On the incentive method and its application to linear programming

Using linear incentives to decompose a convex dynamic optimization problem into subproblems along the time axis makes low-level subproblems the same as those obtained from Lagrangian multiplier decomposition. The authors illustrate how incentive methods are related to multiplier methods, which are reinterpreted by the use of incentives. This viewpoint motivates the derivation of exterior methods for linear programming, in which an optimal solution is obtained by traveling in the exterior of the feasible region. Numerical examples are used to demonstrate the performance of one exterior method