Customer Rating: 
Summary: Checkout GNU scientific library (GSL)
Comment: I totally agree with another reviewer that the biggest drawback of numerical recipes (NR) is that first you have to buy them to see whether you need it or not afterall. Linux these days is way more user-friendly than it used to be. If you have fedora or ubuntu, get gsl with one line
On fedora:
yum install gsl gsl-devel
On ubuntu:
sudo apt-get install gsl-bin gsl-ref-html
That's it, people! What can be harder!
GSL documentation is free online, but you can order a book as well.
GNU Scientific Library Reference Manual - 2nd Edition
P.S. also GSL library can be used together with c++ code.
Customer Rating: Summary: talk about outdated
Comment: this book was likely a looker back in the day, but its 2007 now. Need to have better details for non "C"-users. wish i had bought "Idiots Guide to C".
Customer Rating: Summary: A classic book of numerical algorithms
Comment: This book, although published 15 years ago, is still very useful. In fact, its more recent counterpart "Numerical Algorithms in C++" is a mess, and I wouldn't recommend it to anyone. The explanations of the algorithms that occur in each section of this book are top-notch. It helps with such questions as "Sure you know how to evaluate an integral with pencil and paper, but how do you do it with a computer?" Everything from linear algebra techniques to integration and evaluation of functions to the FFT and spectral applications are explained clearly and coded up in C. The code is great too, with the exception of one problem that several reviewers have already mentioned - the author has a FORTRAN-like programming style in which each implementation has arrays going from 1 to n versus 0 to n-1. This does cause some implementation problems if you want to transfer the algorithms into another programming language. Overall, though, I can't think of one book that does all of the heavy lifting that this one book does as well as it does in the arena of numerical algorithms.
The book is now available online. Just type "Numerical Recipes" into Google and click on the Numerical Recipes Home Page to peruse the entire book free of charge. You might also find the "Numerical Recipes in C Example Book" useful. That book is simply the source programs that demonstrate all of the Numerical Recipes subroutines. Each example program contains comments and is preceded by a short description of how it functions. I know I found it helpful in many cases.
Customer Rating: Summary: Very nice book
Comment: A must buy for students or researchers who need numerical methods. Comprehensive topics. A good place to start to deeper levels. Online book is good for quick look.
Customer Rating: Summary: A classic, and still worth having
Comment: "Numerical Recipes" has been a staple in computing libraries for many years, and for good reason. It provides immediately usable implementations of all the workhorses of numerical computation, in production-quality form. Maybe there are better implementations out there, FFTW for example, but getting something to work correctly always comes before getting it to work fast. Numerical computation is a specialty, and vanishingly few of us are specialists. As a result, getting this much specialist knowledge for the price of a very few hours' wage, fully debugged and documented, is a great bargain.
I have to agree with the critics who point out that the Gnu Scientific Library (GSL) is more complete in some areas, and offers better licensing terms. This collection has its own strengths, though, and not just in documentation. The writeup, however, is the major interface between the software and us, the bio-ware. GSL's collection of 'man' (help) pages serves a purpose, but this book's exposition describes a lot more of the background and rationale for the routines. The code and man pages are self-evident statements of the implementation - but "what" is a very different question than "what else" or "why."
This one may not serve all needs. You'd be amazed how many it does serve, though. If you need more than a Matlab session for numerical computing, you need this.
//wiredweird
