
>
>  Finally, note that bins 11, 12, 23 and 24 are bins for which the amplitude
>  is low, these bins are almost only pi0 contamination, and we do not yet
know
>  if we can extract a good BSA on these bins. It is not a surprise that the
>  comparison shows relatively high discrepancies and high uncertainties on
>  these bins.
>
>  Let me know if you have any questions.
>
>  Best regards,
>
>  Guillaume
>