Sun Feb 24 11:24:28 CET 2013

Guessing types of linear functions : linear variable vs. multiplicative parameter

Is it necessary?  Aren't all types known?

No.  One of the problems to pull this off might be that not all inputs
are known.  I.e. there might be inner nodes that add constants, which
for a linear program interpretation act as inputs.

I.e. it results in responses instead of transfers.

Maybe responses is what we're looking at?

The problem doesn't seem very well-defined yet.