Customer Reviews

Numerical Recipes in C++: The Art of Scientific Computing

5 star:		(6)
4 star:		(1)
3 star:		(5)
2 star:		(3)
1 star:		(11)

 

 

Virtually all code (with the exception of a rudimentary vector/matrix class library -- which, by the way, now features zero-based arrays, by popular demand) in the book is procedural, not object oriented. Thus, the title of the new edition is misleading, since the "++" features of C++ are not utilized.

On the other hand, this is *the* classical recipe book for numerical techniques for a wide variety of applications, and is a "must-have" for anybody looking for a good reference for numerical algorithms.

Bottom line: this edition will be just as useful as any of the other editions (C, Fortran, Pascal, etc) to a C++ programmer. If you are looking for numerical methods using object oriented programming, the only decent books to date are those by Daoqi Yang, and Barton & Nackmann (although the latter is a bit dated).

This book is nice, and contains all standard methods of numerical amalysis and linear algebra.

But!

There is book called "Numerical Recipes in C" that is basicly the same book, and it could be find at cornell university web pages for free

So, DON'T BUY THE BOOK - you can find practically the same book for free at the web!

The code in this edition in so-called C++ is just a copy-and-paste of the C edition code. All the functions have just been made members of ONE "NR" CLASS. The code presented brings none of the improvements that C++ could have brought. This turns this book into an ABSOLUTE [bad deal] as THE C EDITION IS AVAILABLE FOR FREE ... Once again, this would only be useful for the explaination of the algorithms since the code in the C edition is a rather inefficient transcription of Fortran into C.

This book was a good idea 20 years ago. However, if you want to understand numerical algorithms then you're better off using the WWW to find the pages you want OR buying a real Numerical Methods text. (Amazon has them all!)

If you are a developer who just needs the recipe then these aren't the best algorithms to use. The Gnu Scientific Library is a much better place to start, the code is more modern.

I returned this book. The licensing is very restrictive. The book comes with an "Immediate License" that allows you to type the routines into your computer and use for personal and noncommercial purposes. Any other use or distribution requires the purchase of an additional license.

Some of the routines (Quicksort, p 336) are not very well coded: a bunch of one letter variable names, loops that only exit on break.

Don't bother buying this book. It's a rip off. It is a mindless translation of the C version. C++ has, included in the STL, a complex class. You would think that an author of a numeric computing book would take advantage of such a class. Not these authors. Get this; they still require that the array passed to their FFT algorithms have the real and imaginary parts alternate in a real array. Oh, it does use zero based indexing now. Be still my beating heart.

This is very simply the only book that explains all essential numerical algorithms comprehensively and gives examples how one might implement them. The infinite number of reviews and discussions about the code can - in my eyes - be summarized as follows: it works, it lacks robustness, and you will find a better open-source alternative to virtually every single function. Some details that are worth knowing:
-A major advantage of the C++ over the C edition is that arrays are finally zero-based.
-This book doesn't include the source code in any digital form, so you will have to purchase the CD or online access if you don't want to spend hours typing and correcting typos.
-You will need the most recent update (2.11 I guess) for some algorithms to work correctly.
-It's not open source.
So in the end of the day I would recommend the book...
... if you are looking for a comprehensive explanation how an algorithm works - especially if you are not a mathematician or a computer scientist.
... if you just want a straight-forward implementation of an algorithm, and speed or robustness are not your major concerns.

I have dipped into the C version of NR since it came out almost ten years ago. It has been very useful. Having recently got the C++ version I may have had unrealistically high expectations. It is a good book, but I don't judge that it is that much better from the C version. All modern C++ compilers accept C programs so you can still get the formidable benefits of NR without moving to this version. I particularly disliked the size of the header file in the C++ version. It has to be included in all code (or else write your own individual header files) and since it runs to many thousands of characters, surely it was obvious that no one would spend time entering it? A bad decision for the reader and user, but one that definitely encourages the purchase of the downloadable code package.

Book contains the same numerical procedures as in recipes in C piled up in one class. Using this book is like using C without exploitation of object capacities on C++.
What I would recommend is to get the vector library Blitz++ and Numerical recipes in C; both are available for free on-line. Using Blitz++ you achieve speed of Fortran 90 and easy exposition of C++.
And that's what author of this book should have done, write all the procedures based on blitz++.

This book uses C++ only in its title.

The explanations are cursory and tend to overlook many of the points to be considered before applying one of their examples to real problems. While good at providing source code for an experienced signal processing engineer or mathematician, this book can be dangerous for those in search of a quick fix without the background to evaluate what they are really doing.