One of the best collections of facts about numerical linear algebra algorithms to be found today. The presentation is more practical than rigorous, and the coverage is encyclopedic but somewhat terse. This is not a book from which to learn the material for the first time unless one has a very high level of mathematical maturity. A few mistakes are peppered throughout, but they are all well known and documented in the errata sheet. This work is valuable for both people developing code and people trying to make intelligent use of existing software. In short, this is a must own reference book for the desk of a practicing numerical analyst or scientist.
