> We can however add a function to the interface to > get the dual solution of the initial root LP, if that is what you > need. > > Yes. I usse Symphony as a LP solver because it is nearly 5 times faster than CLP. Yann