Look at the linear_program function on this page, use RDF for your floating-point real numbers, and tell us what happens !