[Coin-discuss] MILP

[Lou Hafer]

Luwi,

	Would that it were that easy. The sbb library components, and the
various main programs (sbb, sample2, etc.) are still works in progress. `At
least as much art as science' is probably a good assessment of the state of
the art for solving MIP problems.

	The primary stand-alone front-end for the sbb library is built from
SbbMain.cpp. Try `make unitTest' in the Sbb directory. You should get an
executable called sbb.

	From the sounds of it, the cuts and heuristics currently included in
sbb are not sufficient to solve your MIP. Size is definitely not the main
determinant. Rather, it's the shape of the polyhedron defined by the
constraint system. Unfortunately, though there are tantalizing leads,
determining the correct approach for a MIP by analysing the polyhedron is
still beyond the current state-of-the-art. (I keep hoping someone will jump
in to tell me I'm wrong .)

	If you're feeling ambitious, you might want to try adding some of the
new cut methods that are appearing in the Cgl library. The pattern for using
additional cut generators should be fairly obvious from the code. We (the core
team) will likely get around to integrating these once the move to INFORMS is
past. Help is always appreciated, however.

	Not what you wanted to hear, I imagine. Others may have more practical
(and less generic) advice.

							Lou