Why Broyden’s Nonsymmetric Method Terminates on Linear Equations
Author(s) -
Dianne P. O’Leary
Publication year - 1995
Publication title -
siam journal on optimization
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 2.066
H-Index - 136
eISSN - 1095-7189
pISSN - 1052-6234
DOI - 10.1137/0805012
Subject(s) - mathematics , linear subspace , iterated function , projection (relational algebra) , nonlinear system , variety (cybernetics) , dimension (graph theory) , forcing (mathematics) , set (abstract data type) , pure mathematics , algorithm , mathematical analysis , computer science , statistics , physics , quantum mechanics , programming language
The family of algorithms introduced by Broyden in 1965 for solving systems of nonlinear equations has been used quite effectively on a variety of problems. In 1979, Gay proved the then surprising result that the algorithms terminate in at most $2n$ steps on linear problems with n variables [SIAM J. Numer Anal., 16 (1979), pp. 623–630]. His very clever proof gives no insight into properties of the intermediate iterates, however. In this work we show that Broyden’s methods are projection methods, forcing the residuals to lie in a nested set of subspaces of decreasing dimension. The results apply to linear systems as well as linear least squares problems.
Accelerating Research
Robert Robinson Avenue,
Oxford Science Park, Oxford
OX4 4GP, United Kingdom
Address
John Eccles HouseRobert Robinson Avenue,
Oxford Science Park, Oxford
OX4 4GP, United Kingdom