Numerical Recipes in C & C++ Source Code CD-ROM with Windows, DOS, or Mac Single Screen License

Rating:
Summary: This is the one of the worst examples of coding I've seen
Review: I just got my copy of the CD today. I was surprised to see that they had stripped off all the code comments provided in the books. All that is left is the terse library of routines with the single character variables we loved to use as a BASIC beginner.

Buy this CD for the licence rights, but download the PDF chapters from the NR web site to get the commented code.

Rating:
Summary: The familiar Numerical Recipies routines without so much C
Review: The numerical recipes books have always provided minimal but usable routines for most numerical applications. The C++ conversion supplies routines that have been barely changed from the C versions, but uses C++ to give you the minimum that C++ programmers demand: automatic memory management and freedom from scary macros and side effects . The authors also updated the code to support double precision floats and complex numbers where the C version does not. The implementation of vectors and matrices is so minimal that even matrix multiplication and addition are not supported. All manipulation is element by element. This may make the routines less readable but has one big advantage: you can plug any vector or matrix library into these routine in place of their built in ones and expect everything to work instantly.

This is NOT free software - the license is relatively friendly to commercial software but not to open source software.

While there are other libraries that seem to be much more friendly to open source development than the NR library, I still think the numerical recipes books and libraries are useful to people writing open source software. You can't use the library directly, but where else are you going to find examples of minimal implementations of numeric algorithms? The code is absolutely modular, so you do have some hope of being able to decipher it.

Anyway even the NR routines are based on older routines so the authors admit in the license that comes with the book that they can't claim to own the algorithms they use.

Basically the new version of the code has been improved to the point where there's no need to run screaming, and for quick and dirty math routines, that's good enough for my use anyway.

Rating:
Summary: The familiar Numerical Recipies routines without so much C
Review: The numerical recipes books have always provided minimal but usable routines for most numerical applications. The C++ conversion supplies routines that have been barely changed from the C versions, but uses C++ to give you the minimum that C++ programmers demand: automatic memory management and freedom from scary macros and side effects . The authors also updated the code to support double precision floats and complex numbers where the C version does not. The implementation of vectors and matrices is so minimal that even matrix multiplication and addition are not supported. All manipulation is element by element. This may make the routines less readable but has one big advantage: you can plug any vector or matrix library into these routine in place of their built in ones and expect everything to work instantly.

This is NOT free software - the license is relatively friendly to commercial software but not to open source software.

While there are other libraries that seem to be much more friendly to open source development than the NR library, I still think the numerical recipes books and libraries are useful to people writing open source software. You can't use the library directly, but where else are you going to find examples of minimal implementations of numeric algorithms? The code is absolutely modular, so you do have some hope of being able to decipher it.

Anyway even the NR routines are based on older routines so the authors admit in the license that comes with the book that they can't claim to own the algorithms they use.

Basically the new version of the code has been improved to the point where there's no need to run screaming, and for quick and dirty math routines, that's good enough for my use anyway.