[Clas_hadron] DNP presentation for Mini-Symposium

[Stuart Fegan]

Hi Dave,

With the promise of so much new data for g8, whose effect on the results 
was yet to be determined as statistical only, and with my g9 analysis in 
such a preliminary state at the time of making those plots for my 
Baryons talk, I felt it was safer to show the older g8 points in Glasgow 
last summer as all I'm demonstrating is that consistent measurements of 
'S' are possible using the FROST data, even with all the bound nucleon 
considerations.  I was also under the (perhaps wrong) impression that 
the g8 points I was showing were not directly from Craig's 2007 thesis, 
but an intermediate state of the analysis as it stood in 2011 when I was 
writing my own thesis (I think just before the final photon polarisation 
tables were available).

I'm currently working a few other projects, intending to return to the 
FROST analysis a little nearer the next collaboration meeting, so the 
DNP came a little too soon to get anything new out for g9a, let alone 
update the comparison points.  I've already offered my apologies to 
Natalie for this, and I'll be in touch shortly about obtaining more 
up-to-date g8 results for future presentations.

I see no harm in continuing the analysis on this track, particularly as 
any determination of the sign of the G observable will offer invaluable 
constraint to Stef's likelihood function, which last I saw had a double 
peak for G, one positive, one negative.  Not only that, but with the 
cleaner hyperon event samples now available, and careful normalisation 
of the data sets, it may even be possible to go beyond G and extract (or 
integrate out) some or all of the Lambda recoil observables, although 
that's one or two more CLAS meetings down the road!  There's also my own 
desire to produce a stand-alone g9a note on the KL and KS analyses as 
soon as is practical, into which a section on the maximum likelihood 
extraction can be inserted, like the most recent version I saw of the g8 
note, which can compliment any analysis note Stef may want to produce 
when her own thesis is complete.

Hope all is well in Glasgow,
Stuart

On 21/10/2013 19:25, David Ireland wrote:
> Hi Stuart,
>
> Andy is right about the g8 results. Please do not compare to the 
> results from Craig's thesis. The results I showed are already part of 
> the review process - Craig's never made it that far.
>
> We still await the final re-cooked g8 data, but it is very likely that 
> we will have a factor of 2.5 times more data.
>
> Also, and I cannot stress this more strongly, you *must* extract all 
> the observables you are sensitive to at the same time. If you do not, 
> your results may be misleading. Stef already has a likelihood function 
> that can do the job. The best thing to concentrate on for the moment 
> would be to clean up the event sample as much as possible, and work on 
> calculating the relative effects of background and carbon contamination.
>
> Cheers,
>
> Dave
>
>
>
>
>
>
> On 21/10/2013 10:36, Stuart Fegan wrote:
>> Andy,
>>
>> Yeah, those higher bins were pretty limited by the new event selection
>> leaving next to no statistics to determine carbon scaling via the same
>> method as the thesis analysis.  I've a few tricks up my sleeve though,
>> and hope to have a new measurement of the beam polarisation observable
>> completed in the next week or so.
>>
>> My gut feeling is that S shouldn't be too greatly affected by the Lambda
>> recoil, partly informed by the hazy memory of not seeing much difference
>> when writing the observable extraction macro.  Those diagnostic plots
>> were never properly analysed for bound nucleon background and photon
>> polarisation, so until I have a proper look, it remains just a feeling,
>> which may be very wrong.  G on the other hand, I expect to be
>> particularly prone to Lambda recoil effects, and will need a bit of
>> careful thought to account for this in the extraction.  The cleaner
>> event samples now available may help with this, as the effect of the
>> bound nucleon background should be better accounted for, making it
>> easier to see what's going on.
>>
>> Cheers,
>> Stuart
>>
>> On 10/19/2013 10:56 PM, A.M. Sandorfi wrote:
>>> Hi Stuart,
>>>
>>> Thanks for the qualifications. I too could not view your lower 
>>> energy bins
>>> for S in Natalie's file. Looking at your Baryon'13 slides, I agree 
>>> that at
>>> the lower beam energies, your analysis is closer to g8. Nonetheless, 
>>> your
>>> higher energy g9a results still seem shifted from g8, at least from the
>>> version of g8 you are apparently plotting. Dave Ireland showed what 
>>> I think
>>> is the current g8b version at the Spectroscopy Working Group meeting 
>>> last
>>> February. That's what people should be comparing to.
>>>
>>> Slide 19 of your Baryons'13 presentation, which appears to be slide 
>>> 12 in
>>> Natalie's file, says you averaged the K+L beam asymmetry over the 
>>> two target
>>> polarization settings. It would be interesting to see separate plots 
>>> of your
>>> S for +z and -z target polarizations. That would give some clue as to
>>> whether or not an incomplete integration over the Lambda recoil is 
>>> giving
>>> problems.
>>>
>>> Andy
>>>
>>>
>>>
>>> On 10/18/13 7:31 PM, "Stuart Fegan"<sfegan at jlab.org> wrote:
>>>
>>>> Andy, Natalie, et. al.,
>>>>
>>>> Given that the comments relating to my analysis (slides 12 and 13) 
>>>> take
>>>> up a significant portion of this last message, I suppose I better step
>>>> in and defend them.  Although I broadly agree with the arguments
>>>> regarding the Lambda recoil polarisation, there are a few things I
>>>> should point out regarding my FROST results.
>>>>
>>>> 1) The low energy bins of the 'S' measurement show very good agreement
>>>> with the g8b data shown, in my own humble opinion. However, these 
>>>> plots
>>>> seem to be hiding under the higher energy bins on slide 12 of 
>>>> Natalie's
>>>> pdf, and I can't get at them in my non-proprietary pdf viewer.  
>>>> They're
>>>> from my Baryons2013 slides, and these are available here (slides 19 
>>>> [low
>>>> energy] and 20 [high energy]);
>>>> http://nuclear.gla.ac.uk/~stuartf/talks/SFegan_Baryons2013.pdf. The 
>>>> main
>>>> problem with the higher energy bins relates to the methods used to
>>>> account for the nature of the FROST target, which is pretty far 
>>>> removed
>>>> from the ideal free proton target, and I'll come to that in a moment.
>>>>
>>>> 2) The g8b results shown for comparison with my 'S' measurements are
>>>> themselves preliminary, and pre-date the g8b analysis currently under
>>>> review.  The old g8b results were all I had available when I produced
>>>> these plots for Baryons, but I should have this rectified by the next
>>>> collaboration meeting, where I intend to present a further update 
>>>> to my
>>>> FROST work.
>>>>
>>>> 3) These results were the first major update to the analysis since my
>>>> thesis, coming after the availability of detailed photon polarisation
>>>> tables, and a bit of work harmonising the particle and channel
>>>> identification with the aforementioned g8b review. With very little 
>>>> new
>>>> work done on the observable extraction side of things, these interim
>>>> results demonstrate a need to revise the rest of the analysis, whose
>>>> issues were obscured until this recent availability of a much cleaner
>>>> hyperon event sample, and smaller systematic errors from photon beam
>>>> polarisation.
>>>>
>>>> 4) The nature of the FROST target material means a significant 
>>>> quantity
>>>> of bound nucleon events contribute to the value of "P_gamma S" 
>>>> extracted
>>>> via the binned asymmetry method (where P_gamma is, of course, photon
>>>> polarisation).  Accounting for this contribution has been one of 
>>>> the key
>>>> difficulties in analysing data from this target, and (see point 3 
>>>> above)
>>>> the methods of accounting for this, were yet to be re-examined in any
>>>> detail when those plots were produced (although the values relating to
>>>> this effect were recalculated using the new event samples as described
>>>> in chapter 6.3 of my thesis).
>>>>
>>>> 5) I think you can guess by now that the source of the "new analysis"
>>>> was me.  The results differ, but only because of everything I have
>>>> stated above.  The plots were produced by running the new hyperon 
>>>> event
>>>> samples through the same ROOT macro I used for my thesis, with the new
>>>> photon polarisation tables and recalculated estimates of the dilution
>>>> effect due to bound nucleon events.
>>>>
>>>> So that's about the status of my side of things, hope that's provided
>>>> some useful follow-up information should any questions come up about
>>>> this during the talk.  As I say, I'll be presenting a further 
>>>> update to
>>>> this work in November at the next Hadron Spectroscopy meeting.
>>>>
>>>> Cheers,
>>>> Stuart
>>>>
>>>> On 18/10/2013 18:06, A.M. Sandorfi wrote:
>>>>> Hi Natalie,
>>>>>
>>>>> You have an interesting talk. I have a few comments beyond Volker's:
>>>>>
>>>>> Slide 4: I see you still love the historical table, even though it's
>>>>> incomplete.
>>>>>
>>>>> Slide 5: The wording of your footnote "^" is not correct. You can 
>>>>> point out
>>>>> that the measurements denoted with a "^" were obtained in
>>>>> double-polarization experiments, which can be more prone to 
>>>>> systematics than
>>>>> the other way of measuring the same matrix element, with a
>>>>> single-polarization experiment. But they are both measurements of 
>>>>> the square
>>>>> of a reduced matrix element. The designation of "direct" has no 
>>>>> meaning.
>>>>>
>>>>> Slides 9 through 14: when you compare data from multiple sources, 
>>>>> plotted
>>>>> with different symbols, you need to include a legend on each slide 
>>>>> so the
>>>>> audience can quickly understand what you are comparing. Some times 
>>>>> you have
>>>>> this in words somewhere on the slide, which is not as clear, but 
>>>>> often there
>>>>> is no hint at all.
>>>>>
>>>>> Slide 12: Although the errors are larger, the red g9a points are 
>>>>> generally
>>>>> in poor agreement with g8. There is a likely reason for this. Your 
>>>>> first
>>>>> bullet states that you have averaged +z and -z target 
>>>>> polarizations. An
>>>>> experimental asymmetry measurement constructed as 
>>>>> (par-perp)/(par+perp)
>>>>> contains 11 different observables, one of which is "S", the beam 
>>>>> asymmetry
>>>>> you are after. All of the others are multiplied by a component of 
>>>>> the Lambda
>>>>> recoil polarization, for example Pt(z)*Pr(z')*Lz'. For either 
>>>>> parallel or
>>>>> perp photon polarizations, when you flip the sign of the target 
>>>>> polarization
>>>>> Pt(z), the signs of the two recoil components Pr(x') and Pr(z') 
>>>>> also flip.
>>>>> So when you average the yields from +z and -z target 
>>>>> polarizations, terms
>>>>> like the above one in Lz' add; they DO NOT CANCEL OUT. The
>>>>> (par-perp)/(par+perp) ratio will reduce to "S" only if the Lambda 
>>>>> =>  pi-p
>>>>> decays are completely integrated so that the recoil information,
>>>>> Pr(x',y',z'), is thrown away. That depends on modeling detector 
>>>>> acceptances,
>>>>> holes in CLAS, etc. and certainly has significant systematic 
>>>>> issues. If you
>>>>> want to show slide 12, you must qualify the first bullet with a 
>>>>> statement
>>>>> about the additional systematics involved in averaging over the 
>>>>> recoil.
>>>>> (If you extracted "S" by fitting cos(2phi) moments, then it's a 
>>>>> little more
>>>>> complicated. The unwanted terms can still drop out, but only if the
>>>>> averaging over the Lambda decays is done just right.)
>>>>>
>>>>> Slide 13: this is presumably a new analysis, since the results 
>>>>> differ from
>>>>> figure 7.8 of Fegan's thesis. The source should be quoted or 
>>>>> referenced
>>>>> somehow. Depending upon how this was done, it is susceptible to 
>>>>> the same
>>>>> problems as I mentioned above. Changing the sign of the target 
>>>>> polarization
>>>>> changes the recoil polarization components, so that the results 
>>>>> for G depend
>>>>> critically on a careful average over the pi-p decays so as to 
>>>>> intentionally
>>>>> loose the recoil information.
>>>>>
>>>>> Slide 14: this is missing the g8 results for T. These should also 
>>>>> be plotted
>>>>> here. According to slide 2, you are summarizing both. I realize 
>>>>> that they
>>>>> agree with GRAAL and disagree with your g9a analyses for many 
>>>>> energy bins.
>>>>> That I suspect has to do with the above recoil issue. The Lambda 
>>>>> decay has
>>>>> to be detected to pull the channel out of the background. But once 
>>>>> detected,
>>>>> the recoil information is in the data stream, whether or not one 
>>>>> chooses to
>>>>> look. Once in hand, throwing it away becomes tricky. (Ironically, 
>>>>> there may
>>>>> be less uncertainty in analyzing the recoil and using it, than 
>>>>> there is in
>>>>> trying to throw it ALL away.) In any case, these issues should be 
>>>>> aired.
>>>>>
>>>>> Andy
>>>>>
>>>>>
>>>>>
>>>>> On 10/17/13 7:19 PM, "natalie at jlab.org"<natalie at jlab.org>  wrote:
>>>>>
>>>>>> Hi,
>>>>>> I have linked my presentation for next week's mini-symposium at 
>>>>>> the DNP
>>>>>> meeting on Thursday. I was asked to give an overview talk of kaon
>>>>>> photoproduction at CLAS. Please let me know if you have any 
>>>>>> comments or
>>>>>> suggestions.
>>>>>>
>>>>>> http://www.jlab.org/Hall-B/secure/g9/natalie/dnp_minisym_kaonphoto_walford.p 
>>>>>>
>>>>>> df
>>>>>>
>>>>>> Thanks,
>>>>>> Natalie
>>>>>>
>>>>>> _______________________________________________
>>>>>> Clas_hadron mailing list
>>>>>> Clas_hadron at jlab.org
>>>>>> https://mailman.jlab.org/mailman/listinfo/clas_hadron
>>>>> _______________________________________________
>>>>> Clas_hadron mailing list
>>>>> Clas_hadron at jlab.org
>>>>> https://mailman.jlab.org/mailman/listinfo/clas_hadron
>>
>>
>> -- 
>> Dr Stuart Fegan
>> Postdoctoral Researcher
>> INFN-Genova
>> Via Dodecaneso, 33 - 16146, Genova, Italia
>>
>> E-mail: sfegan at jlab.org
>>
>> _______________________________________________
>> Clas_hadron mailing list
>> Clas_hadron at jlab.org
>> https://mailman.jlab.org/mailman/listinfo/clas_hadron
>>