1. (Path Replication) I don't know where the paper was published or when. But there's a LOT of literature for the topic. 2. (Control Flow Analysis for Nested Functions) This seems the most interesting, but -- since the paper by Olin Shivers in 1988 -- there have been a lot of works on the field. I'd fear that you might possibly waste time. 3. (Partial Inlining of Local Functions) You'd primarily need a good set of applications which you could use to evaluate this. 4. (Semantic Composition of OpenGL Shaders) This is outside of my field of competences, I think. 5. (Cross-Distribution of Opaque Binary Applications) I can't gauge whether this is a pure implementation work, or if there is a constrained problem that should be solved which is interesting outside of the implementation. Unfortunately I've noticed in the last few weeks that I'm completely overloaded and don't have the time to take on any more work. In particular, I can only take on the work if I can use one of my assistants to handle the "front line" work. This means that you'd have to find a thesis which can be supervised by one of them. I suggest you contact them directly. If you can find another group that can supervise one of the above topics however, then I think this would be good for you.