5

Let $E\to M$ be a smooth vector bundle over a smooth manifold $M$.

Let $D$ be a differential operator defined on the space $\Gamma(E)$ of smooth sections of $E$. We fix a section $s\in \Gamma(E)$.

Assume that for every $x\in M$ there exist a natural number $n$ such that $D^n(s)(x)=0$.

Does this imply that $D^k(s)=0$, for some $k\in \mathbb{N}$?

If the answer is "No", what about if we assume that $D$ is an elliptic operator and $M$ is a compact manifold?

The motivation for this question is the following fact which I learned from the book of "R.P.Boas:A primer of real functions".

https://www.jstor.org/stable/10.4169/j.ctt5hh8x5

Fact: Assume that $f:\mathbb{R} \to \mathbb{R}$ is a smooth function with the property that for every $x\in \mathbb{R}$ there exist a natural number $n$ with $f^{(n)}(x)=0$.Then $f$ is a polynomial.

  • (A version of) the real analytic result you mention is discussed in this MO question: https://mathoverflow.net/questions/34059/if-f-is-infinitely-differentiable-then-f-coincides-with-a-polynomial – Sam Hopkins Apr 24 '19 at 02:59
  • @SamHopkins Thanks for your comment.That is about smooth finction. Among answer one can find a reference to Boas book. I learned this result from the book of Boas when i was a master student any way the link you mentioned is about that classical theorem not about sections of vector bundles. – Ali Taghavi Apr 24 '19 at 05:23
  • 2
    right, I just mention to link that in case other people were interested in learning more about the classic result you mentioned. – Sam Hopkins Apr 24 '19 at 13:22
  • @SamHopkins yes I see thanks for the link. – Ali Taghavi Apr 25 '19 at 06:03

0 Answers0