Reading about Sasakian manifolds one come across two slogans:

A) "A Sasakian manifold is an odd-dimensional analogue of a Kahler manifold."

B) "A Sasakian manifold sits between two Kahler manifolds - one above and one below."

I would like to understand the second slogan for the motivating example of the three sphere $S^3$. What are the two Kahler manifolds that it sits between? I would guess that below is the projective line $\mathbb{CP}^1$. But I cannot guess what lies above.


I will answer your question for $S^{2n+1}$, since there is no difference between the case $n=1$ and the case of general $n$.

Let $(M^{2n+1},g,\theta)$ be a Sasakian manifold. One definition of a Sasakian manifold is that its metric cone is Kähler; this is the "one above". Here the metric cone is the manifold $M\times(0,\infty)$ with metric $dt^2+t^2g$. Thus in the case of $S^{2n+1}$, the metric cone is $\mathbb{C}^{n+1}\setminus\{0\}$ with the flat metric (written in spherical coordinates).

Taking the quotient by the $S^1$-action on $(M^{2n+1},\theta)$ determined by the Reeb vector field gives the "Kähler manifold below". For $S^{2n+1}$, the $S^1$-action is scalar multiplication by $e^{i\phi}$ (regarding $S^{2n+1}$ as the unit sphere in $\mathbb{C}^{n+1}$), so the quotient is $\mathbb{C}P^n$.

  • $\begingroup$ Will the metric cone be non-compact in general? $\endgroup$ – Fofi Konstantopoulou Apr 29 '20 at 20:32
  • $\begingroup$ Does this not work for n =0? S1 x R does not seem to be homeomorphic to C. $\endgroup$ – Asvin Apr 29 '20 at 22:24
  • $\begingroup$ @Asvin I think it should be $[0,\infty)$, with $M \times 0$ quotiented out. This part goes to the origin. $\endgroup$ – Ryan Apr 29 '20 at 23:26
  • $\begingroup$ I see. That makes more sense, thanks! $\endgroup$ – Asvin Apr 29 '20 at 23:39
  • 1
    $\begingroup$ The usual definition does not include $0$ in the second factor, the reason being that $(M\times[0,\infty)/\sim,dt^2+t^2g)$ is not smooth at the origin unless $M$ is a round sphere (here $\sim$ identifies $M\times\{0\}$ to a point). I have edited the original answer to reflect that the origin is missing. Also, the metric cone is always noncompact, because of the second factor. $\endgroup$ – Jeffrey Case Apr 29 '20 at 23:45

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.