A Nondifferentiable Optimization Algorithm for Constrained Minimax Linkage Function Generation

This paper describes a nondifferentiable optimization (NDO) algorithm for solving constrained minimax linkage synthesis. Use of a proper characterization of minima makes the algorithm superior to the smooth optimization algorithms for minimax linkage synthesis and the concept of following the curved ravines of the objective function makes it very effective. The results obtained are superior to some of the reported solutions and demonstrate the algorithm’s ability to consistently arrive at actual minima from widely separated starting points. The results indicate that Chebyshev’s characterization is not a necessary condition for minimax linkages, while the characterization used in the algorithm is a proper necessary condition.