After discussion about linearization of differential operators here, I found another problem. That is, the uniqueness of extension of nonlinear operator.
Here, I consider especially nonlinear differential operator $L: \Gamma(E) \rightarrow \Gamma (F)$ where $E, F$ is vector bundles of compact Riemannian manifold $M$
T. Sunada for example says in Sunada that we can extend differential operators of order $m$ (naturally?) to $ H^{k,p}$ if $k> \frac{n}{2} + m$
When I ask Gemini , it said that has something to do with "Sobolev algebra".
Does anyone have any idea about this?
Thank you.