From: Ian on
I have heard that tuples are less efficient when compared to pre-allocated matrices.

I was wondering, if the tuples contain pre-allocated matrices, does this slow down computation time?
I have very elongated tuples, however I seem to get very fast performance still... for instance:
a = {}
a{group}{metric}{subject} = zeros(Something, something);

When I add a new group, the entirety of all ..{metric}{subject}[SxS] are not copied??? because they are isolated in their own tuple or not?