How to handle quadratic objectives with SDDP.jl?

Hello everybody.
I have to solve a problem similar to an economicd dispatch or unit commitment and I have seen that a good alternative is the Stochastic Dual Dynamic Programming and I was thinking about using SDDP.jl.

However my objective to maximize is a quatratic objective, of the style (x-Param)^2 and the rest of constraints are linear
My doubt is if SDDP.jl guarantees convergence for this kind of problems.

I think the problem is convex, but I am not sure.

Thank you very much and best regards

1 Like

Hi @Samuel_M welcome to the forum :slight_smile:

If the problem is convex, yes.

If the problem is not convex, most probably not.

Note that max x^2 is not convex.