8
$\begingroup$

I expect that the holonomy group of an Enriques surface $S$ is $SU(2)\times C_2$. I think this can be proven by the fact that its double cover, which is a K3 surface, has the full $SU(2)$ holonomy, but I failed proving it. The holonomy group should be either $SU(2)$ or its $\pi_1(S)=C_2$-extension $SU(2)\times C_2$. Could someone help me prove or disprove this?

$\endgroup$
2
  • $\begingroup$ $U(2)$ is not a $C_2$ extension of $SU(2)$, though; unless you and I mean something different by the word extension. $SU(2)$ is the kernel of the determinant map $U(2) \to U(1)$. $\endgroup$ Dec 15, 2013 at 10:22
  • $\begingroup$ You are right. $U(2)$ should have been $SU(2)\times C_2$. Thanks for pointing out the mistake. $\endgroup$
    – David K.
    Dec 15, 2013 at 10:25

1 Answer 1

14
$\begingroup$

Assume that you have endowed an Enriques surface $S$ with a Ricci-flat Kähler metric $g$. The holonomy $H$ of $g$ cannot be contained in $\mathrm{SU}(2)$ because the canonical bundle of $S$ is not trivial (though its square is trivial). Meanwhile, the identity component of $H$ has to be equal to $\mathrm{SU}(2)$ because this is the holonomy of the (simply-connected, non-product) K3 surface that is the double cover of $S$ endowed with one of its Ricci-flat Kähler metrics. The fundamental group of $S$ is $\mathbb{Z}_2$, so $H$ must be a $\mathbb{Z}_2$-extension of $\mathrm{SU}(2)$ that lies inside $\mathrm{U}(2)$. There is only one of these, namely the group consisting of those matrices in $\mathrm{U}(2)$ with determinant $\pm1$, so this must be $H$.

N.B.: $H$ is not isomorphic to $\mathrm{SU}(2)\times\mathbb{Z}_2$. In the latter, the set of elements that satisfy $a^2=1$ has $4$ members: $(I_2,1),(-I_2,1),(I_2,-1),(-I_2,-1)$, while, in the former, the set of matrices satisfying $A^2=I_2$ consists of two points, $\pm I_2$, together with the $2$-sphere consisting of the elements of the form $iJ$ where $J$ lies in $\mathrm{SU}(2)$ and satisfies $J^2 = - I_2$.

$\endgroup$
3
  • $\begingroup$ @Dr. Bryant Thanks you very much for the answer. I do not understand how you conclude that the identity component of the holonomy is $SU(2)$. As to $H$, you are totally right. $\endgroup$
    – David K.
    Dec 15, 2013 at 11:58
  • $\begingroup$ @DavidK.: Oh. I am relying on the Berger classification of holonomy for this: We know that the identity component of the holonomy must be a subgroup of $\mathrm{SU}(2)$ and, by Berger, we also know that the only connected proper subgroup of $\mathrm{SU}(2)$ that can be the holonomy of a $4$-manifold is the identity subgroup, which can only happen if the metric $g$ is flat, which is clearly not possible. $\endgroup$ Dec 15, 2013 at 12:08
  • $\begingroup$ @Dr. Bryant Thank you very much for the clarification. I am not familiar with the Berger classification. I will take a close look at it. I appreciate your kind reply. $\endgroup$
    – David K.
    Dec 15, 2013 at 12:14

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.