Hello Everybody,

I'm looking for something specific: a good book on numerical analysis in C, using GNU/Linux. Any recommendations?

I'd like something fairly mathematical and technical; I got Ivor Horton's "Beginning C" book and it was obviously not written with people like me in mind.

It doesn't matter how old it is, and if it covers C on Unix, that's good enough.

Let me know what you think,
Joel

I'm not completely sure what you're looking for, but the "Bible" has long been "Numerical Recipes in C":

http://www.amazon.com/Numerical-Reci...6799529&sr=1-2

It's completely agnostic as far as platform.

CAVEAT : I've got the old 2nd edition (C-oriented). I've not seen the new 3rd edition (C++), and I can't vouch for it one way or the other.

Neither am I. Now that I think about it some more I'm just looking for a more advanced book on C, that has info about GNU/Linux-specific stuff (e.g. libraries, dlopen() etc.), and covers advanced topics germane to numerical analysis, like parallel computing. For example,

GNU/Linux Application Programming
http://www.amazon.com/GNU-Linux-Appl...816789&sr=11-1

is a good book, I've looked at it before, but I think its focus is applications, not numerical stuff. Correct me if this is a false distinction

Thanks,
Joel

Hi -

Now I'm *really* confused ;-)

SUGGESTION:
1. If you need some "C" code for linear equations, calculus or other "numerical analysis" stuff, "Numerical Recipes in C" is arguably your book. For what it does, it's great (at least my old 2nd Edition was great - I haven't seen the 3rd edition, I don't know one way or the other).

"Numerical Recipes in C, 2nd Edition", is freely available online:
http://www.nrbook.com/a/

2. If you're writing a subroutine to compute an integral or an FFT, it totally doesn't matter whether you're running Linux, Windows, or any modern platform. "Numerical analysis" is platform-agnostic (IMHO).

3. If you want something to get more "in depth" in your Linux programming skills, then your links sound like great books. Here's another suggestion, if you wish:

http://www.amazon.com/Programmers-To...ref=pd_sim_b_6

4. Finally, if you want to better understand issues with multiple CPUs (and how to better exploit "parallelization" in your own code), this is a truly outstanding book. Difficult (not for the faint of heart!), but useful:

http://www.amazon.com/Art-Multiproce.../dp/0123705916

'Hope that helps .. PSM

Nevertheless, you've provided some really great recommendations.

Let me just clarify:

I was already aware of Numerical Recipes, and I had a copy of it in my hand at the library (the very edition you speak of), but I couldn't get over to the Unix-specific section before I had to leave...nor could I check out that book or another one that looked exactly like what I wanted.

That I'm aware of also, however the C books I've encountered (with a few exceptions) are mostly like other programming books, going on and on about Windows when I don't use Windows. They tell me very little that's useful other than a few concepts about the language (like I said, I'm a mathematician, so a book like Beginning C was really below my expectations).

It's not the numerical analysis that I expect to be GNU or Unix-specific, but the hints on using the compiler, debugger, etc. I now have a book for that (C in a Nutshell). The problem is that most books that I've seen that cover that stuff don't say anything about numerical analysis, and they have really impractical examples (things like averaging a set of numbers). I would doubt they would say much about optimizing for numerical routines (e.g., with options specific to gcc) or parallel computing if the book is mainly about something other than numerical analysis.

The best I was hoping for was to get it all in one package. However, I've got the second best thing, with Numerical Recipes, and C in a Nutshell, I think I can hack it.

It sure does.

Thanks
Joel

for rather strange values of "freely". Details are here.

An alternative is here.