Publication: Sparse Matrix Optimisation using Automatic Differentiation
Introduction
Applications
Tools
Research Groups
Workshops
Publications
   List Publications
   Advanced Search
   Info
   Add Publications
My Account
About
Impress

Sparse Matrix Optimisation using Automatic Differentiation

- Ph.D. thesis -
 

Author(s)
Richard C. Price

Year
1987

Abstract
This thesis considers the problem of finding a minimum of a multivariate function with and without constraints. The truncated Newton method is used as the method for finding the minima, see [Dixon1988New]. In the constrained case, the Di Pillo Grippo penalty function is used. As these methods require first and second order tensors, differentiation arithmetic is used. A modification is made to the basic method described in [Rall1981ADT] which allows only vector storage to be used. Various modifications of the method are applied to real-life problems, and comparisons are given.

BibTeX
@PHDTHESIS{
         Price1987SMO,
       AUTHOR = "Price, Richard C.",
       TITLE = "Sparse Matrix Optimisation using Automatic Differentiation",
       SCHOOL = "Hatfield Polytechnic",
       ADDRESS = "Hatfield, U.K.",
       YEAR = "1987",
       KEYWORDS = "differentiation arithmetic; sparse; optimisation.",
       ABSTRACT = "This thesis considers the problem of finding a minimum of a multivariate function
         with and without constraints. The truncated Newton method is used as the method for finding the
         minima, see [Dixon1988New]. In the constrained case, the Di Pillo Grippo penalty function is used.
         As these methods require first and second order tensors, differentiation arithmetic is used. A
         modification is made to the basic method described in [Rall1981ADT] which allows only vector storage
         to be used. Various modifications of the method are applied to real-life problems, and comparisons
         are given."
}


back
  

Contact:
autodiff.org
Username:
Password:
(lost password)