Stochastic Approximation Type Algorithms for the Optimization of Constrained and Multinode Stochastic Problems.
BROWN UNIV PROVIDENCE R I CENTER FOR DYNAMICAL SYSTEMS
Pagination or Media Count:
The aim of the paper is the development of a structure for stochastic optimization algorithms of the Monte-Carlo or stochastic approximation type which is analogous to that used in non-linear programming. The developed structure is quite versatile, and seems to consider the elements of the problem in a very natural manner from both the theoretical and practical viewpoints. A second paper is also included in the report, titled Stochastic Approximation Algorithms for the Local Optimization of Functions With Non-Unique Stationary Points.
- Operations Research