Publication: The Truncated Newton Method for Sparse Unconstrained Optimisation Using Automatic Differentiation
Introduction
Applications
Tools
Research Groups
Workshops
Publications
   List Publications
   Advanced Search
   Info
   Add Publications
My Account
About
Impress

The Truncated Newton Method for Sparse Unconstrained Optimisation Using Automatic Differentiation

- Technical report -
 

Author(s)
Lawrence C. W. Dixon , Richard C. Price

Institution
The Numerical Optimisation Center, Hatfield Polytechnic

Year
1986

Abstract
A method is presented which solves unconstrained optimisation problems using a truncated Newton method. Automatic differentiation is used to calculate the derivatives required. By taking advantage of the structure of the optimisation method, only vector storage is required since the Hessian is multiplied by a vector during the computation of the derivatives. Numerical results obtained by solving standard test problems are presented.

BibTeX
@TECHREPORT{
         Dixon1986TTN,
       AUTHOR = "Dixon, Lawrence C. W. and Price, Richard C.",
       TITLE = "The Truncated {N}ewton Method for Sparse Unconstrained Optimisation Using Automatic
         Differentiation",
       TYPE = "Technical Report",
       NUMBER = "NOC TR170",
       INSTITUTION = "The Numerical Optimisation Center, Hatfield Polytechnic",
       ADDRESS = "Hatfield, U.K.",
       MONTH = "October",
       YEAR = "1986",
       REFERRED = "[Dixon1987ADa]; [Dixon1991UoA]; [Fischer1987AD].",
       COMMENT = "This paper was presented at the SIAM National meeting, Boston, 1986. Also to appear
         in J. Opt. Theory and Appl. 60(2), pp. 261--275, February 1989.",
       KEYWORDS = "point algorithm; numerical results; differentiation arithmetic; optimisation
         method.",
       ABSTRACT = "A method is presented which solves unconstrained optimisation problems using a
         truncated Newton method. Automatic differentiation is used to calculate the derivatives required. By
         taking advantage of the structure of the optimisation method, only vector storage is required since
         the Hessian is multiplied by a vector during the computation of the derivatives. Numerical results
         obtained by solving standard test problems are presented."
}


back
  

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