5
$\begingroup$

It is well known that if the first Chern class is proportional to the kähler class given, then every cscK in that class has to be kähler Einstein. So there are two directions to generate examples as far as I can think of:

1). It is a result if we have a cscK in $[w]$ then for any kähler class sufficiently close to $[w]$ the existence of cscK is guaranteed. So maybe we can try perturb our class on manifolds where $C_1(M) < 0$.

2). May be we can consider manifolds where $C_1(M)$ is semiample.

I have no idea. Are there any standard examples which are relatively simple to calculate?

$\endgroup$
2
  • 2
    $\begingroup$ The easiest examples are products of two KE manifolds with different Einstein constants. For example $M=\mathbb{P}^1\times\mathbb{P}^1$ with the product of two Fubini-Study metrics with different volumes. This is clearly cscK and not KE. Note that $M$ here is Fano. $\endgroup$
    – YangMills
    Apr 27, 2020 at 15:35
  • $\begingroup$ Also, your "fact" number 1) above is wrong. It is true if $M$ doesn't have nontrivial global holomorphic vector fields, but false in general, see e.g. LeBrun-Simanca $\endgroup$
    – YangMills
    Apr 27, 2020 at 15:38

2 Answers 2

4
$\begingroup$

Here is a non-compact example.

Consider the half-space $ \mathbb{C} \times \mathbb{H}$ as a subset of $ \mathbb{C}^2$. We use the Kahler potential $$\Psi = \frac{x_1^2}{x_2} - \log(x_2).$$ Here, $z_1 =x_1+ \sqrt{-1} y_1$ and $z_2 =x_2+ \sqrt{-1} y_2$.

This metric has constant scalar curvature but the Ricci potential is $3 \log(x_2)$, so is not Kahler-Einstein.

This Kahler manifold is known as the Siegel-Jacobi space and has been studied in quite a bit of detail. To give two examples, Mathieu Molitor wrote a paper "Gaussian distributions, Jacobi group and Siegel-Jacobi space" describing the geometry in considerable detail. It was also studied by Jae-Hyun Yang in "Geometry and Arithmetic on the Siegel-Jacobi Space."

$\endgroup$
6
  • 2
    $\begingroup$ Are there any examples that are compact? $\endgroup$
    – qwe
    Apr 23, 2020 at 12:58
  • $\begingroup$ I don't know of any examples off the top of my head. Presumably they exist since cscK seems like a much weaker condition than Kahler-Einstein. However, any example cannot be Fano. $\endgroup$
    – Gabe K
    Apr 23, 2020 at 14:29
  • $\begingroup$ Why can't it be Fano? $\endgroup$
    – qwe
    Apr 23, 2020 at 15:26
  • $\begingroup$ For Fano manifolds, csck metrics are precisely Kahler-Einstein and their existence depends on an algebro-geometric condition known as k- poly stability. This is a famous result of Chen, Donaldson, and Sun. $\endgroup$
    – Gabe K
    Apr 23, 2020 at 18:59
  • 2
    $\begingroup$ This is wrong -- there are many Fano examples, see my comment above. What is true (and has nothing to do with Chen, Donaldson, Sun) is that on a Fano manifold a cscK metric which is cohomologous to $c_1(M)$ must be KE (proof extremely easy). But there are in general many other Kahler classes, e.g. on $\mathbb{P}^1\times\mathbb{P}^1$. $\endgroup$
    – YangMills
    Apr 27, 2020 at 15:36
1
$\begingroup$

$\newcommand{\Proj}{\mathbf{P}}$Examples of compact constant scalar curvature Kähler manifolds that are not Einstein are constructed by solving explicit ODEs in On existence of Kähler metrics with constant scalar curvature, Osaka J. Math., 31 (1994), 561–595. If $M$ is a compact almost-homogeneous space with real hypersurface orbits and two ends, and if the connected automorphism group is reductive, then the set of Kähler classes containing a Kähler metric of constant scalar curvature is a real-algebraic hypersurface that separates the Kähler cone.

The simpest non-product example is $\Proj^{3}$ blown up along two skew lines. We may view this Fano threefold as the completion of the line bundle $\mathcal{O}_{\Proj^{1}}(-1) \otimes \mathcal{O}_{\Proj^{1}}(1) \to \Proj^{1} \times \Proj^{1}$ (with the line bundles pulled back to $\Proj^{1} \times \Proj^{1}$ by projection to the respective factors). The Kähler cone has three parameters, which may be viewed loosely as the sizes $a_{1}$ and $a_{2}$ of the $\Proj^{1}$ factors and the size $2b$ of the $\Proj^{1}$ fibre. [See note below] If we scale to make $b = 1$, the resulting slice of the Kähler cone is the open quadrant $a_{1}$, $a_{2} > 1$.

Slice of the Kahler cone of P^3 blown up along two skew lines

The point $(a_{1}, a_{2}) = (2, 2)$ is proportional to the anticanonical class, which contains an Einstein-Kähler metric by work of Koiso and Sakane (Non-homogeneous Kähler-Einstein metrics on compact complex manifolds, in Springer Lecture Notes 1201, 1986, 165-179).

The set of classes containing a constant scalar curvature representative consists of the diagonal $a_{1} = a_{2}$ and a branch of the right hyperbola $(a_{1} - 1)(a_{2} - 1) = 1$.

Note: More literally, let $\omega$ denote the unit-area Fubini-Study form on $\Proj^{1}$. In the Kähler class $(a_{1}, a_{2}, b)$, the zero section has Kähler form cohomologous to $(a_{1} - b)\omega \times (a_{2} + b)\omega$ and the infinity section has Kähler form cohomologous to $(a_{1} + b)\omega \times (a_{2} - b)\omega$.

$\endgroup$

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.