Computation and Theory in Nonlinear Optimization.
Final rept. 15 Jan-14 Nov 88,
Pagination or Media Count:
Researchers have made substantial progress in the decomposition of large-scale nonlinear problems for accelerated convergence. A Newton method for nonsmooth equations has been developed and conditions for its convergence determined. The original proposal suggested work in the general area of large-scale optimization in particular, methods for decomposition based on the so-called Bundle method, decomposition of large-scale nonlinear problems, including parameter optimization in networks of queues, and accelerating the convergence of bundle-type methods, including developing the theory base in areas such as implementable second-order models of functions to be optimized. A particular area of work suggested as a major topic of investigation was the development of a computationally implementable and efficient Newton-type algorithm for nonsmooth problems. jhd
- Operations Research