Octave

Matlab's free linux brother is quite a life saver. Although it has not turned to be as fast as Matlab for me somehow.

I run a modified code borrowed originally from Tim Baileys' website for running an EKF-SLAM simulation on robot whose motion model is differential.

Mainly intended for Fedora Core users

The easiest way for installing octave would be

yum -y install octave*

although octave-forge would suffice for most common purposes.

For those who are interested in the source kindly download from here

I have had quite a number of hiccups in working with octave. GNUPlot does not live up to the front end of Matlab. Moreover, the code itself without the GUI runs much slower on octave.

However, I found that using ATLAS libraries will quicken the code. I have however not delved into the intricacies of why and how. Installing ATLAS libraries and making octave use them from yum did not work. However, downloading the source and configuring using this works for me. (this is borrowed from another site btw)

./configure --enable-shared --enable-dl --disable-static LDFLAGS="-L/usr/lib64/atlas"

where the "/usr/lib64/atlas" is the path of the installed ATLAS libraries.