Agreed that: A) it and Scilab are great projects B) these would suffice fine for pretty much any undergraduate course in linear algebra, but many courses need Simulink and I'm not sure Scilab's Xcos is good enough. Since Matlab is like $20 for students, it is basically free and makes sense especially when you consider the IDE. Sure, Python and Julia are great options once you can code. There are many students who honestly just aren't very good coders by their sophomore and junior years. Engineering students have to code here and there in various projects, but often don't have a full class just for programming. Long story short, inverting a matrix in Matlab/Octave/Scilab is more straightforward than in Python. I think Julia's syntax has enough similarities with Matlab and is similar, but the tooling would be confusing to students.