IPF

IPF allows one to find a matrix S, close to an input matrix T, but such that the row sums of S are R, and the column sums of S are C.

Its useful in a range of tasks (I use it in traffic matrix problems), but is often used in statistics for examining independence assumptions in contingency tables.

See
http://en.wikipedia.org/wiki/Iterative_proportional_fitting
for more info (the verion implemented above is the RAS version from that page).

I have a more advanced version designed tofund S satisfying an arbitrary set of linear constraints, but its not quite in hand-out-able form, so ask me if you need it.

See internal documentation and ipf_test.m for details of how to use it.