Free Software for Numeric Computations

In this course, we use NAG routines to provide functionality in certain areas such as probability distributions and linear algebra. The NAG library is a proprietary product which must be licensed from NAG (Numerical Algorithms Group). However, most of the things we get from the NAG library can also be found in a variety of public domain sources, which are available for your use in whatever computing environment you may find yourself. The following list is by no means complete, but may get you started in finding some sources for reliable numerical software which you can legally use on your own computer without paying any fees.

There are two large electronic depositories for this kind of software which you should be aware of, namely netlib and statlib. The links point to the SCF help files for these two repositories.

For probability distributions, there is an excellent library, available through statlib, known as dcdflib, which has most of the common probability routines.

A number of people have compiled lists of useful free software for numerical computations. Among them are:

Index of resources for numerical computation, compiled by Ajay Shah. This focuses almost entirely on free software.
Numerical Evaluation of Special Functions by D. W. Lozier and F. W. J. Oliver. This has information about both free and commercial routines.
The GAMS Problem Decision Tree, developed by the National Institute of Standards and Technology. This presents a wide variety of software, both free and commercial, in a decision tree format, making it very useful when you have a specific problem in mind.
Freely Available Numerical and Statistical Software, a list compiled by Gordon Smyth at University of Queensland.
Graphics and analysis software, compiled by S. Baum at Texas A&M University, is a well-annotated list of packages and libraries for graphics and data analysis.
Free Scientific Software Under Linux. As the name implies, this list is oriented towards the Linux operating system, but some of the routines may be suitable for other flavors of UNIX. (If you'd like to know what Linux is all about, click here).