[G8b_run] Two-Pion Analysis: Response to Ken's Comments

[Ken Livingston]

Hi Volker,
Thank for the reply. I think we should have a meeting this week to 
briefly discuss the submission of this and Mike's note, and any 
supporting documentation.

Cheers,
Ken



On 03/02/2012 08:48 PM, Volker Crede wrote:
> Hi Ken,
>
> here are our responses to your comments. Thanks again for looking into 
> this.
>
> Best wishes
>
>        Chuck & Volker
>
> ========================================================================
>
> Chapter 1.
> p1 It would be useful to label the topologies for reference later. Okay.
>
> p6 What proportion of events have more than one photon and not 
> analyzed. Chuck is still working on adding this information to the 
> analysis note.
>
> p7 Eloss - would be good to use an example plot. What example plot do 
> you have in mind? Well, we can think of something.
>
> p20 Top Timing - I find this confusing. "Once again" - where was it 
> previously used? "Once again" does not add any information here, so we 
> will take it out.
>
> Fig 1.16 Would be useful to see contours. Honestly, we have not seen 
> much improvement over the current picture by using contours. We will 
> keep looking into this and perhaps update this picture later.
>
> Fig 1.17 Why 5% c.l. And in text --- removes "much of the background" 
> how much is "much"? We consider our mass distributions pretty much 
> background-free. I talked to Marco Battaglieri about this and they 
> also assumed background-free distributions in their two-pion 
> analysis. I have attached two pictures showing this, one for the 
> missing-proton mass and one for the missing-pion mass. The 4-vecs used 
> are not kinematically fitted otherwise we would observe spikes for the 
> missing particles. The black solid line is the distribution with all 
> cuts used in the analysis, but without a CL cut for the event. 
> Background is visible, but is almost negligible under the missing-mass 
> peak. The blue line represents the distribution with an additional CL 
> cut of 5%. We loose very little signal, but the background is 
> essentially cut out. The CL cut of 5% is a compromise. Usually a 10% 
> cut is used, but since our background is so small, I think we can 
> afford 5% to retain more good statistics. I will describe this better 
> in the analysis draft.
>
>
> Chapter 2.
> p27 bottom. "Thusly" should be "thus". Okay.
>
> Eq 2.4 some terms not defined. Okay, will add more definitions.
>
> Fig 2.2 is not needed. Since Chuck put quite some effort into this for 
> his dissertation, we prefer to leave it in ;-)
>
> Eq 2.9, 2.10 They should, in general, have different I_0 - eg 
> I_0_para, I_0_perp, or some reference made to how they were 
> normalized. No mention of phi_0 in these eqns. Well, the discussion at 
> this point is quite general and the equations are meant for the ideal 
> case of having no acceptance effects. I will add a reference/comment 
> as to how we normalized the distributions in the analysis. I agree 
> that the definition of beta is missing since we continue to talk about 
> phi-distributions.
>
> 2.4 p34 The vertical offsets result from a systematic error in the 
> scaling of the data. I agree and for this reason, we have used the 
> offsets as a source of the systematic error to account for the 
> normalization issues. The asymmetric distribution for Topology 3 
> (missing proton) in Fig. 2.8 remains strange, though. Since this 
> topology also has other problems, we simply decided not to use it for 
> our final results.
>
> 2.5 p35. Remind us which topologies are which (worth re-writing them 
> here again). Okay.
>
> Fig 2.9 "hinting" in the caption is too vague a word. Okay, we will 
> use "showing" or a similar word.
>
> The different vertical offsets means that the data were not scaled 
> correctly when making asymmetries. This is correct, but there are also 
> statistical effects. The size of the offset is correlated to the 
> statistics for the corresponding distributions. We do not expect an 
> offset of exactly zero. Nevertheless, the effect on the observables is 
> almost negligible.
>
> I don't understand the thing about the detector acceptance dropping 
> out for topology 0, but not topology 5. The diagram with all the X and 
> Ys with arrows doesn't help me! When we form the asymmetry for one 
> particular topology, the acceptance in principle drops out (as we all 
> know). We have done this individually for each of our four topologies 
> to extract the observables and then averaged over our results. This we 
> have called Topology 0. For Topology 5, we have first added all the 
> phi distributions for all the different topologies, then formed the 
> asymmetry, and then extracted the observables. Since the topologies 
> have all slightly different acceptances, these effects do not cancel 
> out. A comparison of Topologies 0 and 5 gives us a feeling for how big 
> these acceptance effects are. We will think of a better way to make 
> this point clear.
>
> p42 2.7.2 Where does it say how the PARA and PERP histograms were 
> normalized? I guess nowhere at this point; it needs to be added of 
> course. Thanks for pointing this out.
>


-- 
=======================================================
Ken Livingston

Dept. of Physics&  Astronomy,        Tel: +44 141 330 6428
University of Glasgow,               Fax: +44 141 330 5889
Glasgow G12 8QQ.
Scotland. UK.
=======================================================