What is an example of a manifold $M$ with $\dim(M)>1$ whose Lie algebra $\chi^{\infty}(M)$ of smooth vector fields admit an elliptic operator $D:\chi^{\infty}(M)\to \chi^{\infty}(M)$ such that $D$ is a Lie algebra derivation on $\chi^{\infty}(M)$?

Does every manifold admit such an operator?

Is there a Riemannian manifold for which the Laplace operator $D=\Delta$, naturally defined on $\chi^{\infty}(M)\simeq \Omega^1(M)$, would be a derivation of $\chi^{\infty}(M)$?


First a classical result about elliptic operators: in dimension $\ge 3$ the order of elliptic operators is even. In dimension 2, say in $\mathbb R^2$ you have elliptic vector fields such as $$ \bar \partial=\frac12(\partial_x+i\partial_y),\quad \partial=\frac12(\partial_x-i\partial_y). $$ Of course vector fields are derivations and I suspect that, since a derivation on a manifold must be a differential operator, that differential operator must be first-order from Leibniz' formula and thus should be a vector field since no mutiplication by a (non-zero) function is a derivation.

  • $\begingroup$ Thanks for your answer. Is there a complete classification of all derivations of $\chi^{\infty}(M)$?(as a vector field analogy of the fact that every derivation of $C^{\infty}(M)$, the space of smooth functions on $M$ is in the form $f\mapsto X.f$ for a unique vector field on $M$? what is your argument that every derivation on the space of vector fields is of first order?What is your argument that every elliptic operator on the sections of a vector bundle of rank greater than 1 is off even degree? $\endgroup$ Apr 27 '19 at 18:20
  • $\begingroup$ Finally is not the operator $(u,v,w)\mapsto (u_x-v_y,u_y+v_x,w)$ counted as an elliptic operator? $\endgroup$ Apr 27 '19 at 18:24
  • $\begingroup$ Do not you think that the classical result you mentioned is about "functions" not vector fields?what is a proof of that classical result(or a reference, please)? $\endgroup$ Apr 27 '19 at 18:30

I would like to apply the hints presented in the answer by Bazin to give the following answer.

Every derivation of $\chi^{\infty}(M)$ is inner and it is obvious that every inner derivation is a non elliptic operator. The reason is that every non vanishing vector field is locally look like $\partial/\partial_x$, whose corresponding adjoint operator is obviousely non elliptic.

So the non ellipticity of a derivation operator lies in its adjoint-ness not merely on its order.It is first order but this order situation is not enough to conclude it is non elliptic. In fact there is an example of a first order elliptic pde in dim 3 as follows: $$D(u,v, w)= (u_x-v_y, u_y+v_x, w)$$

This operator $D$ satisfies the definition of an elliptic PDE but is of first order.

One can construct a first order elliptic PDE in dimension 4 without a term of order 0. Construct a linear PDE on $\chi^{\infty}(\mathbb{R}^4)$ whose principal symbol is the $4\times 4$ matrix representation of quaterniouns $$\xi_1+\xi_2 i + \xi_3 j+\xi_4 k$$ where $(\xi_1,\xi_2,\xi_3,\xi_4)$ is a cotangent vector.

So it seems that the first line of the answer by Bazin does not work.

Remark: The above linked paper contains a reference to a result by F.Takens that every derivation of $\chi^{\infty}(M)$ is an inner derivation.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.