The last two weeks I spent mainly working on doing optics fits and trying
working on my presentation for APS on Saturday.  A few revisions have
been made and the latest copy can be found:

http://www.jlab.org/~riordan/gen-APS07-riordan.pdf

I've had a little success with getting fits to work in correcting the
extreme regions of the magnet.

Page 1 -
This is how much the vertex reconstruction is ruining our momentum
reconstruction.  Everything is within 1% and there is a constant shift
which I believe is taken care of by shifting the center around.
Basically, this says that the vertex reconstruction and the momentum
reconstruction optics problems are probably separable problems.

Page 2 -
I allowed a chi^2 minimization code to fit to all the regions in the
magnet.  Unfortunately, once we get away from the center the momentum
reconstruction is unphysical and we get poor reconstruction once we get
any amount away from the center.

This is due to the leading and xbend coefficients playing off of each
other.

Page 3 -
I fit the x bend coefficient in the center across several y values.  Then
I held this constant for constant y.  The momentum spectrum here is much
more reasonable.  The fits were poor in the very extreme regions due to
poor statistics.

Page 4 - 
Projecting the exact same events to the shower we see that some of the
inconsistancies are likely due to miscalibrated blocks, so we get a
reasonably believeable spectrum.

Page 5 -
This is the momentum resolution across X is.  Except for the very extreme
regions with poor statistics (x>0.4m), we get acceptable values.

Page 6 and 7 -
I broke apart the previous plot into various x regions and fit the
resolution (using a gaussian and constant background).  Page 7 is the
results for sigma as a function of x.  Everything is reasonably stable.


B.tr.bendx[0]>0.4 represents about 8% of our statistics.