Will asked for more information about the parameters and shapes.  The following plots (both inline and in the linked pdf files) include all this information.  The following use the L2 mock trigger as a sort of filter for the EEMC trees, as discussed here.

Fitting the MC Templates to Template Functions

• Signal
  • p0 = const
  • p1 = mean
  • p2 = sigma
  • p3 = skew
  • p4 = scale factor for second Gaussian
  • p5 = second sigma
  • p6 = second skew
• Conversion background
  • p0 = const
  • p1 = mean
  • p2 = sigma
  • p3 = skew
• Other background ("pol3")

Reduced Chi^2

Fitting the Data to the Templates

Recall:

• Spin summed case
  • Free parameters are three weights and the MC energy scale factor alpha (less than one if MC energy is higher than data)
  • Initial values are the relative normalizations of the three templates in MC for the weights, and alpha = 1.
• Individual spin cases
  • Free parameters are only the weights, not alpha which is fixed to the value from the spin summed case
  • Weights initialized to the values from the spin summed case

The mass distributions, per pT, with the functions, sum of functions, data, and residual are here.

Reduced chi^2

Fractions in the Peak

Alpha Scale Factor and Pi0 Peak Position

Note: the peak position is directoy correlated with the alpha scale factor, and so fixing the alpha factor for the spin sorted fits to the value for the total fit forces the pi0 peak position for the spin sorted fits to all be at same position (the position from the spin summed fit).  The jump of the last point is somewhat atypical, but the variance of all the points about the mean is still on the order of 2 MeV.