A Generalized Gradient in Optimal Control.
Abstract
Gradient methods or methods of steepest descent in control-space have been found to be quite useful for the numerical solution of a large class of problems in optimal control. However, there is a large class of problems for which these methods have not been successful. To deal with this larger class of problems, a generalized gradient method is proposed. The convergence rate of this method is analyzed for a certain class of problems and the analysis is verified numerically. The analysis is then used to suggest improvements in the basic algorithm. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Jul 01, 1973
- Accession Number
- AD0767552
Entities
People
- David D. Barnett
Organizations
- University of California, Berkeley