[d2n-analysis-talk] Referee comments on A1n letter

[David Flay]

Hi Diana,

On Fri, Jan 30, 2015 at 6:33 PM, Diana Parno <dparno at uw.edu> wrote:
>
>
> Reviewer 1, item 5 (migration between x bins in BigBite): I’ve already
> talked with David a bit about this, and he remembers doing a small study on
> how electron energy loss affected bin assignment (but will have to dig into
> his notes). I took a very look at the kinematics plots for the 4.74-GeV
> data set and, in the DIS range we’re reporting in this letter, most of the
> counts are fairly central within the x-bin. We see about a 25% drop in
> counts between one edge of the x-bin and the other, with an approximately
> linear falloff in between. My gut says this is not a huge effect given our
> energy resolution but I don’t have numbers to back this up.
>

​Concerning electron energy loss, I looked at this back in June of 2011 for
the elastic tail (pages 5 and 6):

https://hallaweb.jlab.org/wiki/images/2/25/DF_LHRS_6_23_11.pdf​

you can see that this effect looks quite small; however, I never quantified
it any further. Of course, this was for the LHRS data, but the concept
should be the same for BigBite.



> Reviewer 1, item 7 (world fits): David and I have already talked about
> this and it seems the COMPASS data were casualties of last-minute thesis
> chaos. He’s already planning to redo the world fits on the timescale of the
> next week or two.
>

​I have redone the fit to the A1p data, and I don't see a significant
change in the fit; in fact, it's accounted for by my error band (which I
show in the long paper draft).   My impression is that these data won't
significantly affect our results. ​

Reviewer 1, item 12 (including g1p data in the world fit, scaled by some
> reference value of F1): This question seems tougher to me. I am not
> immediately sure how well fixed F1p is in this kinematic range (although it
> must have been measured a lot?) and what kind of systematics we might
> introduce this way. I’m sure David and Zein-Eddine have thought about this …
>

​My thought process in this analysis was to minimize as much model
dependence as possible; if we use our "favorite" model of F1p to construct
g1p/F1p, we would introduce another systematic error here (as I would
generally consider other models in the construction, and see how they all
differ, let alone considering a single model's systematic error), in
addition to my fit of g1p/F1p.  By fitting only those data that are
extractions of g1p/F1p, we don't have to worry about an extra systematic
error.  ​

Reviewer 1, item 13; Reviewer 2, item 5 (constructing PDF ratios from
> different experiments): While I see the reviewer’s argument, this strikes
> me as a lot of work to do on other people’s data and I am not persuaded
> that the results would be very comparable to our data points. At the same
> time I don’t want to discount the other work that’s out there, and the
> caption is already rather long for adding in another explanation of why we
> don’t plot [fill in work here]. And there is the fact that both reviewers
> thought something was clearly missing.
>

​I believe those more recent HERMES results are Delta u/u and Delta d/d
(and anti-quark counterparts); that would be a non-trivial calculation to
construct the ratios we need.  Those HERMES results can be seen here:

http://arxiv.org/pdf/hep-ex/0407032.pdf


Best regards,

David

--------------------------------------------------------------------
David Flay
Post-doctoral Research Associate
University of Massachusetts, Amherst
Department of Physics
Lederle Graduate Research Tower, Rm 423
710 N Pleasant St
Amherst, MA 01003-9305
office phone: (413) 545-0586
--------------------------------------------------------------------
-------------- next part --------------
An HTML attachment was scrubbed...
URL: https://mailman.jlab.org/pipermail/d2n-analysis-talk/attachments/20150202/85ab0218/attachment.html