Parametric Techniques for Multistage Stochastic Allocation.

摘要

This paper treats a problem of multistage allocation under conditions of risk. After qualitative features of the problem are discussed, an efficiency procedure is formulated and is shown to have desirable computational, utility-theoretic, and asymptotic properties. Several techniques are then developed which are applicable to an approximation of the efficient frontier by parametric complementary pivoting. These include a parametric algorithm for a minimax approximation, some parametric decomposition methods, and an interactive algorithm for certain problems with a block-angular structure. A listing for a Fortran 4 code solving the parametric linear complementarity problem is given in an appendix.