From the simonster version / EEGLAB: (caromk notes at end)

PLEASE READ THE LICENSE FILE FIRST - GNU GPL

SOFTWARE REQUIREMENTS:

      ANSI C compiler
      BLAS and LAPACK math libraries
      MATLAB version 5.0 or higher (*)


(*) MEX implementation only


HOW TO COMPILE:
The ICA software requires the BLAS and LAPACK libraries. A public domain implementation of the library packages is available from NETLIB at http://www.netlib.org/clapack/.
NOTE: hardware optimized implementations of the BLAS routines are available for most architectures and have proven to increase execution speed by 4-500%.

Makefiles have been constructed for several platforms. To compile the ICA software using an existing Makefile; modify the library paths to fit the paths on you system. Next, make the designated Makefile.

Example:
      make -f Makefile.alpha


If you compile the ICA software for multiple architectures, make sure to run a 'make-clean' between each make session.

Example:
      make -f Makefile.alpha clean


A Makefile must be constructed to compile the ICA software for a platform for which no Makefile exists. You may modify any of the existing Makefiles to fit your system.

Defining MMAP causes the ICA software to use memory mapping rather than memory allocation for data storage. Memory mapping allows the ICA software to return freed memory to the kernel, thus significantly decreases memory usage. However, memory mapping may not work on all systems.

Caromk practical notes:

1 - Most of the Makefiles require the CLAPACK libraries, which I found a huge
pain to get working and gave up. The Makefile.linux demonstrates how to do it
with LAPACK/BLAS/F2C, but is klugey, as it uses static links and an ignore flag,
     because I don't know how to make a proper Makefile.

2 - To use with MATLAB, you need to modify two line in binica.m. (a) Define ICABINARY
in the appropriate place for your system. (b) Define the name of the parameter file for your system in the SC variable. It should be in the same directory as binica.m. Also, (c) Add the directory containing this code to your Matlab path.