[7/12/23 ] Formulating the problem of finding the largest convex subset (in a way that can be computed) is difficult in general, and I think that limiting the subsets to a certain shape could make things easier.
\(n\)-ellipses, ellipses with \(n\) focal points, are always convex, and versatile as \(n\) varies.
The resources I intend to use are:
The item by Erdös is exactly what I’m looking for and it suggests that \(n\)-ellipses might be able to approximate any convex, closed plane curve.