Normal conical algorithm for concave minimization over polytopes (Q1177231)

The class of conical algorithms, introduced in 1964 by the author, designed to find the global minimum of a concave function over a polytope uses valid cuts to exclude parts of the feasible domain, while partitioning the latter through an exhaustive subdivision cone splitting process. To ensure faster convergence and overcome some major drawbacks a sufficient convergence condition for the class of conical algorithms is given, and on the basis of this condition a new algorithm, called ``normal'', which is easy to implement and performs better in solving moderate size \((n\leq 50)\) problems on microcomputers, is presented.