Hi Adam,
I have done this, and it appears there are typos in the paper
(I've heard the same from others). Take a look at '~/blast/exp/rpex/
thesis_cbc/macros/walcher_data/walcher_fit.C'. For an example of how
to use the covariance matrix from a ROOT fit, see '~/blast/exp/rpex/
thesis_cbc/macros/fit_arrington.C'. If you use TH1::Fit or
TGraph::Fit, then replace
minuit.mnemat(&cov[0][0],39);
with
gMinuit->mnemat(&cov[0][0],39);
note that 39 is the number of fit parameters, ie. a 39x39 matrix.
--Chris
TA-53/MPF-1/D111 P-23 MS H803
LANL, Los Alamos, NM 87545
505-665-9804(o) 665-4121(f) 662-0639(h)
On Aug 31, 2006, at 12:24:22, Michael Kohl wrote:
> Hi Adam,
>
> although there are uncertainties given in Friedrich/Walcher's paper
> (=diagonal elements of the covariance matrix), one can expect that
> there are correlations between the parameters (off-diagonal
> elements) leading to correlated errors for the parameterization.
> In order to really understand what's going here I think it is
> unavoidable to take the effort and reproduce the fits, parameters,
> and their errors on our own.
>
> Regards,
>
> Michael
>
>
> On Thu, 31 Aug 2006, Adam Jon DeGrush wrote:
>
>> Hi,
>>
>> Attached is a plot of what I showed in yesterday's meeting with
>> the Friedrich and Walcher curve corrected. Since part of our
>> discussion centered around the sizable error in the term
>> corresponding to the location of the bump in GEp, I added/
>> subtracted this error to generate two addtional curves. The curve
>> with a positive uGEp/GMp ratio in the region ~0.3 GeV^2 is from is
>> with it error added, the one that continuously falls off is from
>> the error subtracted.
>>
>> Regards,
>> Adam
>>
>>
>>
>>
>> On Wed, 30 Aug 2006, Michael Kohl wrote:
>>
>>> Hi,
>>> below are the minutes of today's analysis meeting.
>>> The next meeting will be on Wednesday September 6, 2006 at 13:30
>>> at Bates.
>>> Minutes:
>>> -Status ffr paper
>>> +Text has been reviewed.
>>> +Figures still being updated, asymmetry plot has been updated.
>>> +Chris showing plots of GE and GM. World data should really be
>>> original published values for GE and GM from experiments that have
>>> really performed Rosenbluth separation on actual measurements (not
>>> global fit of other data)
>>> +Some authors combine their cross section measurements with existing
>>> measurements to perform Rosenbluth separation. For this letter,
>>> really only experiments from which a single-expperiment
>>> extraction is
>>> possible should be plotted.
>>> +Yet another discussion on how to extract the form factors GE and GM
>>> from the BLAST ratio measurement combined with previous cross
>>> section data. Fit versus equation solving. Discussion about nature
>>> of cross section error when using parameterization, statistical
>>> versus systematic. Although the error of the cross section has
>>> a statistical and systematic component, for the purpose of using it
>>> with the the BLAST ratio measurement to deduce the separate form
>>> factors from solving the equations, the character of the total
>>> cross
>>> section error becomes systematic (no fitting involved). However,
>>> in a
>>> fit method where GE and GM result from a fit to the ratio and
>>> the cross
>>> section values, it only works if the cross section error is
>>> treated as
>>> point-by-point statistically independent and statistical in its
>>> nature.
>>> And this exactly causes problems here when using a parametrization
>>> because the statistical fluctuation is replaced by an error band.
>>> +Will sort things out in offline phone meeting tomorrow at 4pm. Will
>>> use the Bates conference line +1-866-867-8301, passcode 4073393
>>> -d(e,e'p)
>>> +hPz still Q2 dependent (~10% from lowest to highest bin). Possible
>>> origins: Diluted data sample (->check cuts); Resolution effects
>>> (->study effect of convoluting angles and momenta with finite
>>> resolution on asymmetry by varying resolution); or polarization
>>> profile along z corresponding to Q2 profile through correlation of
>>> z and Q2 (unlikely)
>>> +need to compare asymmetries and hPz vs z with MC with and without
>>> accounting for spin angle profile. Remaining variation of hPz
>>> with z
>>> could be due to polarization profile.
>>> +proton ff ratio from quasielastic d(e,e'p) at low pmiss (see
>>> plot in
>>> meeting directory) from super ratio. Ratio data from deuterium
>>> slightly below hydrogen data.
>>> +Friedrich/Walcher parameterization showing bump even for the ratio,
>>> in fact parameter Q_b for the location of bump is much smaller
>>> for GEp
>>> (0.07) than for GMp (0.35), therefore bumps in GEp and GMp seem
>>> not to
>>> cancel. However, Fig.3 in F.-W.'s paper shows bump at 0.2-0.3
>>> GeV for
>>> all form factors. Something wrong in the Friedrich/Walcher
>>> paper? One
>>> should verify the fit to the data, like Chris did for Arrington's
>>> parameterization.
>>> -Comparison of MC for different radiative generators
>>> +3 radiative generators available: Mascarad-dgen (Vitaliy), pionmc
>>> (Aki) and ElRadGen (Tavi+Sasha)
>>> +still bugs in implementation of ElRadGen, need more time
>>> +ElRadGen still unpolarize only
>>> +need to compare W spectra at generator level
>>> Best regards,
>>>
>>> Michael
>>> +-------------------------------------+--------------------------+
>>> | Office: | Home: |
>>> |-------------------------------------|--------------------------|
>>> | Dr. Michael Kohl | Michael Kohl |
>>> | Laboratory for Nuclear Science | 5 Ibbetson Street |
>>> | MIT-Bates Linear Accelerator Center | Somerville, MA 02143 |
>>> | Middleton, MA 01949 | U.S.A. |
>>> | U.S.A. | |
>>> | - - - - - - - - - - - - | - - - - - - - - -|
>>> | Email: kohlm@mit.edu | K.Michael.Kohl@gmx.de |
>>> | Work: +1-617-253-9207 | Home: +1-617-629-3147 |
>>> | Fax: +1-617-253-9599 | Mobile: +1-978-580-4190 |
>>> | http://blast.lns.mit.edu | |
>>> +-------------------------------------+--------------------------+
>>
>
>
> +-------------------------------------+--------------------------+
> | Office: | Home: |
> |-------------------------------------|--------------------------|
> | Dr. Michael Kohl | Michael Kohl |
> | Laboratory for Nuclear Science | 5 Ibbetson Street |
> | MIT-Bates Linear Accelerator Center | Somerville, MA 02143 |
> | Middleton, MA 01949 | U.S.A. |
> | U.S.A. | |
> | - - - - - - - - - - - - | - - - - - - - - -|
> | Email: kohlm@mit.edu | K.Michael.Kohl@gmx.de |
> | Work: +1-617-253-9207 | Home: +1-617-629-3147 |
> | Fax: +1-617-253-9599 | Mobile: +1-978-580-4190 |
> | http://blast.lns.mit.edu | |
> +-------------------------------------+--------------------------+
>
>
>
>
>
>
>
>
>
>
>
>
>
This archive was generated by hypermail 2.1.2 : Mon Feb 24 2014 - 14:07:33 EST