Summary
We show that the following three conditions are equivalent for a generalized matrix normN onM k (C):
-
(i)
for eachx∈C k there is a vector norm · x (depending onx) such that for allA∈M k (C), Ax x ≦N(A) x x;
-
(ii)
for eachA∈M k (C) there is a vector norm · A (depending onA)_such that for allx∈C k,Ax A ≦N(A) x A; and
-
(iii)
the spectral radius ϱ(A)≦N(A) for allA∈M k (C).
We call generalized matrix norms satisfying (i) (and, thus, (ii) and (iii)locally compatible and cite examples, for instance, to show that a locally compatible generalized matrix norm is not necessarily globally compatible with any single vector norm.
Similar content being viewed by others
References
Householder, A.: The theory of matrices in numerical analysis. New York: Blaisdell 1964
Johnson, C.: Multiplicativity and compatibility of generalized matrix norms. Lin. Alg. and its Applics. (to appear)
Lancaster, P.: Theory of matrices. New York: Academic Press 1969
Stewart, G.: Introduction to matrix computations. New York: Academic Press 1973
Author information
Authors and Affiliations
Additional information
The author wishes to thank Professor G. W. Stewart and the referee for helpful remarks.
Rights and permissions
About this article
Cite this article
Johnson, C.R. Locally compatible generalized matrix norms. Numer. Math. 27, 391–394 (1976). https://doi.org/10.1007/BF01399602
Received:
Issue Date:
DOI: https://doi.org/10.1007/BF01399602