[simcop2387]

That might work nicely with a lot of the sparse matrix research going on too, since I suspect that the trust matrix should be very sparse