[Clascomment] OPT-IN:Single and double spin asymmetries for deeply virtual Compton scattering measured with CLAS and a longitudinally polarized proton target

[Silvia Pisano]

Dear Volker,

thanks a lot for your feedback.

While working on your suggestions, let me forward you the exchange we
already had with Sebastian Khun about the C12 BSA plots, and the
conclusions we drew from it.

Best regards,

silvia


----------------
Sebastian's comment:

"Overall a nice paper with very interesting results. I have only one
"serious" concern: You are showing the A_LU results for both 12C and 14NH3
in Fig. 17 and then comment that they are "compatible". However, the
deviation is at 3 sigma, which is certainly not insignificant! Furthermore,
it looks like nearly ALL the 12C data are shifted UP from the 14NH3 ones,
to the extent that I am wondering whether they are even consistent with an
average of zero (as physics says they must be). It is one thing to say that
we do not have precise enough data to be sensitive to medium effects (which
is undoubtedly true). But I would like to see a bit more discussion about
the 12C "anomaly" and, in the discussion of systematic uncertainties
(V.12), an accounting of the uncertainty coming from the fact that we
really don't know A_LU for the 10% "nuclear background" (at least not very
well) from the data alone."

----------------

Our answer:

Your point about the c12 has been raised during our Ad Hoc review. We
checked already at that time the form of the distributions, fitting it with
the addition of a constant term. The results are in

http://www.lnf.infn.it/~pisanos/dvcs/prd/c12/?play=http://www.lnf.infn.it/~pisanos/dvcs/prd/c12/test_bsa_c12_nh3_noden.png

First plot shows the comparison among the two asymmetries, and, even though
the c12 - red points - are usually bigger than the nh3 ones - blue points,
they are compatible for almost all of the bins. The middle and right plots
show the fit to the c12 (middle), nh3 (right) bsa with the standard bsa
function, where p0 is the constant term and p1 the sinphi modulation. As
you can see, p0 for c12 is compatible with 0 in 1.5 sigma. The other
modulations are perfectly compatible.

Bear in mind also the fact that these are raw asymmetries, not pi0
subtracted, and such a subctraction cannot be performed on c12 due to the
even more limited statistics on pi0 events.

Given this statistical limitation, it is hard to draw any conclusion, or
also think of implementing any kind of systematics from this. The BSA on
c12 that you see is integrated over the whole part B, and trying to
estimate a systematics from it would be a sort of Ansatz, not including
possible dependences etc. Furthermore, it would be estimated on a different
observable, i.e. the raw asymmetry, instead than on the real, subtracted
one.

Your comment, in any case, pushed us to go back and investigate more
throughly the C12 analysis. We found out that due to the limited C12
statistics the BSA is very sensitive to the statistical fluctuation coming
from the variation of the exclusivity cuts. In order to be sure that there
is no sistematic effects but only a statistical one, we compare the BSA on
c12 and on nh3 in the two extreme cases of the exclusivity cuts, i.e. the
one corresponding to the 2.5 sigma and the one relative to the 3.5 sigma
from the central value of the cut estimated following the procedure
described in the analysis note.

At the following links you can see the results relative to these two
scenarios:

2.5 sigma case:
http://www.lnf.infn.it/~pisanos/dvcs/prd/c12/?play=http://www.lnf.infn.it/~pisanos/dvcs/prd/c12/test_bsa_c12_nh3_2.5sigma.png

3.5 sigma case:
http://www.lnf.infn.it/~pisanos/dvcs/prd/c12/?play=http://www.lnf.infn.it/~pisanos/dvcs/prd/c12/test_bsa_c12_nh3_3.5sigma.png

together with the results on the fit of the bsa on c12. As you can see, the
asymmetry are perfectly compatible, and no shift is observed in the c12 (p0
in the fit box).

Since these are the two extreme cases and not effects are seen, and being
the final contamination - dilution factor - around 8%, we concluded that
not further systematics was needed.

This said, however, in the systematics studies we did, the one relative to
the variation of the exclusivity cuts takes somehow into account possible
effects from nuclear background, since, as you can see from Fig. 15,
varying the cuts it varies the c12 contamination, and should then include
any sensitivity in the BSA emerging from an increased nuclear backgroud.

We realize now that that kind of plots in the final paper could be
misleading,  and therefore we have decided to remove it and to add a little
more explanation in the text. Here is a summary of our reasons to remove
the figure:

1) That figure is supposed to justify why we present the BSA with this
data, despite the fact that there is some background from the N14 and
therefore it is not the standard free proton BSA. We think that this is not
a concern since our background is just 8% and we are applying tight
exclusivity cuts which ultimately select N14 protons at quasi-free
kinematics;

2) It would be a concern to have such a background if our results were
disagreeing with the BSA measured on free protons (FX et al) but Figure 19
shows that we are consistent;

3) C12 and in particular C12 BSA is not used anywhere in extracting the
results. We only used the total counts of C12 to get the dilution factor.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: https://mailman.jlab.org/pipermail/clascomment/attachments/20141222/e6450126/attachment.html