Introduction to Matrix Computations

by Gilbert W. Stewart

Numerical linear algebra is far too broad a subject to treat in a single introductory volume. Stewart has chosen to treat algorithms for solving linear systems, linear least squares problems, and eigenvalue problems involving matrices whose elements can all be contained in the high-speed storage of a computer. By way of theory, the author has chosen to discuss the theory of norms and perturbation theory for linear systems and for the algebraic eigenvalue problem. These choices exclude, among other things, the solution of large sparse linear systems by direct and iterative methods, linear programming, and the useful Perron-Frobenious theory and its extensions. However, a person who has fully mastered the material in this book should be well prepared for independent study in other areas of numerical linear algebra.

441 pages, Hardcover

First published May 28, 1973

Reviews

[Nick Black]

Amazon third party 2009-07-22. I don't complain about $0.25 textbooks.

You know, it seems that a lot of large third-party vendors seem to price their wares on Amazon based off the other offers listed. Over years of buying obscure (ie this doesn't apply to actively used material) scientific textbooks, I've noticed that either:

a) All copies are priced reasonable-to-high, or
b) There's an absurd peak of one or more copies (usually more than 1) at some extreme bargain price -- basically a bimodal distribution with blooming at the top end.

Furthermore, a large number seem the product of estate sales (thus seeding many ridiculously low prices via liquidation agents). Conjectures:

On average, a single highly-undervalued offering will be picked up before I find it, explaining very few singleton bargains of the extreme variety. Sometimes I get lucky and find the exception -- likewise, sometimes a merchant gets unlucky and prices based off the exception. Now, under the Law of Large Numbers this wouldn't hold water, because multiple merchants hitting the unlucky case before it's removed would be extremely unlikely (a classic exponential). BUT, we're not operating under large numbers, just a few nutty scientists with phat cash and dreams of phat libraries. So once one or two people get their copy, there's much less pressure on the bargain price, and it can multiply.

So this kind of works out like Conway's Game of Life; if a price is low, on the next turn there will be a new copy next to it, but a glider might smash things along the way and it fizzles out (I would of course have liked to use a nuclear simile, but one tries to avoid using "glider" and "reactor" in the same sentence, whereas explosions are clearly a non-linear process).

[KAl]

Good