Coding Distributions as a Convex Constraint

34 Views Asked by At

In convex optimization, how can we impose a constraint that a variable has certain distribution? e.g. elements of vector $v$ have power law distribution?

1

There are 1 best solutions below

2
On BEST ANSWER

It is easy to see that in general such a constraint cannot be guaranteed to preserve convexity. For example, any non-continuous distribution or any continuous distribution which has a probability zero mid-region will very likely break convexity.