Thank you for the tip. There is an other possibility to get rigorous (100% sure even for ill-conditioned and badly scaled problems) error bounds for the optimal value of the LP with interval arithmetic, Christian Jansson has many publications on this topic.
I think that the exact simplex solver is very interesting but it is very slow for real life problems (have not tried yet though). The above mentioned rigorous error bounds can be computed quickly even for large problems. Ali _______________________________________________ Help-glpk mailing list [email protected] http://lists.gnu.org/mailman/listinfo/help-glpk
