[Halld-offline] one final followup on the kinematic fit

[Kei Moriya]

Dear folks,

This is just one final followup on the kinematic fit talk
I gave yesterday. I have placed 2 files,
http://dustbunny.physics.indiana.edu/~kmoriya/2011.06.03.kfit/errorsInMeV.pdf
http://dustbunny.physics.indiana.edu/~kmoriya/2011.06.03.kfit/errorsInGeV.pdf
that look at the kinematic fit results.

Richard had asked me during the meeting yesterday whether the
errors I had put in the initial photon momentum had any effect
on the final kinematic fit result, and later on provided me with
the actual errors that we expect from the tagging system.

I tried to implement these errors into the code, with
sigma_x = sigma_y = 1.8 MeV and sigma_z = 5 MeV,
but I mistakenly put the errors in GeV first, which resulted
in the second file above. The first file above is in the
correct units of MeV (the topology is gamma + p -> pi+ pi+ pi- (n))

The point I want to make is that when the initial photon
is given huge errors of 1.8-5.0 GeV, the kinematic fitter
has enough freedom to wiggle around the momenta to get
a very low chi2, which shows up in the confidence level
distribution in p.2 of each file. Notice the slope in
CL, which is weighted towards the higher end for the GeV
errors file.

The effect of the errors on the final kinematically fit photon
momentum can be seen in p. 5, where the kinematically fit
momentum is distributed according to the errors that were
put in.

Finally, the pull distributions are in p. 4
(I have also fixed the gaussian widths, thanks again Richard).
Whereas for the actual errors that we want, the pull distributions
are much wider than the normalized gaussian, with the huge errors
from the photon beam, the pulls are way too tight.

As David pointed out yesterday during the meeting, the kinematic fit
can do only as well as the errors it is given by the tracking code,
and we can see that even having the wrong errors on the photon can really
change the kinematic fit results.

I will go back to looking into the tracking errors again,
and hope to disentangle the effects of the multiple scattering
and energy loss.

	Kei