Eduard L. Stiefel
(April 21, 1909 - November 27, 1978)
Virtual Workshop
on
Advanced Linear and Nonlinear Conjugate Gradient and Related Methods
in honor of
Eduard Stiefel's
birthday centenary
The conjugate gradient algorithm which is now more than 55 years old includes a class of very elegant unconstrained optimization algorithms characterized by strong local and global convergence properties and by low memory requirements.The history of these algorithms begins with the research of Eduard Stiefel at Eidgenössische Technische Hochschule Zürich - Swiss Federal Institute of Technology, and of Cornelius Lanczos (1893-1974) and Magnus Hestenes (1906-1991) at the Institute for Numerical Analysis (National Applied Mathematics Laboratories of the United States National Bureau of Standards in Los Angeles).
The purpose of this Virtual Workshop is to celebrate the birthday centenary of Eduard Stiefel, who first designed this class of algorithms. In 1951 he established the n-step method for solving systems of algebraic linear equations with symmetric and positive definite matrix, published as: Über einige Methoden der Relaxationsrechnung. Z. Angew. Math. Phys., 3 (1952), pp.1-33 (submitted in July 1951). Invited by Olga Taussky-Todd (1906-1995), he visited the Institute for Numerical Analysis (INA) of the National Bureau of Standards - UCLA, Los Angeles, from July 1951 until February 1952. There he found out that Magnus Hestenes in collaboration with other scientists from the INA (Lanczos, Forsythe, Motzkin, Rosser, Stein) had discovered the same method too. Hestenes called it the conjugate gradient method. He published it as iternal report: Iterative methods for solving equations, NAML Report 52-9, National Bureau of Standards, Los Angeles, CA, July 1951. During this visit Hestenes and Stiefel combined their efforts and published a 28-page two-columns conjugate gradient paper as: Methods of Conjugate Gradients for Solving Linear Systems, Journal of Research of the National Bureau of Standards, vol.49, No. 6, December 1952 (Research Paper 2379), being one of the most influential papers in numerical analysis. Some insights into this paper were made by Dianne P. O'Leary. Since then over 40 nonlinear conjugate gradient algorithms have been established, and practically an innumerable number of papers have been published on this subject, insisting both on their theoretical and practical aspects. An excellent survey of the development of different versions of nonlinear conjugate gradient methods, with special attention to global convergence properties, is presented by Hager and Zhang.
Call for Papers
The purpose of this Virtual Workshop is to honor Eduard Stiefel's birthday centenary by providing an opportunity to researchers to present and discuss different topics on recent developments in the field of Linear and Nonlinear Conjugate Gradient and their applications in optimization community and industry. The idea is that of promoting scientific exchange and consider possibilities of further cooperation through network facilities.
All the contributions will be placed on a special site dedicated to this Virtual Workshop. The access to the contents of papers is based on username and password. These are allocated only to contributors after having submitted paper(s) as PDF files to the following address: nandrei@ici.ro
The contributions should emphasize all aspects of Linear and Nonlinear Conjugate Gradient and Related Methods referring to: new conjugate gradient algorithms, convergence analysis, computational results and comparisons, complexity, applications, open problems etc.
International Workshop Committee:
The deadline for submission is January 31, 2009.
We believe this Virtual Workshop will have astrong impact on the development and consolidation of linear and nonlinear conjugate gradient methods, identifying emerging subjects of research and promoting the collaboration among researchers working in this area.