In the late '90s I did first year undergraduate maths. As part of that, I did a pull-forward module from the second year on linear algebra. Or rather, I started it. Half-way through the course I discovered that I'd been failing to correctly distinguish the lecturer's us, vs and nus (this took a rather long time as he mumbled, so you had to reverse as much as you could from what he wrote on the blackboard). So, I gave up the course and never came back to it as I'd switched to computer science the next year.
Skip forward a decade and I'm catching up on lost maths. I have finally finished reading a Springer yellow book! On the downside, it's from their undergrad series. On the upside, it's written by Serge Lang, a famous Bourbakiste. On the downside, it is linear algebra, probably the simplest 'proper' maths. On the upside, at least it's not Lang's 'Introduction to Linear Algebra', a simpler text.
Moreover, I think I understood it pretty well. Conveniently I knew much of the material already, but I learnt plenty new and I now have a much stronger grasp of much of the material. I found the presentation very good. Strangely, the effective culmination of the book, a theorem about Jordan Normal Form, was by far the most terse and difficult to follow proof, for apparently no good reason! (I skipped the appendix on Iwasawa decomposition, which I may come back to after warming up my group theory.)
Would I recommend it? If you're somehow in the state of being reasonably mathematically literate but don't know linear algebra (a strange state!), yes.
Posted 2011-01-06.