The Final Comparison of Simulations to Data
In comparing the two simulation codes, before arriving at 'final' versions with which to compare the outputs, I went through various steps and changed either some of the input parameters or the fortran code itself. The most significant change was to input the form factor models from SIMC into MCEEP. The final simulation comparison plots shown here were obtained with no collimator model used in either simulation and with all radiation and energy loss turned on. The form factor fit by John Arrington was used in both simulations - more details can be found in the link to the codes above. In this case, I have used reduced acceptance cuts of 20% less than nominal in theta, phi and delta-P as well as a cut on the W2 distribution. The same cuts applied to the data were used on the simulations. A particle ID cut on the gas cherenkov was also used on the data; efficiency corrections for particle ID, track reconstruction and deadtime have all been applied to the data.
A table (pdf file) summarising the integrated yields for the data and simulations for each step in the modification process are given here. The integrated yields for different cuts on the W2 distribution (again using the 20% reduced acceptance cuts) are shown on the second page.
The final step for comparing the MCEEP and SIMC simulations to the data, was to subtract the dummy cell background from the data. Unfortunately, during our experiment, we did not take any data on the dummy cryotarget cell. Instead, I have used data from a carbon foil, taken at the same time at similar kinematics. By integrating the yields upto the same point in the carbon and hydrogen spectra, I was able to rescale the W2 distribution from the carbon data and use it as the background distribution in the hydrogen spectrum; the rescaled background is the light blue trace in the first figure. NOTE: in each figure, the red trace is from MCEEP and the green trace is from SIMC. Then I simply subtracted this background off the hydrogen W2 distribution to leave the background reduced data - this is shown in the second figure, on both logarithmic and linear scales.
You can see from the W2 distribution that both
simulations agree well in the tail but in the peak region, MCEEP is
higher and perhaps one bin offset from SIMC; the data is also wider in
the peak but I have not included any resolution or smearing effects in
the simulations. Some further work on checking the data and any
associated calibrations in the software codes is now required. The
ratio of yields is now,
- MCEEP/SIMC = 1.047
- Data/MCEEP = 0.95
- Data/SIMC = 0.995
But from the figure, one can see that the ratio of data to SIMC yield of 0.995 is somewhat misleading due to the disagreement in shape, especially around the peak region.
Looking at the Q2 distribution with the same set of cuts applied, the data trace is now larger than the two simulations, whereas originally MCEEP was larger. This is due to the change in the form factor model and also due to the change in cut from a Q2-omega graphical cut, to the linear cut on the W2 distribution. The graphical cut on Q2-omega was perhaps misleading as it resulted in a very narrow cut on the peak in the W2 distribution as can be seen in this link. Note also that for the Q2 distribution plot and the HRS variables plots, the data is not background subtracted because the carbon run used for the background data included the sieve plate while the hydrogen data did not. Thus, the distributions for these variables with the sieve plate in are sufficiently different (they have spikes from the sieve holes!) that I could not use them to subtract out the background.
The full encapsulated postscript versions of the plots shown are available here:-