I took my first graduate course as an undergrad last semester and it seemed that graduate student exams are much more open ended. Not only do they require you actually know more stuff (i.e. remember the theorem, the general approach) but they also require a lot more creativity, time management and general problem solving smarts.
But flipping open a text at the graduate level (Birkhauser, Springer-Verlag type), not only are most books filled with theorems, lemmas, proofs etc from head to toe, covering a very large array of concepts, but the much needed exercise at the end of the chapter is replaced with a list of references. Where are the exercises? And the course I took was sort of commonly taught I believe - stochastic networks.
I can see how this can be very problematic as one climb up the academic ladder. How would you ever find a text on "lower dimensional topology of autonomous robotic system", "Brain chaos in semi-aquatic faunas" or "Renal-sarcomere interaction theory" (three completely made up course).
How should graduate students approach a subject or an emerging field where there exercises scarce aside from what is introduced during the lecture or in existing literature?