Sensitivity analysis based heuristic algorithms for mathematical programs with variational inequality constraints (Q2276884)

The authors consider some algorithms for the bilevel programming problem of minimizing F(x,y) subject to \(L\leq y\leq U\), \(x\in K(y)\), and Z(x,y)(z- x)\(\geq 0\) for all \(z\in K(y)\) [cf. \textit{P. Marcotte}, Math. Program. 34, 142-162 (1986; Zbl 0604.90053)]. For many applications the preceding variational inequality constraint defines x as a relatively smooth function of y. After presenting some results about sensitivity analysis of variational inequalities formal statements of three heuristic algorithms for the bilevel programming problem are given. The first two of them are descent algorithms and they differ from one another in the manner in which stepsizes are determined, either according to the Armijo rule or a priori. The third one is a generalization of the equilibrium decomposition optimization algorithm proposed by \textit{C. Suwansirikul, T. L. Friesz} and \textit{R. L. Tobin} [Transp. Sci. 21, 254-263 (1987; Zbl 0638.90097)]. Finally some numerical examples drawn from equilibrium network design, hierarchical mathematical programming, and game theory are discussed.