On Davidon's collinear scaling algorithms for optimization
\textit{W. C. Davidon} [SIAM J. Numer. Anal. 17, 268-281 (1980; Zbl 0424.65026)] presented a new class of algorithms for unconstrained optimization. These algorithms are related to quasi-Newton methods, and are based on local collinear scalings and local conic approximations, which extend local affine scalings and local quadratic approximations used in derivations of quasi-Newton methods. The algorithms of Davidon have not been studied theoretically or computationally. Algorithms based on collinear scalings and conic approximations related to quasi-Newton methods that have been investigated todate are all derived based on an approach by \textit{D. C. Sorensen} [ibid. 17, 84-114 (1980; Zbl 0428.65040)]. In this note, the author shows that all these algorithms derived using the approach of Sorensen are in fact different from the algorithms that Davidon presented.
- A Large-Step Analytic Center Method for a Class of Smooth Convex Programming Problems
- Collinear scaling and sequential estimation in sparse optimization algorithms
- Conic Approximations and Collinear Scalings for Optimizers
- Deriving collinear scaling algorithms as extensions of quasi-Newton methods and the local convergence of DFP- and BFGS-related collinear scaling algorithms
- scientific article; zbMATH DE number 3799576 (Why is no real title available?)
- Least-Change Secant Updates of Nonsquare Matrices
- Local andQ-superlinear convergence of a class of collinear scaling algorithms that extends quasi-newton methods with broyden's bounded-⊘ class of updates† ‡
- Quasi-Newton Methods, Motivation and Theory
- The Q-Superlinear Convergence of a Collinear Scaling Algorithm for Unconstrained Optimization
This page was built for publication: On Davidon's collinear scaling algorithms for optimization
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q1332350)
