Publication: Algorithmic differentiation techniques for global optimization in the COCONUT environment
Introduction
Applications
Tools
Research Groups
Workshops
Publications
   List Publications
   Advanced Search
   Info
   Add Publications
My Account
About

Algorithmic differentiation techniques for global optimization in the COCONUT environment

- Article in a journal -
 

Area
Optimization

Author(s)
Hermann Schichl , Mihály Csaba Markót

Published in
Optimization Methods and Software

Year
2012

Abstract
We describe algorithmic differentiation as it can be used in algorithms for global optimization. We focus on the algorithmic differentiation methods implemented in the COCONUT Environment for global nonlinear optimization. The COCONUT Environment represents each factorable optimization problem as a directed acyclic graph (DAG). Various inference modules implemented in this software environment can serve as building blocks for solution algorithms. Many of them use techniques based on various forms of algorithmic differentiation for computing approximations or enclosures of functions or their derivatives. The algorithmic differentiation in the COCONUT Environment not only provides point evaluations but also range enclosures of derivatives up to order 3, as well as slopes up to order 2. Care is taken to ensure that rounding errors are treated correctly. The ranges of the enclosures can be tightened by combining the evaluation routines with constraint propagation. Advantages and pitfalls of this method are also outlined.

BibTeX
@ARTICLE{
         Schichl2012Adt,
       author = "Schichl, Hermann and Mark\'{o}t, Mih\'{a}ly Csaba",
       title = "Algorithmic differentiation techniques for global optimization in the {COCONUT}
         environment",
       journal = "Optimization Methods and Software",
       volume = "27",
       number = "2",
       pages = "359--372",
       year = "2012",
       doi = "10.1080/10556788.2010.547581",
       url = "http://www.tandfonline.com/doi/abs/10.1080/10556788.2010.547581",
       eprint = "http://www.tandfonline.com/doi/pdf/10.1080/10556788.2010.547581",
       abstract = "We describe algorithmic differentiation as it can be used in algorithms for global
         optimization. We focus on the algorithmic differentiation methods implemented in the COCONUT
         Environment for global nonlinear optimization. The COCONUT Environment represents each factorable
         optimization problem as a directed acyclic graph (DAG). Various inference modules implemented in
         this software environment can serve as building blocks for solution algorithms. Many of them use
         techniques based on various forms of algorithmic differentiation for computing approximations or
         enclosures of functions or their derivatives. The algorithmic differentiation in the COCONUT
         Environment not only provides point evaluations but also range enclosures of derivatives up to order
         3, as well as slopes up to order 2. Care is taken to ensure that rounding errors are treated
         correctly. The ranges of the enclosures can be tightened by combining the evaluation routines with
         constraint propagation. Advantages and pitfalls of this method are also outlined.",
       ad_area = "Optimization"
}


back
  

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