5
$\begingroup$

Let $M$ be a smooth manifold endowed with $\nabla$ a flat torsion-free connection. Let $\operatorname{Diff}(M)$ be the group of smooth diffeomorphisms of $M.$ Obviously if $ \phi \in \operatorname{Diff}(M)$, then $\phi^* \nabla$ is torsion-free and flat, but equivalent (through $\phi$) to $\nabla.$

Is it known whether the set of nonequivalent affinely flat connections is infinite (if it is not empty)?. In the case of the $2$-dimensional torus it is known that the set of flat, torsion free connections quotient out by the group of diffeomorphisms is a $4$-dimensional manifold.

$\endgroup$
  • 2
    $\begingroup$ Ali, "torsion free" is also a typo and should be "torsion-free" (and also, "$k$-dimensional" with a dash), and "quotiented out" $\endgroup$ – YCor Mar 1 '16 at 11:27
4
$\begingroup$

The set of affinely flat structures has been studied by many authors. Among them, we can quote William Goldman who has studied the deformations of $(X,G)$-structures. Let $X$ be a manifold, $G$ a Lie group which acts on $X$, we suppose that if two elements of $G$ coincide on an open subset of $X$, they coincide on $X$. A manifold $M$ is an $(X,G)$-manifold if it is endowed with an atlas $(U_i,\phi_i)$, where $\phi_i\colon U_i\subset M\rightarrow X$ is a diffeomorphism between $U_i$ and its image. We suppose that $\phi\circ {\phi_j}^{-1}$ is the restriction of an element of $G$. A flat torsionless connection defined on the $n$-dimensional manifold $M$ defines on it a $(\mathbb{R}^n,\operatorname{Aff}(\mathbb{R}^n))$-structure.

The $(X,G)$-structure of $M$ can be lifted to its universal cover $\tilde M$, and there exists a developing map: $D\colon\tilde M\rightarrow X$ such that for every element $\gamma\in\pi_1(M), D\circ \gamma =h(\gamma)\circ D$. We thus have a representation $h\colon\pi_1(M)\rightarrow G$ called the holonomy representation of the $(X,G)$-structure on $M$. It is a result of Thurston that if $M$ is closed the set of holonomy of $(X,G)$-structures defined on $M$ is an open subset of $\operatorname{Rep}(\pi_1(M),G)$, the set of representations $h:\pi_1(M)\rightarrow G$.

Goldman, William M. "Varieties of representations." Geometry of Group Representations: Proceedings of the AMS-IMS-SIAM Joint Summer Research Conference Held July 5-11, 1987 with Support from the National Science Foundation. Vol. 74. American Mathematical Soc., 1988.

Available here:

http://www-users.math.umd.edu/~wmg/geost.pdf

For the dimension 2, a theorem of Benzecri shows that an oriented surface is endowed with an affine structure if and only if its genus is one.

$\endgroup$
  • 2
    $\begingroup$ It could also be helpful to note that the map assigning to an $(X,G)$-structure its holonomy is a local homeomorphism on moduli, so the $(X,G)$-structure moduli space is locally homeomorphic to an open subset of the representation variety of the fundamental group. $\endgroup$ – Ben McKay Mar 1 '16 at 10:33

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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