Criss-cross method

From Glossary

Jump to: navigation, search

Criss-cross method. A method in linear programming that chooses a pivot by possibly crossing from primal to dual, and vica versa. The least-index criss-cross method is a finite pivot method that chooses a pivot based on the least index of a column or row for which there is improvement in either the primal or dual infeasibility. It terminates at a basis when one of the following termination conditions is reached:

  1. Both the associated primal and dual solutions are feasible (implies optimality because complementary slackness holds by construction).
  2. There is evidence of primal and/or dual infeasibility (like the tests in the simplex method).
(Note: any basis is used to start and there is only one phase.)


Views
Personal tools