I've been utilizing very wide bins in my response matrix up until now assuming that the variation of the response over a single bin is small.  To check this, I've put together a couple of plots.  First is just the response matrix with 1 GeV/c bins plotted next to the matrix with variable bins.

Note that both response matrices have been normalized in the usual fashion.  The next plot shows projections (of detector jet pT) of the response matrix for multiple bins of particle jet pT.

The choices of particle jet pT correspond to some of the larger bins used in the unfolding procedure.  The same plot but with larger bins is attached.  The bin low edges used in the unfolding are:

pTbinEdges = {0., 0.2, 0.6, 1., 1.5, 2., 3., 5., 8., 12., 17., 23., 30.}