[Sane-analysis] Reminder of Sane Meeting at 3:30pmToday Wed Oct 8th in f226

[Whitney R. Armstrong]

Hi Oscar,

To continue on my early results from simulating with your pion cross section 
fit...

I am noticing that with the exponential fit found in Fersch's thesis I can get 
very nice results compared to simulation if I assume that instead of the ratio 
(pairs/total) that the fit was to the ratio (pairs/DIS). It makes more sense to 
fit the data with an exponential modeling the (pairs/DIS) ratio because there 
is no upper bound on it, unlike the ratio (pairs/total) <=1. 

I think this works because the measurements for the most part took place 
kinematically where the DIS rates dominate. That is, the contribution to the 
denominator from pairs can be ignored and the ratio works out that
R(pairs/total) ~ r(pairs/DIS)

Then just by calculating R = r/(r+1) the ratio agrees very well with data.  You 
can see this result as the green line. 

http://quarks.temple.edu/~whit/SANE/analysis_main/results/detectors/cherenkov/simulated_cherenkov3_952.png

The only thing that seems to be missing is the factor of 2 for pairs produced 
at the target.  

Cheers,
Whit


On Thu, Oct 09, 2014 at 04:07:48PM -0400, O. A. Rondon wrote:
>Hi Whit,
>
>Some comments follow.
>
>Cheers,
>
>Oscar
>
>Whitney R. Armstrong wrote:
>> Hi Oscar,
>>
>> I have added comments below.
>> On Thu, Oct 09, 2014 at 12:15:50AM -0400, O. A. Rondon wrote:
>>> Hi,
>>>
>>> Here are the items we discussed at the meeting that I said I would
>>> look up:
>>>
>>> - question of trend of d2 with E' or x.
>>>
>>> I see a similar trend as Whit but for the SLAC data vs lower x limit
>>> in the d2 integral, not for SANE. See plots on page 5 here (plots on P.
>>> 4 are the same, except they have a typo (fixed for those on p. 5,
>>> changes things a bit)
>>> https://userweb.jlab.org/~rondon/analysis/asym/world/A180_Aperp-q2-x_d2-xmin.pdf
>>>
>>
>> To be clear, when you say "not for SANE", you mean "not for Hovannes's
>> SANE analysis".
>>
>
>Yes, those are the data we have been showing.
>
>>> The low x and min E' limits are related, but not identical.
>>
>> It is *very* interesting that I am getting almost exactly the same x
>> dependence as E155x shown on page 5. (for my results see
>> http://quarks.temple.edu/~whit/SANE/analysis_main/results/d2p_vs_cut_2.png)
>> The x-axis shows increasing E' cut starting at 800 MeV in steps of 100MeV.
>> Mapping the E' lower limit to x, roughly corresponds to 0.2 to 0.4.
>>
>
>That is one thing I thought about, too. Namely, d2(low x) >0 corresponds
>to a region dominated by g1.
>
>On p. 6 of my d2 report, top plot, I show that it makes a difference
>whether one uses E155 g1 calculated at constant Q^2= 5 GeV^2 vs
>calculated at the Q^2 of each x bin. But there is little Q^2 dependence
>if g1 is calculated from PDF's.
>
>And the bottom plot shows that at middle x, x^2 g2bar is very sensitive
>to what model of g1 is used.
>
>> The agreement between my analysis and E155x highlights the need for the
>> cherenkov window cut to remove tracker pairs.
>>
>>>
>>> For SANE d2 vs min E', I see the same trend we saw at the meeting, shown
>>> above on p. 7, and with different g1 inputs on the top plot of page 5
>>> (or 4)
>>
>> I believe this to be a consequence of Hovannes's analysis not correctly
>> treating the background.
>>
>
>I'm not sure, because at E' = 1.5 GeV, where there is little background
>left, d2(Hovhannes) ~-.004 is many sigmas away from your Nachtmann d2(95
>or 96), which is zero within errors. Both d2's should be converging to
>the same value at this E'.
>
>Also, I like your small error bars, but they are about four times
>smaller than the ones I show. They are total errors, but the stat errors
>are 90% of the bars. I don't think the statistics are that different
>between your analysis and Hovhannes. Since he no ADC cut he actually
>should have more events.
>
>>>
>>> - question of CLAS e+/e- fit
>>>
>>> This is discussed in great detail in my wiki page
>>> https://hallcweb.jlab.org/experiments/sane/wiki/index.php/Scaling_fits_to_Hall_C_positron_data
>>>
>>>
>>> Summaries of e+/e- models that I've looked into (saves actually reading
>>> the wiki page ;-) ), including the direct fits to Hall C pairs data for
>>> 5.9 GeV (top left), Hovhannes Wiser pi0 simulation (top right), and CLAS
>>> (Vipuli) fit (bottom left), are here
>>> https://userweb.jlab.org/~rondon/analysis/asym/pairs_models.pdf
>>>
>>> The plain CLAS fit is the blue curve (pairs for just CLAS target). The
>>> red curve is my expectation of what BigCal would see (includes tracker).
>>>
>>> The Hall C fits for 4.7 GeV data are here
>>> https://userweb.jlab.org/~rondon/analysis/asym/pairs_models_5_9-4_7.pdf
>>>
>>> Finally, the fit by Vipuli in all detail is here
>>> ("Report on pair Background")
>>> https://userweb.jlab.org/~rondon/sane/mtg7/index.html
>>>
>>> and the code I used to get the SANE e+/- background from the CLAS fit (a
>>> nontrivial job) is in this spreadsheet
>>> https://userweb.jlab.org/~rondon/sane/analysis/pairs/pair_run_plan_short.ods
>>>
>>>
>>> with this code I get f(E'=900 MeV)= 0.25 for CLAS only, and  f=0.62 for
>>> SANE, without the tracker, 0.71 with the tracker.  I estimated the
>>> tracker X0 using 9 mm of plastic.
>>
>> This is an underestimate for the tracker thickness. The tracker is at
>> least 12mm of plastic (one layer was a double layer) plus some
>> additional thickness for the WLS fibers and glue.
>
>OK. That is easy to adjust in my code, but actually it would make the
>dilution even worse.
>
>By the way, the method I use to get the pair dilution for SANE starting
>from the CLAS (Vipuli) dilution is explained in detail on slides 5 and 6
>here
>http://hallcweb.jlab.org/experiments/sane/rondon/snbkgd3-sdd.pdf
>
>> I am using the fits from the thesis of Robert Fersch.
>> http://www.jlab.org/Hall-B/general/thesis/Fersch_thesis.pdf
>> It appears to be doing a good job based on my agreement with SLAC d2. It
>> could certainly do better for cuts less than 1200MeV.
>>
>
>I'd be very cautious using Fersch's model, because for 5.7 GeV, his data
>go only up to 30 deg. and his fit doesn't agree well with the data even
>at this angle, see his Fig. 6.2.
>
>On the other hand, Vipuli's fit is for 5.7 GeV data for angles between
>36 and 44 deg. She sees little angular dependence, so her fit has only
>two parameters. I suggest using her fit rather than Fersch's, because
>hers is based on data closer to our kinematics.
>
>> I am currently implementing your pion cross section fits into my
>> simulation.  This should help understand lower E' background better.
>
>Great. Here are the new parameters for the pi0(PT) cross section:
>
>sigma_pi0(PT) = a*exp(C*PT), a = 10219+/-361, C = -8.675+/-0.065
>
>(the old parameters were similar a = 11150, c= -8.67, but had no errors)
>
>
>
>
>
>> Cheers,
>> Whit
>>
>>>
>>> Cheers,
>>>
>>> Oscar
>>>
>>>
>>>
>>>
>>>
>>>
>>>
>>>
>>> _______________________________________________
>>> Sane-analysis mailing list
>>> Sane-analysis at jlab.org
>>> https://mailman.jlab.org/mailman/listinfo/sane-analysis
>>
>
>
>_______________________________________________
>Sane-analysis mailing list
>Sane-analysis at jlab.org
>https://mailman.jlab.org/mailman/listinfo/sane-analysis