Solving dynamic control problems via deterministic global optimization
A significant multi-stage stochastic program from the area of financial planning is posed as a nonlinear stochastic control problem. The dynamic policy, called fixed-mix, results in a nonconvex optimization model. A deterministic global optimization algorithm specialized for this problem class produces a guaranteed optimal solution for realistic size applications. The proposed branch and bound type deterministic algorithm guarantees finite {element_of}-convergence to the global solution through the successive refinement of converging lower and upper bounds on the solution. These bounds are obtained through a novel convex lowering bounding and the subsequent solution of a series of nonlinear convex optimization problems. Computational results obtained with an efficient C implementation of the proposed procedure GLOFP, demonstrate the efficiency of the approach on a set of real world financial planning problems. These tests confirm that local optimization methods are prone to erroneously underestimate the efficient frontier. The concepts can be readily extended to other non-convex dynamic policies.
- OSTI ID:
- 36315
- Report Number(s):
- CONF-9408161--
- Country of Publication:
- United States
- Language:
- English
Similar Records
Global Solution Strategies for the Network-Constrained Unit Commitment Problem with AC Transmission Constraints
An Iterative Approach for Solving the SCOPF Problem Applying LP, SOCP, and NLP Subproblems