> I am not sure I understand it. OK, I first solve the > simplex problem asuming the variables are real and > then pass the solution through the LPX pointer to > lpx_integer. I was wondering why there is this > intermediate step why the function could not do it for > me in one sweep. Unless I misunderstood something.
Yes, firstly you need to obtain optimal basic solution to lp relaxation using lpx_simplex and then call lpx_integer to obtain a mip solution. The process is made two-pass in order to allow the user having alternate ways for specifying optimal basis for lp relaxation. For example, the user could read an optimal basis from a file on solving a set of instances with identical lp relaxation. _______________________________________________ Help-glpk mailing list [email protected] http://lists.gnu.org/mailman/listinfo/help-glpk
