Asymmetric minimization with a convex fourth degree approximation
作者:
J.VAN REMORTEL,
D. J. WILDE,
期刊:
International Journal of Systems Science
(Taylor Available online 1976)
卷期:
Volume 7,
issue 1
页码: 39-64
ISSN:0020-7721
年代: 1976
DOI:10.1080/00207727608941899
出版商: Taylor & Francis Group
数据来源: Taylor
摘要:
The Newton-Raphson method is a basic technique in unconstrained minimization which exactly minimizes a quadratic, symmetric approximation to an objective function. Functions with some degree of asymmetry can be optimized more efficiently by exact minimization of asymmetric approximations. In this paper, a special separable convex asymmetric fourth degree function is proposed whose minimum can Vie found in a finite number of steps, despite the high degree of the approximation. The method may be regarded as a. quasi-Newton method with a few operations added to measure and correct for asymmetry.
点击下载:
PDF (724KB)
返 回