4
$\begingroup$

I am wondering if there is a simple example of a manifold such that, given a value for the scalar curvature $R$, I can find a manifold such that the Ricci tensor has all zero components except for one component which takes the value $R$.

I feel like this can be achieved using a warped product of two metrics to separate out one coordinate and then just solve the differential equations so that the first coefficient vanishes, but obviously the coefficient of the non-zero component needs to be $R$.

$\endgroup$
2
  • $\begingroup$ Aren't components dependent on the choice of coordinates? Or are you asking for an example of a manifold where it is possible to choose coordinates in such a way? $\endgroup$
    – user74900
    Commented Mar 21, 2019 at 9:07
  • $\begingroup$ Yes, if it is possible to choose coordinates in this way. $\endgroup$ Commented Mar 21, 2019 at 9:15

1 Answer 1

4
$\begingroup$

Your intuition is correct that warped product metrics will do the trick.

Maybe try to redo the computations yourself, but if I'm not mistaken the Ricci curvature of $g=dt^2+f(t)^2dx^2$ (where $x\in\mathbb{R}^n$ and $dx^2$ denotes the euclidean metric on $\mathbb{R}^n$) is given by : $$\text{Ric}_g=-(n-1)\frac{f''(t)}{f(t)}dt^2-\left(f''(t)f(t)+(n-2)f'(t)^2\right)dx^2.$$ So any positive non constant solution to $f''f+(n-2)(f')^2=0$ will give you a solution, these are given by : $f(t)=C_2(C_1+(n-1)t)^{\tfrac{1}{n-1}}$. However note that the scalar curvature will not be constant in that example and that the singularity which happens when $f$ vanishes is genuine (the scalar curvature will blow up).

Maybe we can get something better (smooth for instance) using $\mathbb{R}\times \mathbb{S}^n$ or $\mathbb{R}\times \mathbb{H}^n$, this way the ODE becomes $f''f+(n-2)(f')^2=\pm 1$. The trick which solved the ODE in the first case doesn't work here though and a more careful analysis is required.

I don't know if one can find less symmetric examples, maybe studying the vector field given by the eigenvector of the Ricci tensor associated to the nonzero eigenvalue can give something.

$\endgroup$
2
  • $\begingroup$ Is it not the case that two ODEs have to be solve simultaneously though, as you want $dx^2$ coefficient to be $0$ but you want the coefficient of $dt^2$ to be equal to the scalar curvature $R$? $\endgroup$ Commented Mar 21, 2019 at 13:45
  • $\begingroup$ Here $(\partial_t,\tfrac{1}{f}\partial_{x^1},\dots,\tfrac{1}{f}\partial_{x^n})$ is an orthonormal basis of $TM$ hence the scalar curvature is $-(n-1)\tfrac{f''}{f}-n\tfrac{f''f+(n-2)(f')^2}{f^2}$. Your requirement is more or less equivalent to asking for the Ricci tensor to be of rank one as a bilinear form (which is an invariant statement). $\endgroup$ Commented Mar 22, 2019 at 8:41

You must log in to answer this question.

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