[Clascomment] CLAS paper color transparency

[GUIDAL Michel (57321)]

Hi Kawtar,

Thanks for your rapid reply to my comment. Let me reply in turn:

>> I have one main question on your CT analysis. I wonder why you have rho
>> peaks (i.e. M_pi+pi- spectra) so "clean" (your fig.4), especially 
>> compared to the ones that were obtained in the analysis with a "simple" 
>> proton target (e1-6 analysis, published in Eur.Phys.J.A39:5-31,2009; 
>> look at particularly at Figs. 4 and 11).

> Answer: We had this discussion more than two years ago. We can not use a 
> proton detection for a nuclear target because protons can originate from 
> many processes.

More than the cuts in W and t which are, to my eyes, very similar
between "your" analysis and "ours", this different detection
"topology" (i.e. you detect pi+ and pi- and we detect p and pi+) is 
probably the main origin of the differences between "your" and "our" 
M_pi+pi- spectra. It is actually a bit disturbing/intriguing that the 
shape of the M_pi+pi- spectrum changes so much according to the detection 
topology.
To be fully convinced of what is going on, it would be interesting to 
show/prove with simulations, with all the channels contributing
to the pi+pi- final state (and there are many, I already quoted a few
of them) how the M_Pi+pi- spectrum "evolves" according to the topology, in 
particular that almost all "non-rho0" pi+pi- pairs disappear when you
detect only pi+ and pi-. I could do that. You could also.

> However, you can apply our cuts to your data and 
> convince yourself. For more details about different cuts effects, you are
> invited to see figure 19 at (P. 39) of version 5 of our analysis note.

>> It looks to me that the cuts you used are not so different from the ones 
>> used in the e1-6 analysis (for instance, I think that you use W>2 while 
>> W>1.8 was used in e1-6; you also make a cut to select low t's but these 
>> low t's are also part of the e1-6 analysis and since low t's dominate the 
>> cross section, the integrated spectra (over t) shoud largely reflect the 
>> low t spectra).

> Answer: It is not true that low t dominates the kinematical region of CLAS
> acceptance.

What you say is not correct in "our" topology, i.e. detecting p and pi+: 
the acceptance is almost flat as a function of t. You would have to
look at Fig.4.23 (p.61) of "our" analysis note which can be found
in "http://www.jlab.org/~guidal/rho" (which is unfortunately not
accessible right now due to the cyber attack... but you might
have already a copy somewhere). In any case, M_pi+pi- spectra are 
corrected for acceptance and so, small or large acceptance (as long as
it is not 0) shouldn't affect the shape of the M_pi+pi- spectra.

>> In the e1-6 analysis, non-resonant 2-pion contributions (which have plenty 
>> of sources: Delta++pi-, Delta0pi-, incoherent ppi+pi-, etc...) under 
>> the rho0 peak could sometimes reach 50%. And actually a systematic 
>> uncertainty of ~20% was associated with this "background" subtraction 
>> since actually no existing model on the market coud satisfactorily 
>> reproduce all the dependences (xB, Q2, t,...) of these M_pi+pi- 
>> spectra. So, my question is why the situation (shape, normalization,... 
>> of the 2-pion spectra) is so different in your case? And, as a 
>> corrolary, how can you be so confident in your non-resonant 2-pion 
>> contribution subtraction under the rho0 peak since I am not aware of 
>> any fully reliable model in this energy region for this
>> process (and what systematic error did you attribute to it)?

> Answer: Here, we are not measuring a cross section but ratio of yields. We 
> are most sensitive to changes in background's shapes between deuterium 
> and solid targets. Our simulations reproduce well these shapes, in 
> addition to the fact that the contributions from background in our 
> kinematical range are small. For more details on background subtraction 
> and its systematics, please review sections (7.2) and (8.3) that are
> respectively at (p. 45) and (p. 55) of version 5 of our analysis note.

Fitting well only a 1-dim M_pi+pi- spectra is not a proof that
you the do the correct thing. There are a million of ways to do so
with different ingredients (and only one correct). You need to fit 
simultaneously several other distributions to begin to have confidence in 
the relative weight of the different "ingredients" you enter into the fit.
For instance, was it the same weights for the different contributions
that you used to fit your Fig.24 and Fig.20 ? (Fig.20 was used for
acceptance corrections and Fig.25 for background subtraction, so
for different purposes, and so the ingredients of the MC can be 
different).
Looking at Fig.34, shouldn't I conclude that the ratio between
D and Fe can enormously change according to the parametrization
you choose for the rho ?

To be clear, I don't want to be agressive with my questions
and don't mean to put your results in doubt. I just want
to understand why things are so different between "our" analysis
and "yours", this might be of use for everyone in the future (I think I 
already made some progress by identifying the main source as probably the 
detection topology, more than any cuts).

Amities,

Michel

***
Michel GUIDAL
Institut de Physique Nucleaire
Bat 100 - M052
91406 ORSAY Cedex
Tel: (33) 01 69 15 73 21
Fax: (33) 01 69 15 64 70
E-mail: guidal at ipno.in2p3.fr
***