defining an objective without an analytic expression for fmincon

[Noam]

I want to solve a problem with an objective that doesn't have a closed-form analytic expression.  To give a concrete example, I have a set of 2D variables representing points on which I enforce a set of analytic constraints, and the function I want to minimize is intersection of these points' convex hull with some given shape. 

If I were to try and solve this optimization problem in Matlab, I would formulate the constraints and then write a function that computes objective on each iteration explicitly, by computing the convex hull, intersecting it with the given shape and computing the area. Can I somehow supply such a function that computes the objective to YALMIP and have it call fmincon with that function as the objective callback?

[Johan Löfberg]

It is really not the intended use of YALMIP which is eveloped for structured optimization. If you have such a complicated unstructured function, isn't it much easier to call fmincon directly by simply passing the function pointer to fmincon

There is a hack called https://yalmip.github.io/command/sdpfun/, but really, if you use this, you typically shouldn't use YAMIP to begin with

[Noam]

Thank you, Johan. I completely understand this is not YALMIP's original goal. Alas, my constraint-set is so easy to formulate in YALMIP and since I am still trying different ideas for these constraints and constantly modifying them, this chimera can still be very useful for me. Thanks for the tip!

[Johan Löfberg]

OK. You'll see if it works