7
$\begingroup$

Let $(M^n,g)$ be a closed smooth Riemannian $n$-manifold with positive scalar curvature (or positive Ricci curvature) and $(S^n, g_{st})$ be the standard round $n$-sphere.

Whether there exists a non-zero degree harmonic map $f$ from $M^n$ onto $S^n$, $f:M^n\to S^n$?

$\endgroup$

1 Answer 1

5
$\begingroup$

A simple example where the answer is 'no' is when $M=\mathbb{RP}^2$ (with, say, the standard metric of Gauss curvature $K\equiv1$, though, in dimension $2$, only the conformal structure on $M$ matters in the definition of harmonic map).

There is no non-constant harmonic map $f:\mathbb{RP}^2\to S^2$ (when $S^2$ given the standard metric with $K\equiv1$). In particular, there is not one that has nonzero degree.

The reason is that such a map would lift to $\tilde f:S^2\to S^2$ as a non-constant harmonic map, and it is well-known that such a map would have to be either holomorphic or anti-holomorphic when $S^2$ is regarded as $\mathbb{CP}^1$, i.e., the Riemann sphere. Since $\mathbb{RP}^2$ is $\mathbb{CP}^1=S^2$ divided by the anti-holomorphic involution $[z,w]\to [-\bar w,\bar z]$, it would follow that the holomorphic mapping $\tilde f$ would have to satisfy $\tilde f\bigl([z,w]\bigr) =\tilde f\bigl([-\bar w,\bar z]\bigr)$, which would be impossible unless $\tilde f$ were constant.

$\endgroup$
4
  • $\begingroup$ Thanks. Do you have higher-dimensional examples? Since I need the existence to attack the problem of scalar curvature. In particular, whether the map $Id: (S^n, g)\to (S^n, g_{st})$ is a harmonic map? Here $Id$ is the identity map, $n\geq 5$ and the scalar curvature of $g$ is positive. $\endgroup$ Commented May 12, 2021 at 13:54
  • $\begingroup$ @JialongDeng: Are you asking for the condition on a metric $g$ on $S^n$ that the identity map $\mathrm{Id}:(S^n,g)\to(S^n,g_{st})$ be harmonic and the scalar curvature be positive? Since this is an underdetermined system of second order PDE, there are likely to be so many examples that they can't be classified. $\endgroup$ Commented May 12, 2021 at 14:22
  • $\begingroup$ Yes. Can it be helpful if we add the condition: $|v|_g\geq |v|_{g_{st}}$ for any $v\in TS^n$? $\endgroup$ Commented May 12, 2021 at 15:20
  • $\begingroup$ @JialongDeng: Actually, that won't be all that helpful. If a metric $g$ satisfies the stated conditions, then multiplying it by a sufficiently large constant will yield a new metric $g'$ that satisfies this additional inequality as well. $\endgroup$ Commented May 12, 2021 at 15:37

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .