Here I do a consistency check on the RooUnfold package comparing the unfolded detector level distribution from simulation to the corresponding particle and parton level distributions...

This is a test which should have been performed when I started using the RooUnfold package, but I did not think of it at the time. Anselm recently suggested that Xuan try it, and so I will as well.

To test that the RooUnfold (specifically, the SVD algo) works as it should, I have used the simulated detector level dijet mass spectra as the distribution to be unfolded. This is just the projection onto the x-axis, of the unfolding response matrix. Since the relationship between the particle level and the detector level is know in the simulation and encoded in the response matrix, unfolding the detector level spectrum should give back the particle level spectrum.

Figure 1: This top pannel shows the detector level dijet mass spectrum (Red), the corresponding particle level dijet mass spectrum (Black), and the detector level spectrum which was unfolded using the response matrix constructed from the detector and particle level information (Blue). The bottom pannel shows the ratio of the unfolded detector level spectrum to the particle level spectrum.

Figure 2: Same as figure 1, but now the correction is done and the comparison made to the parton level.