Hi friends,

I'm wondering if there is any paper that formally proves that the ILQR (not DDP) algorithm converges to a locally optimal solution. The proof seems possible since ILQR is basically a quasi-Newton method. However, the closest paper I can find is this one: Article Discrete LQR and ILQR methods based on high order Runge-Kutt...

But the paper is very new and the proof is rather brief and seems to rely on some hidden assumptions. Has anyone checked the correctness of this paper? I would appreciate any comments or suggestions regarding this matter!