1
$\begingroup$

We have a nowhere vanishing vector field $X$ on a Reimannian manifold $(M,g)$, such that $\mathcal{L}_X g=2\alpha X^\flat \otimes X^\flat$ for a constant non zero $\alpha$. So, for every vector fields $Y,Z$ orthogonal to $X$, we have $\mathcal{L}_X g(Y,Z)=0$. Also, for every two arbitrary commutitive vector fields $Y,Z$ orthogonal to $X$, we can write $dX^\flat (Y,Z)=0$.

Can we deduce that $K(X,Y)=0$ for every vector field $Y$ orthogonal to $X$?

Personally, I think, we can. Because in this case we can deduce that $M$ locally has to be as a warped product manifold $M=I\times_{f(s)} F$ with $X=\mu \dfrac{\partial}{\partial s}$. Where $\mu$ is a smooth function on $I$, and $I$ is an interval. Now, the condition $\mathcal{L}_X g=2\alpha X^\flat \otimes X^\flat$ yields that, $f(s)$ has to be a constant.

Am I right?

$\endgroup$
3
  • 1
    $\begingroup$ I am flagging this question because the author does not assume any topological constraint on the manifold, despite arguing the already received answers are not satisfactory, though mentioning 1d manifolds, where the concept of sectional curvature does not exist. Also, he asks to $alpha$ be non-zero, what is not possible on the closed setting. I recommend the OP make further clarifications. $\endgroup$ Commented Jan 4, 2023 at 18:53
  • 1
    $\begingroup$ In general, you shouldn’t change a question after it has been answered. Instead, it’s better to make an edit to the end of the question specifying the setting you are considering. $\endgroup$
    – Gabe K
    Commented Jan 5, 2023 at 14:31
  • $\begingroup$ @Gabe K: Thank you, you right. $\endgroup$ Commented Jan 5, 2023 at 17:45

2 Answers 2

2
$\begingroup$

This is not necessarily true. As a counter example, take $M=\mathbb{S}^3$ and consider the Killing field induced by some rotation. Then the Lie derivative of the metric will vanish (i.e., $\alpha=0$), but the sectional curvature is constant and positive.

$\endgroup$
1
  • $\begingroup$ Thank you for your answer. The constant $\alpha$ is non zero in our setting. $\endgroup$ Commented Jan 4, 2023 at 15:45
1
$\begingroup$

On a Riemannian manifold without boundary it must hold that $\alpha=0$ --- i.e., $X$ is a killing field, like @Gabe K said.

In general, it is easy to establish that for a general vector field $X\in\mathfrak{X}(M)$,

$$ \operatorname{tr}_{g}\mathcal{L}_{X}g=2\operatorname{div}{X}.$$

On the other hand,

$$ \operatorname{tr}_{g}(2\alpha X^{\flat}\otimes X^{\flat})=2\alpha |X|^{2}_{g}.$$

So by comparing,

$$ \operatorname{div}{X}= \alpha |X|^{2}_{g}.$$

But by the divergence theorem,

$$\int_{M} \operatorname{div}{X}\, \operatorname{Vol}_{g}=0.$$

Which is to say,

$$ \alpha \int_{M} |X|^{2}_{g}\, \operatorname{Vol}_{g}=0.$$

Hence either $\alpha=0$ or $X=0$ identically.

$\endgroup$
1
  • $\begingroup$ Integral of divergence is zero in a compact manifold without boundary. Our manifold is not necessarily compact. Examples of such vector fields can be constructed in 1- dimensional case. $\endgroup$ Commented Jan 4, 2023 at 16:05

You must log in to answer this question.

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