Why try to optimize Octave for ARM? Instead you should spend time optimizing the {BLAS, LAPACK, FFT, sparse matrices, etc.} libraries on which Octave depends. I don't think there is a lot in the Octave source code itself that would benefit from a lot of time spent on optimization. For the most part, if there are processor optimized version of the aforementioned libraries, then Octave will perform quite well for vectorized code.

I have been contributing small bit to Octave for a couple of years now. I can attest to the fact that the core Octave developers are helpful, do good work, and are always looking for more help. Currently I work on two different parts: an IDE for Octave called OctaveDE (which I am looking for more help on, but is more UI work than math work) and I am rewriting the image reading and writing routines to be Matlab compatible. There is also a lot of OpenGL work going on implementing a faster plotting backend as an alternative to GNUPlot. Because Octave is ofttimes in "Matlab catchup mode", any help would be greatly appreciated and you would be given direction on the best way to contribute early while you get used to the code base and the way the core developers work.

http://blogs.netapp.com/dave/2007/09/netapp-sues-sun.html The victim has become the offender.

As for whether you own your own ideas or not, many companies make you sign away patent right for anything you come up with (while working for them, whether on the job or not) that is related to their business interests. Sometimes very skilled or well-known people can weasle their way out of clauses like this, but for the most part you probably signed away your right to patent tech stuff when you signed you offer acceptance letter. go read it.