Nonnegative-orthant optimality certificate

ID: nonnegative-orthant-optimality-certificate

Consider the original linear program
One unit of costs twice as much as one unit of , so is a plausible answer. To certify it, multiply the three linear inequalities by , respectively, and add:
The proposed feasible point attains , so its cost is the global minimum.
In the conic dual problem convention, the data and certificate are
Direct matrix multiplication gives and . At , the slack variable is , so . The first linear inequality has positive slack and zero multiplier; the two positively weighted linear inequalities are tight. This is complementary slackness.
Figure 1.
A feasible decision and a weighted-inequality certificate meet at the same cost
. The shaded feasible set lies above in the nonnegative orthant. Dashed lines show costs . The lowest attainable line touches . The right panel verifies the dual variable : its weighted constraint coefficients reproduce the objective function, and its weighted right-hand sides give the lower bound . Details.
The entire dual feasible set can also be read off: write with . Then and . Maximizing this lower bound gives . A point such as is still a valid certificate of the weaker bound .

New to topics? Read the docs here!