It comes up all the time when trying to build second order optimization methods. With the eigensystem of your objective in hand you have a complete understandin

It comes up all the time when trying to build second order optimization methods. With the eigensystem of your objective in hand you have a complete understanding of the (non-) convexity of your energy landscape, which is useful to ensure you always have good search directions, etc.