I came across the recent paper Convex optimization approach to discrete optimal control but, since my background in optimization is weak, I am having trouble parsing it.

My question is about its applicability. I have a potential application in which I have a nonconvex nonlinear optimization problem [a composite function of several real-valued vectors] that I can optimize successfully using gradient descent (at least getting a good local optimum). But I want to add in a control 'node' with a binary control vector that can be also be optimized with my existing gradient descent algorithm. Does this paper appear to be able to solve that problem? Or if not, does anyone know of how I might go about this?

  • $\begingroup$ Are you still interested in this question? $\endgroup$ Commented Nov 26, 2023 at 7:42


You must log in to answer this question.