[Frost] tagger sag

[dugger at jlab.org]

Volker,

I do not think you fully read what I wrote. It does not matter. I do not
want to be critical.

I only had a suggestion. That is all. I can tell that there is no sense in
trying to change your mind regarding the accuracy of the FSU kinematic
fitter when the pulls are not optimal.

This has become boring.

Sorry to be an irritant.

Sincerely,
Michael



> Mike,
>
> just to avoid some confusion here: Sung has not yet applied any momentum
> corrections (only eloss corrections). If I recall correctly, momentum
> corrections had a way bigger impact on the g8b pull distributions than the
> photon energy correction. I think this needs to be the next step and Sung
> is working on it. What I am saying is that we cannot easily compare mean
> values from g8b and Frost. Also, the sigma improvement is only based on
> the adjustment of the covariance matrix (and has nothing to do with the
> corrections we are discussing here).
>
> For the moment, we can only get rough estimates from the fitter for the
> tagger sag; perhaps on the level that we see if something happened in the
> cooking. But I remember a plot that Sung showed me a few weeks ago that
> indicated a much smaller problem ([E_calculated - E_measured]/E_measured)
> for the g9a data.
>
> Best wishes
>
>     Volker
>
>
> On Wed, 9 Jun 2010, Michael Dugger wrote:
>
>>
>> Sung,
>>
>> I really hate to be a bother. From what I can tell, the plots you show
>> are
>> from the kinematic fitter. The FSU kinematic fitter showed the exact
>> same
>> kind of agreement for g8b. I do not know why. There were many
>> discussions regarding this. Please see
>> http://www.jlab.org/~dugger/tmp/meanValues_048326.A10.gif (this plot
>> goes
>> along with the email I got from Charles Hanretty shown at the bottom of
>> this email). Charles and I had the exact same conversations regarding
>> the
>> tagger sag for g8b.
>>
>> The pull parameters for the photon energy are a much better gauge of how
>> well the measured photon energy compares to the calculated when using a
>> kinematic fitter. A kinematic fitter can not be relied upon to give
>> accurate results when the pulls are not close to optimal. This appears
>> to
>> very much the case for the calculated photon energy using the FSU
>> kinematic fitter for g8b (and probably for g9a as well).
>>
>> You can see the difference in the pull parameters before and after the
>> tagger sag correction for g8b data in Fig. 7 on page 8 of
>> http://www1.jlab.org/ul/Physics/Hall-B/clas/public/2009-030.pdf
>>
>> Your photon pulls shown on your web page at:
>> http://www.jlab.org/Hall-B/secure/g9/sungkyun/6-1-10.html
>> Give the following parameters:
>>
>> Carbon->
>> mean 0.122
>> sigma 1.124
>>
>> CH2->
>> mean 0.251
>> sigma 1.235
>>
>> For Chuck using g8b data:
>>
>> Before tagger sag correction->
>> mean 0.3027
>> sigma 1.219
>>
>> After tagger sag correction->
>> mean 0.0049
>> sigma 1.040
>>
>> At first glance, it looks entirely reasonable to suspect that you have
>> the
>> same sort of photon energy problems that Chuck had.
>>
>> I also want to make it clear that the sag correction was supposed to be
>> included in the g8b data and it was never determined what went wrong.
>> Why
>> should we assume that the tagger sag is fixed when it has not been
>> verified?
>>
>> One more thing that might be of interest: It appears that the
>> [E_calculated - E_measured] is nearly perfect for the comparisons
>> done by the kinematic fitter. However, it would be very surprising to
>> get
>> this same type of result from your most recent study at:
>> http://www.jlab.org/Hall-B/secure/g9/sungkyun/6-9-10.html
>>
>> I only propose that you find out [E_calculated -
>> E_measured]/E_measured using your most current study (without kinematic
>> fitter) and include enough statistics to get the numbers to the 0.1%
>> level. This would serve as a way to verify that the tagger sag is
>> correctly included in g9a.
>>
>> No matter what, you are going to have to get your photon pulls to be
>> much
>> better before you can reliably use the kinematic fitter in your
>> analysis.
>> That is the price you pay for using such a powerful technique.
>>
>>
>> ************************
>> Date: Wed, 19 Aug 2009 15:18:41 -0400 (EDT)
>> From: Charles Hanretty <hanretty at hadron.physics.fsu.edu>
>> To: Michael Dugger <dugger at jlab.org>
>> Subject: Re: E(true) and E(meas)
>> Parts/Attachments:
>>    1 Shown    161 lines  Text
>>    2   OK      12 KB     Image
>> ----------------------------------------
>>
>> Mike,
>>    Attached is a plot I have of the mean values generated from just 1
>> 1.3GeV file
>> (cooked_048326.A10.B00).  I fit each slice of the (Etrue/Emeas)vEmeas
>> plot
>> by using the function
>> TH2::FitSlicesY (168 slices).  Looking at this, one could see the
>> similarity between this plot and
>> the one you showed at the meeting.  I've made this final plot
>> (Etrue-Emeas v Emeas) but it does not look how I expect it to or want it
>> to
>> so I'm trying to figure out if I made an error somewhere.  One thing to
>> note
>> is that the differences between Etrue and Emeas (for me) seem to be
>> smaller
>> than the differences you are dealing with.  This might flatten out some
>> of
>> the bumps we're looking
>> for.  For example, your Etrue/Emeas v Emeas plot has a
>> y-range of 0.95-1.05.  I started off with this range but had a lot of
>> empty
>> space and therefore have changed my y-range to 0.998-1.002.  Anyway,
>> this
>> is an update; there is
>> still work to be done.
>>
>> -Chuck
>> *************************************88
>>
>> -Michael
>>
>>
>> On Wed, 9 Jun 2010, Sungkyun Park wrote:
>>
>>> Hi Michael,
>>>
>>> My understanding is that we do not worry about the problem of the
>>> tagger sag in FROST data.
>>> When we make the cooking file, that problem is corrected.
>>>
>>> I make a same plot made at g8b.
>>> The first plot is about E-id vs E{true photon beam} - E{measured photon
>>> beam}
>>> The second plot is about E{measured photon beam} vs E{true photon beam}
>>> - E{measured photon beam}
>>>
>>> These plots are made in topology #gamma p -> p #pi+ #pi- (all detected)
>>> and the updated eloss correction is applied.
>>> E{measured photon beam} is the initial photon energy.
>>> E{true photon beam} is the photon energy after kinematic fitting.
>>> I used 0.05 as the confidence level cut. I used only a run 55570 in
>>> period 3. The electron beam energy of period 3 is 1.645 GeV. So The
>>> energy range of E{measured photon beam} is from 0.33 GeV to 1.56 GeV.
>>> When we compare plots in CLAS-NOTE with plots attached with this email,
>>> FROST data do not have any problem of the tagger sag.
>>>
>>> Sung
>>> Florida State University
>>>
>>>
>>> ----- Original Message -----
>>> From: Michael Dugger <dugger at jlab.org>
>>> Date: Wednesday, June 9, 2010 1:04
>>> Subject: [Frost] tagger sag
>>> To: frost at jlab.org
>>>
>>>>
>>>> Hi,
>>>>
>>>> I know that the study Sungkyun performed was to help verify the
>>>> eloss
>>>> momentum corrections, and I think that this is a reasonable way to
>>>> look at
>>>> his study. However, this sort of study is also very useful in
>>>> determining
>>>> the tagger sag. Since the eloss correction is not as pronounced for
>>>> the
>>>> Carbon and CH2 targets, these targets can give us a good picture on
>>>> the
>>>> tagger sag situation.
>>>>
>>>> If you look at the CLAS note:
>>>> http://www1.jlab.org/ul/Physics/Hall-B/clas/public/2009-030.pdf
>>>>
>>>> Fig. 5 on page 6 shows the tagger sag correction to be on the order
>>>> of
>>>> 0.5% for g1c data. For g8b data, the sag correction was between 0.2
>>>> to
>>>> 1.0% (Fig. 6 page 7). This means that we need to look for energy
>>>> effects
>>>> on the order of 0.1% to be sure that the tagger sag is not an issue.
>>>>
>>>> For g8b it was found that the tagger sag correction was important
>>>> in
>>>> obtaining reasonable pulls for the FSU kinematic fitter (see table
>>>> 1 on
>>>> page 7).
>>>>
>>>> One way to get a better look at the energy study is to use:
>>>> [E_calculated - E_measured]/E_measured
>>>> instead of
>>>> E_calculated/E_measured.
>>>>
>>>> Since Mike Williams used this sort of parametrization for g1c (and
>>>> I
>>>> did the same for g8b), we can more easily compare results between
>>>> the
>>>> different run periods to see if the shape of the energy correction
>>>> distributions look similar. The tagger sag has a "signature" three
>>>> bump
>>>> pattern that should be clearly visible once we get to the 0.1% level.
>>>>
>>>> -Michael
>>>> _______________________________________________
>>>> Frost mailing list
>>>> Frost at jlab.org
>>>> https://mailman.jlab.org/mailman/listinfo/frost
>>>>
>>>
>> _______________________________________________
>> Frost mailing list
>> Frost at jlab.org
>> https://mailman.jlab.org/mailman/listinfo/frost
>>
>