3
$\begingroup$

Let $(M,g)$ be a $2n$-dimensional almost Hermitian manifold ($n\geq 2$) with a almost complex structure $\cal J$ (not necessary integrable). i.e., $${\cal J}^2=-I,\quad\qquad g({\cal J} X,{\cal J} Y)=g(X,Y).$$ Suppose that $\{X_i,{\cal J}X_i\}$ be any local orthonormal ${\cal J}$-frame and the following relations hold for $i\neq j$ $$K(X_i,X_j)-K(X_j,{\cal J}X_j)=\frac{1}{2n-2}\{\rho(X_i,X_i)-\rho({\cal J}X_j,{\cal J}X_j)\}\tag{1}$$ $$K(X_i,X_j)-K(X_i,{\cal J}X_j)=\frac{1}{2n-2}\{\rho(X_j,X_j)-\rho({\cal J}X_j,{\cal J}X_j)\}\tag{2}$$ where $\rho$ and $K$ are Ricci curvature tensor and sectional curvature respectively. Then

Can be deduce that $(M,\cal J,g)$ is of constant curvature?

This question comes from the study of conformally flat almost Hermitian manifolds.

Your suggestions will be appreciated.

$\endgroup$
4
  • 2
    $\begingroup$ The answer is 'no', at least when $2n=4$. In that case, your conditions only involve the Weyl curvature of the underlying metric $g$, so, in particular, if $g$ is conformally flat, then your conditions are satisfied, and there are many conformally flat metrics in dimension $4$ that do not have constant curvature. Then just choose any $g$-compatible almost complex struture $\mathcal{J}$, and you have a counterexample. Perhaps you meant to ask whether it can be deduced that $g$ is conformally flat? (This does not follow immediately since your conditions don't force the Weyl curvature to vanish.) $\endgroup$ Sep 6, 2017 at 14:17
  • $\begingroup$ Dear prof. Bryant, many thanks for your comment. May be the case dim $=4$ is a exception (such as Milnor exotic sphere! :) and I didn,t cheacked it. can be extend your conterexample to dim $=5$? That is surprising for me that the condition (2) is not useful for conformally flatness because the right hand side of (2) is $X_i$-free. $\endgroup$
    – C.F.G
    Sep 6, 2017 at 14:42
  • 1
    $\begingroup$ I don't understand your comment or what you mean by 'not useful for conformal flatness'. The case $\dim = 5$ cannot happen, since we must be on an even dimensional manifold. The fact is that, in dimension $4$, your conditions only constrain the Weyl curvature of the underlying metric $g$, not the Ricci curvature, but they don't force the Weyl curvature to be zero; it's just required to lie in a certain $2$-dimensional subbundle that is defined by the $\mathrm{U}(2)$-structure given by $(g,\mathcal{J})$ on $M$. If you add that $g$ is Einstein, then, yes, you'll get constant curvature. $\endgroup$ Sep 6, 2017 at 14:51
  • $\begingroup$ Can you write your counterexample as a Answer? Please accept my apology for this request and thank you for sharing your wisdom with me. $\endgroup$
    – C.F.G
    Sep 6, 2017 at 15:05

1 Answer 1

4
$\begingroup$

The answer to the question as asked is 'No', at least when $2n=4$. In that case, the conditions (1) and (2) only involve the Weyl curvature of the underlying metric $g$, so, in particular, when $g$ is conformally flat, then the conditions are satisfied, and there are many conformally flat metrics in dimension $4$ that do not have constant curvature.

To construct an example of $(g,\mathcal{J})$ that does not have constant curvature just choose any conformally flat metric $g$ in dimension $4$ that does not have constant sectional curvature and any $g$-compatible almost complex structure $\mathcal{J}$.

Since the Ricci curvature is not constrained, there are probably examples that are not conformally flat, but they do, at least, have to be self-dual, i.e., $W_-$ must vanish (which is implied by (1) and (2)). Meanwhile, the conditions (1) and (2) imply that $W_+$ must take values in a certain rank $2$ subbundle of the self-dual Weyl curvatures that is defined using $\mathcal{J}$: If $\omega$ is the the (self-dual) $2$-form defined so that $\omega(v,w) = g(\mathcal{J}_xv,w)$ for all $v,w\in T_xM$, then the self-dual $2$-forms can be split as $\Lambda^2_+(T^*M) = \mathbb{R}\omega\oplus E$ where $E$ is a bundle of rank $2$. Then (1) and (2) are equivalent to the requirement that the Weyl curvature be self-dual and take values in the rank $2$ subbundle $\omega\circ E\subset S^2\bigl(\Lambda^2_+(T^*M)\bigr)$.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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