8
$\begingroup$

Suppose that $M$ is a smooth manifold with a measure $\mu$ and let $L^2(M, \mu)$ be a space of all square-integrable functions on $M$.

Recall that $L$ is a hypoelliptic differential operator if for every $f \in \mathcal{D}(L)$, if $Lf$ is in $C^\infty(M)$ then $f$ is also in $C^\infty(M)$.

Could anyone give a reference regarding the example when $L$ is hypoelliptic but its adjoint w.r.t to $\mu$ is not hypoelliptic? Could this happen to the Hörmander operators, when $L$ is defined as $$ L = \sum_i X_i^2 + X_0 $$ and the Lie algebra generated by $\{X_i\}$ spans the entire tangent space?

I am mostly interested in the case when $\mu$ is induced by the Riemannian metric and $M$ is complete.

Thanks!

$\endgroup$
6
$\begingroup$

Hormander's operator $L=X_0+\sum_{1\le j\le k} X_j^2$, where the $X_j$ are real smooth vector fields with the Lie algebra of $\{(X_j)\}_{0\le j\le k}$ generating the tangent space is hypoelliptic as well as its adjoint since the Lie algebra condition does not change by taking adjoints.

On the other hand, $\frac{\partial}{\partial t}+t\Delta_x$ is hypoelliptic whereas its adjoint $-\frac{\partial}{\partial t}+t\Delta_x$ is not hypoelliptic,

Bazin.

$\endgroup$
  • 1
    $\begingroup$ Could you explain why the operators $\frac{\partial}{\partial t} + t \Delta_x$ and $-\frac{\partial}{\partial t} + t \Delta_x$ are hypoelliptic and not, respectively? $\endgroup$ – Bob Yuncken Jul 22 '15 at 14:13

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.