Minimal Topological Generating Sets

• Minimal topological generating sets are the smallest collections of homeomorphisms or isotopy classes whose topological closure equals the entire mapping class group.
• They are studied using methods such as abelianization and combinatorial relations (e.g., lantern and chain), ensuring that no proper subset can replicate the full group generation.
• Results demonstrate that for finite- and infinite-type surfaces, minimal generating sets can range from 2 to 4 elements, depending on characteristics like genus and number of ends.

A minimal topological generating set is a smallest possible collection of homeomorphisms or isotopy classes whose closure in the mapping class group of a topological space—under the relevant topology—equals the entire group. The study of these generating sets, especially for mapping class groups of finite- and infinite-type surfaces, is a fundamental topic in low-dimensional topology, geometric group theory, and quantum topology. The determination of such sets, their minimal cardinality, and their algebraic structure yields insights into the symmetries and dynamics of surfaces, as well as into computational and representation-theoretic applications. This article surveys the definitions, classification theorems, and methods for identifying and proving minimality of generating sets across a range of topological contexts.

1. Definitions and General Framework

Let $X$ be a topological space, typically a surface (possibly with punctures, boundary, or of infinite type). The (topological) mapping class group $\mathrm{Mod}(X)$ is the group of isotopy classes of orientation-preserving self-homeomorphisms of $X$. A (topological) generating set for $\mathrm{Mod}(X)$ is a subset $S$ such that the subgroup $\langle S \rangle$ is dense in $\mathrm{Mod}(X)$ with respect to a specified topology—typically the compact-open topology for infinite-type surfaces, resulting in a non-discrete, Polish group structure.

A generating set is minimal if no proper subset also generates the group topologically. Constraints may be imposed, such as requiring all generators to be involutions (order-2 elements), torsions, or of a particular algebraic form.

2. Finite-Type Surfaces: Orientable and Nonorientable Cases

Orientable Surfaces

For closed orientable surfaces $\Sigma_g$ of genus $g\geq 3$, the mapping class group $\mathrm{Mod}(\Sigma_g)$ is minimally generated by two elements. Classical results date to Humphries, who showed that $2g+1$ Dehn twists about nonseparating curves form a minimal generating set if only such twists are allowed. Wajnryb and later Korkmaz reduced the cardinality to 2 by allowing arbitrary mapping classes: $$\mathrm{Mod}(\Sigma_g) = \langle T, S \rangle, \qquad T = t_{a_2}, \quad S = t_{b_g} t_{c_{g-1}} \cdots t_{b_1} t_{a_1},$$ where $t_{a_i}$, $t_{b_j}$, $t_{c_k}$ are Dehn twists about standard curves (Altunoz et al., 20 Nov 2025).

Generation by involutions is also tightly analyzed. For $g\geq 8$, exactly three involutions suffice, and this is sharp; no two-involution generation exists due to the presence of nonabelian free subgroups. For surfaces with punctures, analogous results hold, with new bounds for the number of involutions as a function of genus $g$ and number of punctures $p$.

Nonorientable Surfaces

The mapping class group $\mathrm{Mod}(N_g)$ of a closed nonorientable surface of genus $g$ presents additional complexities. Crosscaps and crosscap transpositions $u_i$ must supplement Dehn twists for generation. For $g \geq 19$, $\mathrm{Mod}(N_g)$ is minimally generated (in the usual algebraic sense) by two elements: a cyclic crosscap rotation $T$ of order $g$ and a composite $G_1 = u_{g-1}\, t_{\gamma_{10}}\, t_{c_2}^{-1}$, where $u_{g-1}$ is supported on a one-holed Klein bottle (Altunoz et al., 20 Nov 2025). The twist subgroup $\mathcal{T}_g$ (index two in $\mathrm{Mod}(N_g)$) admits similar two-generator theorems for $g\geq21$ (odd genus) and $g\geq50$ (even genus).

Minimality is certified via the non-cyclicity of the group abelianization.

3. Infinite-Type Surfaces (Big Mapping Class Groups)

For orientable, infinite-genus surfaces $S(n)$ with $n\in\mathbb{N}$ ends each accumulated by genus, the mapping class group $\mathrm{Mod}(S(n))$ is no longer countably generated algebraically. Instead, minimal topological generation refers to the countable subsets whose closure in the compact-open topology equals the full group, a Polish group (Altunöz et al., 19 Dec 2025, Altunöz et al., 6 Jan 2026).

Results by End Number

• $n \geq 8$ ends: Three elements suffice to topologically generate $\mathrm{Mod}(S(n))$. These can be chosen to implement an $n$-cycle permutation of ends, a transposition of two ends, and a composite of Dehn twists and a handle-shift.
• $3 \leq n \leq 7$: Four-element sets are required by current constructions, though it is conjectured that three suffice.
• $n=2$ (Jacob's Ladder): A generating set of three elements is explicit, involving rotations and products of Dehn twists and handle-shifts.
• $n=1$ (Loch Ness Monster): Two elements (a handle-shift and a Dehn-twist product) suffice.

For involution-only generators (elements of order two), four involutions suffice if $n\geq 16$; for $n=1,2$ precisely three are minimal (Altunöz et al., 6 Jan 2026). Any group generated by two involutions is virtually cyclic, which is incompatible with the highly non-abelian structure of these mapping class groups.

4. Algebraic Types of Generating Sets

• Dehn twist only: For orientable closed surfaces, the minimal generating set using only Dehn twists is of size $2g+1$ (Humphries).
• Torsion elements: Mapping class groups can often be generated by small sets of torsion elements. For orientable mapping class groups, two torsion elements suffice under suitable genus bounds, and results exist for three order-3 elements or four order-4 elements (Altunoz et al., 20 Nov 2025).
• Involutions: Minimal involution generating sets are of separate algebraic interest due to their geometric and group-theoretic properties. In both finite- and infinite-type settings, at least three are required, with explicit upper bounds depending on genus or the number of ends.

In nonorientable cases, pure-twist generating sets tend to require more elements ($\geq g+1$).

5. Minimal Generation in Related Groups: Braid and Tangle Categories

Parallel analyses exist for braid groups, pure braid groups, and mapping class groups of classical surfaces such as the disc, sphere, and projective plane. For example, $B_n(S^2)$ is generated by two braids $x_0$ and $x_1$ of finite order, with minimality established by abelianization and explicit relations (Gonçalves et al., 2012). For pure braid groups, the minimal generating set size matches the rank of the abelianization.

Further, minimal generating sets in diagrammatic categories, such as those arising from rotational Reidemeister moves for framed and unframed tangles, have been classified. These structural results mirror approaches in mapping class group settings by seeking combinatorially irreducible sets of moves generating all equivalence relations in the respective category; e.g., the minimal set of rotational Reidemeister moves for oriented, unframed link diagrams is of size eight, and for framed links, five (Becerra et al., 18 Jun 2025).

Group/Setting	Minimal Generator Count	Generator Types
$\mathrm{Mod}(\Sigma_g)$, $g\geq3$	2	arbitrary mapping classes, or 3 involutions (for $g\geq8$)
$\mathrm{Mod}(N_g)$, $g\geq19$	2	rotation+crosscap-twist composite
$\mathrm{Mod}(S(n))$, $n\geq8$ (topological)	3	$n$-cycle, transposition, product of Dehn twists/handle-shift
$\mathrm{Map}(S(n))$, $n\geq16$ (involutions)	4	involutions
$B_n(S^2)$, $n\geq3$	2	finite-order braids
Rotational Reidemeister (unframed)	8	moves (diagrams)

6. Methods, Proof Techniques, and Minimality Criteria

Minimality proofs employ group abelianization, the explicit understanding of group presentations, geometric moves (crosscap slides, handle-shifts), and combinatorial group theory. For mapping class groups, lantern and chain relations facilitate expressing standard generators as products of more global elements (rotations or composites). For infinite-type surfaces, topological density arguments—invoking the structure of Polish groups and closure under conjugation—are fundamental.

A generating set is certifiably minimal if abelianization forces a lower bound on generator number, and if any proper subset fails (constructively) to generate the group or dense subgroup in the desired topology.

7. Open Problems and Current Directions

Active directions include:

• Lowering the minimal involution count for infinite-type mapping class groups below current bounds for large $n$ (Altunöz et al., 6 Jan 2026).
• Extending minimality theorems to surfaces with mixed end types or non-standard behaviors.
• Reducing generating set size for punctured or boundary-rich surfaces, particularly in the context of minimal torsion or involution generators.
• Exploring computational and algebraic consequences for quantum invariants, Teichmüller dynamics, and homological actions, given that minimal generating sets often lead to highly efficient presentations with broad implications.

The existence and explicit form of minimal topological generating sets connect deep geometric, algebraic, and dynamical properties and remain a central focus across topology and related fields (Altunoz et al., 20 Nov 2025, Altunoz et al., 20 Nov 2025, Altunöz et al., 19 Dec 2025, Altunöz et al., 6 Jan 2026, Gonçalves et al., 2012, Becerra et al., 18 Jun 2025).