Abstract
Given a finite set of $r$ points in a closed surface of genus $g$, we consider the torsion elements in the mapping class group of the surface leaving the finite set invariant. We show that the torsion elements generate the mapping class group if and only if $(g, r) \neq (2, 5k+4)$ for some integer $k$.
Introduction
In geometric topology, the Mapping Class Group (MCG) of a surface $\Sigma_{g,r}$ describes the group of its topological symmetries—the ways one can deform the surface back onto itself. While it has long been established that these groups are generated by infinite-order "Dehn twists," this paper investigates whether the group can be reconstructed using only torsion elements: periodic symmetries that return the surface to its original state after a finite number of repetitions (like a rotation).
Methodology
The author employs a mix of algebraic topology and group theory, focusing on the first homology group ($H_1$) of the mapping class group. The proof follows a "divide and conquer" approach:
- Short Exact Sequences: Breaking the MCG into its commutator subgroup and its abelianization.
- Commutator Generation: Proving that the commutator subgroup is always generated by torsion elements.
- Homology Analysis: Showing that the images of torsion elements generate the first homology group, particularly for low-genus surfaces ($g=0, 1, 2$).
- Geometric Constructions: Using explicit models to find specific periodic homeomorphisms and expressing them in terms of Dehn twists.
Key Results
The paper provides a complete classification: the mapping class group is generated by torsion elements except when $(g, r) = (2, 5k + 4)$.
- Sphere ($g=0$): Generated by rotations of order $r$ and $r-1$.
- Torus ($g=1$): Generated by elements of orders $n \in \{2, 3, 4\}$.
- High Genus ($g \geq 3$): Generated entirely by involutions (elements of order 2).
The Exceptional Genus-2 Case
The exception $(g, r) = (2, 5k + 4)$ is rooted in symmetry. In these cases, the torsion elements generate a subgroup of index 5. This is due to a unique $\mathbb{Z}_5$-action on a genus-2 surface that has only 3 fixed points, which is insufficient to "reach" enough of the surface's boundary components to generate the full group.
Future Work
The paper leaves open the question of the minimal number of periodic generators needed to create the group regardless of genus or boundary points. Additionally, while the existence of certain symmetries is proven, their explicit expression in terms of Dehn twists remains an area for further investigation.
Summary updated