RECON STUDIES:

Eugene Geis, Taylan Akdogan, Aaron Maschinot


#########################
EG: recon studies

	_ Inbending H-data from 2003 (run # aroun 2000)
		_ shows Proton Mom vs Proton theta
		       electron Mom vs electron theta
		_ proton better resolved
	_ Compares to inbending data of this (new libblast, calibration)
		_ e- momentum at least as good or better
		_ proton momentum much worse
		_ can not blame 2004 offsets since electron ok
		_ Something broke ? Maybe proton beta correction ?

	_ Needs corrected electronics map. Wrong emap was severely
	  affecting acceptance (CC-TOFbottom confusion)
	_ however e-map does not explain bad res runs of this year..

	
	_ Looks at wire 12, 23, 13 xdrift in a cell after ep cut
		_ x_1(23):x_2(31) correlates well for protons, where track angle ~0
		_ there are gaps for electrons where track angle is not corrected for
		_ can clearly identify the region where the wire-drift stub/sign 
		  assignement is wrong.
	_ For dx, xdrift intrinsic resolution estimated to be to ~ 0.8/sqrt(6) mm ~ 0.3
		_ improves for perpendicular tracks (protons) to ~0.2 mm
		_ may explain why proton resolution generally better
		_ note: this selecting only linear region. Many e- cross non linear
		  region (especially 3rd chamber)
	
####################################


TA: wire chamber calibrations

	_ Multi super layer line construction
		_ use hit info from both SL to get segment at once
		_ large # of combinatorial (average ~ 512) makes choice
		  of segment very cpu intensive. 
		_ Can be reduced by excluding unphysical stubs
		_ now compare distance of each wire-hit to segment line
		
	_ First & Second iteration 

		_ determine t0,x0 relative to segments
		_ dynamically re-adjust segment
		_ shows plot of wire distance to segment line (Dx)
		_ across whole cell Dx has sigma ~ 440 um

	_ in next iteration correct t2d according to profile
		of Dx (at each x)

		_ now all Dx cluster well but differently for L,R of wire
		_ this because field lines are different L,R
		_ simply means empirical t2d will be different for L,R

	_ Iteration-4
		_ now Dx width reduced to 240 um
		_ which gives a new t2d (numerically corrected)

	_ note this should work also for cross cell tracks !

	_ NOTE that before any iteration, the intrinsic wire estimate obtained
	  via dx above - for the linear region only - is 200-300 (note for EG
	  dx=x0+x2-2x1, while for TA Dx is just segment-x_i). 

	_ Maybe this means that with above method we can significantly improve (x0.5)
  	  on position per wire

	_ Montecarlo study show delta_p recon roughly linear with pos uncertainty 
		(albeit with linear fitter, see chris)	  


##############################33


AM: wch alignement sensitivity study

	_ moved subdetector in x,y,z
	_ offset not significant up to large shifts
	_ significant changes in reconstruction accuracy 
		only seen for 10 cm shifts !
	_ likely  not a significant source of uncertainty
	  at this time
	_ some thoughs about relative chamber-to-chamber misalignemnets