[Eg1_run] INclusive EG1 meeting this Wednesday...

[kuhn]

We meet July 11 at 11:00 a.m. in F226. Main point of order: Response to 
PWG, long p paper. Here are the call-in instructions:

Call-in instructions:

 1. First dial the conference call facility:
     1. In US, participant dials the toll free number: 866-740-1260
     2. International participant dials: 303-248-0285
 2. Then enter the passcode*8156718*, followed by the # sign



I also enclose a summary of where I think we stand (previously sent to 
the "editorial board" only):

> I wanted to give you an update and some thoughts about where we are 
> with the long proton paper:
>
> 1) I changed the discussion of the CC matching cuts in the paper by 
> removing large swaths of it - I agree with the PWG committee that this 
> topic is too technical for a general audience (but I do believe we 
> need to document what we did in the analysis note, so it shouldn't be 
> removed from there).
>
> 2) I think I'm pretty much done with the paper. You can find the 
> newest version and all files at
> http://www.jlab.org/Hall-B/secure/eg1/EG2000/Kuhn/LongPpaper/
> In particular, I copied the comments from the PWG committee regarding 
> the paper and my answers ("PaperComments.txt"). Our biggest remaining 
> task on this front, before going off to the AHC, is fixing the figures 
> listed in my file "FigsToBeFixed.tex".
>
> 3) Of course, this (#2) will be Rob's next task. For now, it is most 
> urgent that we complete our 3rd round of responses to the PWG 
> committee. As far as I know, Rob has answered most of their comments 
> which are listed at
>
> http://clasweb.jlab.org/rungroups/eg1-dvcs/wiki/index.php/Review_eg1b_note#Comments_to_paper_draft 
>
>
> (please post your response to the web, Rob!) with the exception of the 
> pesky "t-test question" (first comment on Chapter 7). Rob promised to 
> give us an update on his findings with regard to this, but meanwhile I 
> had a few more thoughts that I would like to pass along - let me know 
> what y'all think:
>
> 1) The first "suspicious" t-test result (see Table 7.1, p.116 in the 
> analysis note - Rob, please post this one as well!) is for the 
> inbending 1.6 GeV, pos. target polarization data - a systematic 
> difference between the results for the 2 HWP settings of about 15% of 
> the average statistical error. One can even see a hint of this in Fig. 
> 7.1 p. 114 - the green data points (HWP-) tend to be a bit further 
> away from zero than the the yellow ones (HWP+).
> I can't believe that something changed "systematically" just by 
> changing the HWP (the difference is much too large for a parity 
> violating effect ; so, I assume that what really happened was just the 
> ELAPSING TIME! Meaning, while we use a constant value for target*beam 
> polarization (PbPt - Rob, can you check that this is correct???) for 
> the WHOLE data set (over both HWP states), in reality of course the 
> target polarization changes over time. This could explain why the 
> asymmetries in one case are a little larger (again, ONLY 15% of the 
> average statistical error) than in the other case. If this is correct, 
> than the t-test discrepancy is NOT a systematical error at all - 
> precisely by averaging over both HWP states (for both the data AND the 
> PbPt), we get the right results. So, if Rob (or anyone else - does 
> anyone have the history of Pt from NMR?) can somehow corroborate at 
> least some part of this argument, we should simply reproduce it as 
> response to the PWG committee.
>
> 2) The other suspicious t-test also involves the 1.6 (and 1.7) GeV 
> data (see table 7.4 p.119): The total 1.6 GeV data set (averaged over 
> all target polarizations AND all torus polarities) is once again 
> incompatible (by about 16% of the average statistical error) with the 
> 1.7 GeV outbending (only) data set (also averaged over 2 target 
> polarizations).
> To explain this, we should first note that the 1.6 GeV outbending data 
> are statistically insignificant, so we are really comparing 1.6 GeV 
> inbending with 1.7 GeV outbending, which could mean that the average 
> Q^2 for any 2 formally identical bins that are being combined might in 
> fact be slightly different. Furthermore, there may be a small change 
> in eta and therefore in the contribution from eta*A2 even going from 
> 1.6 to 1.7 GeV (maybe Rob can do a quick numerical check on how big an 
> effect this could be). Finally, it is important to realize that these 
> 2 data sets have by far the highest statistical precision, so 16% of 
> the average statistical error is clearly a small effect in absolute 
> value. I think it would be very helpful if Rob could calculate what 
> the average ratio for the KNOWN systematical error divided by the 
> statistical error comes out to be for the 1.7 GeV data set (say) - if 
> it is more than 0.16, we can claim that this t-test difference is 
> already accounted for by our quoted systematic error. If not, I would 
> have to come back to my suspicion that we are simply too optimistic 
> about our knowledge of PbPt -- maybe Rob can figure out by how much we 
> would have to increase the quoted systematic error on that to match 
> the observed discrepancy of 0.16*average statistical error.
>
> Hopefully, all of this can be done by our next meeting next week 
> (after which Rob will be moving back to VA and probably will be out of 
> commission for a while).

-------------- next part --------------
An HTML attachment was scrubbed...
URL: https://mailman.jlab.org/pipermail/eg1_run/attachments/20120709/1d366bc8/attachment.html