On averaging and representation properties of the BFGS and related secant updates (Q747768)
From MaRDI portal
 | This is the item page for this Wikibase entity, intended for internal use and editing purposes. Please use this page instead for the normal view: On averaging and representation properties of the BFGS and related secant updates |
scientific article; zbMATH DE number 6495756
| Language | Label | Description | Also known as |
|---|---|---|---|
| default for all languages | No label defined |
||
| English | On averaging and representation properties of the BFGS and related secant updates |
scientific article; zbMATH DE number 6495756 |
Statements
On averaging and representation properties of the BFGS and related secant updates (English)
0 references
19 October 2015
0 references
The author considers the secant method for the general quasi-Newton iterative procedure in which the function to be optimized has a derivative which is symmetric and positive definite. In unconstrained optimization, the eigenvalues of the successive Hessian approximations become progressively smaller and eventually converge to zero. The standard approach to rectify the matter is to restart the algorithm using a fresh Hessian approximation, referred to as secant update. The paper displays several representation theorems for various classes of secant updates.
0 references
Newton's method
0 references
secant methods
0 references
quasi-Newton methods
0 references
optimization
0 references
BFGS method
0 references
Schnabel class
0 references
Dennis class
0 references
symmetric rank-2 updates
0 references
0 references
0 references
0 references
0 references
0.7990001440048218
0 references
0.7964490056037903
0 references
0.7919536828994751
0 references
0.7915086150169373
0 references
