Compiled Matlab Applications in AFNI

Several AFNI Matlab applications are now available as compiled applications that do not require Matlab. These include:

McRetroTS (a compiled version of RetroTS)

GroupAna (group analysis for 4 or 5 way ANOVA)

ExamineXmat (Design matrix graph displays)

Applications have been compiled explicitly for 64-bit Linux Intel, Mac Intel and Mac PPC platforms. Other platforms are not supported (though users are welcome to compile their own if they choose). These applications are all available here:

http://afni.nimh.nih.gov/sscc/staff/glend/matlab_compiler/

Please use the version appropriate for your platform. Each application requires its own compiled version and the corresponding ctf file. Both are contained in the platform tgz file. Additionally you will need to install the MCR (Matlab Compiler Runtime) libraries if Matlab is not already installed on your system. Then two environment variables (LD_LIBRARY_PATH  for Linux or DYLD_LIBRARY_PATH for the Mac and XAPPLRESDIR) will need to be set to refer to those libraries. The MCR library installer is also downloadable separately from the same location. For the Mac platforms, this is a .DMG file, and for the Linux-64 platform, this is a binary file. In either case, install the MCR library and then edit the example setup script provided for each platform or the user startup file to point to the installation of the MCR library.

The first time a compiled application is run (or updated with a new version), the .ctf file for the application is expanded creating a new directory that contains an encryped copy of all the Matlab functions used.

These compiled Matlab applications will run a little slower than the native Matlab equivalents because a "headless" version of Matlab needs to be started for each time the application is run, but that should be the only difference in application speed. The compiled versions do offer some advantages. The first advantage is obvious; the user does not to purchase Matlab or a Matlab Toolbox (the Statistical toolbox and the Signal Processing toolboxes are used by these applications). The second advantage is these applications are more easily scriptable (though Matlab can be scripted from a shell command line too). Another advantage is the user version of Matlab need not match the distributed version and doesn't need to be updated for AFNI functionality because the MCR library should provide a current enough version of Matlab.

For more info on McRetroTS or for some more specifics on installation of the compiled binaries, see this page: