---
title: Framed Mapping Class Groups
url: https://www.emergentmind.com/topics/framed-mapping-class-groups
type: topic
---

# Framed Mapping Class Groups

Framed mapping class groups are groups attached to oriented surfaces endowed with a tangential datum, typically a trivialization of the tangent bundle, a nowhere–vanishing vector field, or puncturewise rotation data. In one standard formulation, one fixes a framing on a surface with boundary or marked points and takes the subgroup of the mapping class group preserving its isotopy class. In other formulations, one enlarges the mapping class group by adjoining integer or circle-valued rotation parameters at punctures, producing central or normal extensions. Across these variants, the governing structures are winding-number functions, quadratic refinements and spin structures, Arf-type invariants, relative homology, and monodromy representations arising from abelian differentials, singularity theory, Teichmüller quantization, and configuration-space constructions [2002.02471] [1703.02258] [2002.02472] [1003.5365] [1001.5366].

## 1. Foundational definitions and variant formalisms

For a closed oriented surface $\Sigma_g$ with a nonempty finite set of marked points $Z$, a framing of $(\Sigma_g,Z)$ is a trivialization of the tangent bundle of $\Sigma_g \setminus Z$, equivalently a nowhere–vanishing vector field on $\Sigma_g \setminus Z$, well-defined up to isotopy through such vector fields. If $\phi$ is fixed, the framed mapping class group is its stabilizer
$$
\PMod(\Sigma_g, Z)[\phi] = \{\, f \in \PMod(\Sigma_g, Z) \mid f \cdot \phi \text{ is isotopic to } \phi \,\}.
$$
For an oriented surface $\Sigma_{g,n}$ with $n \ge 1$ boundary components, a framing is a trivialization of $T\Sigma_{g,n}$, equivalently a nowhere–vanishing vector field, and the relative framed mapping class group is
$$
\Mod(\Sigma_{g,n})[\phi] = \{\, f \in \Mod(\Sigma_{g,n}) \mid f \cdot \phi = \phi \text{ up to relative isotopy} \,\}.
$$
Here relative isotopies are homotopies through nonvanishing vector fields fixed on $\partial\Sigma_{g,n}$ [2002.02471].

The same expression also appears in extension-theoretic settings. For a finite-type punctured surface $(S,P)$, a framing at a puncture is a choice of tangent direction, and a framed mapping class records both an ordinary mapping class and an integer rotation amount at each puncture. In the labeled-puncture case there is a short exact sequence
$$
0 \to \mathbb{Z}^s \to FMod(S,P) \to Mod(S,P) \to 0,
$$
with central kernel $\mathbb{Z}^s$; if punctures may be permuted, the kernel remains isomorphic to $\mathbb{Z}^s$ but is only normal, not central [1003.5365].

A third formalism compresses puncturewise framing data to a single global phase. For a closed surface $\Sigma_g$ and the unordered configuration space $\mathcal{C}_n(\Sigma_g)$, the weakly framed configuration space is the unit $S^1$-bundle in the square determinant line bundle,
$$
\mathcal{C}_n^{\mathtt{f}}(\Sigma_g):=\mathrm{U}(\Delta^2(\mathcal{C}_n(\Sigma_g)))\xrightarrow{\ \pi\ } \mathcal{C}_n(\Sigma_g),
$$
and the associated $\mathtt{f}$-based mapping class group fits into
$$
0\longrightarrow \mathbb{Z}\longrightarrow \mathfrak{M}^{\mathtt{f}}(\Sigma_g,*^{\mathtt{f}})\longrightarrow \mathfrak{M}(\Sigma_g,*)\longrightarrow 1.
$$
This is a central $\mathbb{Z}$-extension of the punctured mapping class group [2206.11475].

The terminology also encompasses braid-theoretic models. For integers $n>0$ and $r>0$, the framed braid group $FB_n$ is realized as a mapping class group $M(S_n^{(r)})$ of a disk with $n$ inner boundary components carrying $r$ marked arcs each. Under this identification, the braid generators $\sigma_i$ correspond to half-twists exchanging adjacent boundary components, while the framing generators $\tau_i$ correspond to rotations of an inner boundary component through the marked arcs [1702.03918].

| Setting | Framing datum | Group obtained |
|---|---|---|
| $(\Sigma_g,Z)$ | Trivialization of $T(\Sigma_g\setminus Z)$ | $\PMod(\Sigma_g,Z)[\phi]$ |
| $\Sigma_{g,n}$ | Trivialization of $T\Sigma_{g,n}$ fixed on $\partial\Sigma_{g,n}$ up to relative isotopy | $\Mod(\Sigma_{g,n})[\phi]$ |
| $(S,P)$ punctured finite type | Tangent directions and integer rotations at punctures | $FMod(S,P)$ |
| $\mathcal{C}_n(\Sigma_g)$ | Global $S^1$-phase in $\Delta^2$ | $\mathfrak{M}^{\mathtt{f}}(\Sigma_g,*^{\mathtt{f}})$ |

These constructions are closely related but not identical. The stabilizer formulation emphasizes preservation of a fixed framing class, whereas the extension formulations encode framing change as additional group coordinates.

## 2. Winding numbers, spin structures, and orbit invariants

A framing determines numerical invariants on immersed curves and arcs. For a framing $\phi$ on a surface with boundary, one has a winding-number function on oriented simple closed curves, and, after choosing legal basepoints on each boundary component, a relative winding-number function on legal arcs,
$$
\phi: \mathcal{S}^+(\Sigma_{g,n}) \to \tfrac{1}{2}\mathbb{Z}.
$$
Its fundamental properties are reversibility, twist-linearity,
$$
\phi(T_c^k(a)) = \phi(a) + k \langle a, c \rangle \,\phi(c),
$$
and homological coherence,
$$
\sum_i \phi(c_i) = \chi(S)
$$
for a subsurface $S$ with oriented boundary curves $c_i$ [2002.02471].

In the compact-surface-with-boundary framework, Kawazumi formulates the same structure via rotation numbers. A framing $f$ determines
$$
rot_f(\ell)=\langle \xi(f),[Vec(\ell)]\rangle\in \mathbb{Z}
$$
for a smooth immersion $\ell:S^1\to \Sigma$, and for boundary components the Poincaré–Hopf identity gives
$$
\sum_{j=0}^{n} rot_f(\partial_j\Sigma)=\chi(\Sigma)=1-2g-n.
$$
With $\nu_j(f):=rot_f(\partial_j\Sigma)+1$, this becomes
$$
\sum_{j=0}^{n}\nu_j(f)=2-2g.
$$
The homotopy set of framings is an affine $H^1(\Sigma;\mathbb{Z})$-torsor, and the action of $\Mod(\Sigma)$ on framings is measured by these rotation numbers [1703.02258].

The mod $2$ reduction of winding or rotation data produces quadratic refinements. For an embedded loop $\ell$, the associated quadratic form satisfies
$$
\omega_f([\ell]) = rot_f(\ell)+1 \in \mathbb{Z}_2.
$$
When the boundary restriction is trivial, the corresponding Arf invariant is
$$
Arf(q)=\sum_{i=1}^{g} q([\alpha_i])q([\beta_i]) \in \mathbb{Z}_2.
$$
In the relative framing setting relevant to blown-up zeros of abelian differentials, one has the generalized formula
$$
Arf(\phi,\mathfrak{B}) = \sum_{i=1}^g (\phi(x_i)+1)(\phi(y_i)+1) + \sum_{j=2}^n \big(\phi(a_j)+\tfrac{1}{2}\big)\big(\phi(\Delta_j)+1\big) \pmod 2,
$$
independent of the distinguished geometric basis $\mathfrak{B}$ [1703.02258] [2002.02472].

These invariants classify mapping class group orbits in several regimes. For absolute framings on $\Sigma_{g,n+1}$ with $g\ge 2$, the boundary vector $\nu=(\nu_0,\dots,\nu_n)$ determines exactly one orbit if some $\nu_j$ is odd, and exactly two orbits if all $\nu_j$ are even, distinguished by the Arf invariant. For $g=1$, boundary data do not suffice; one must also use
$$
\tilde A(f)=\gcd(rot_f(\alpha),rot_f(\beta),\nu_j(f)\text{ for }0\le j\le n).
$$
In the relative case, the generalized Arf invariant classifies orbits for $g\ge 2$, while the pair $(\tilde A(f),(f))$ classifies genus-$1$ relative framings [1703.02258].

A recurring consequence is that framed mapping class groups are determined by preservation of winding data. In higher genus this often reduces, mod $2$, to preservation of a spin structure or quadratic refinement; in genus $1$ an additional gcd-type invariant survives.

## 3. Relative homology and crossed-homomorphism descriptions

For surfaces with marked points or boundary, the natural linear target of a framed mapping class group is relative homology. In the marked-point case,
$$
0 \to H_1(\Sigma_g; \mathbb{Z}) \to H_1(\Sigma_g, Z; \mathbb{Z}) \to \widetilde{H}_0(Z; \mathbb{Z}) \to 0,
$$
and the pure automorphism group of relative homology fits into
$$
1 \to \RelAut(H_1(\Sigma_g, Z; \mathbb{Z})) \to \PAut(H_1(\Sigma_g, Z; \mathbb{Z})) \to \Sp(2g, \mathbb{Z}) \to 1,
$$
where
$$
\RelAut(H_1(\Sigma_g, Z; \mathbb{Z})) \coloneq \Hom(\widetilde{H}_0(Z;\mathbb{Z}),\, H_1(\Sigma_g;\mathbb{Z})).
$$
The corresponding relative homological representation is
$$
\Psi^{rel}: \PMod(\Sigma_g, Z) \to \PAut(H_1(\Sigma_g, Z; \mathbb{Z})),
$$
and similarly for surfaces with boundary, using
$$
p_*: H_1(\Sigma_{g,n}, \partial\Sigma_{g,n}; \mathbb{Z}) \xrightarrow{\sim} H_1(\Sigma_g, Z; \mathbb{Z}).
$$
The new feature, compared with the classical symplectic representation, is the relative transvection part carried by $\RelAut$ [2002.02471].

The decisive structure is a crossed homomorphism measuring mod $2$ change of winding number. For a framing $\phi$ on $(\Sigma_g,Z)$,
$$
\Delta_\phi(f,c)=\phi(f(c))-\phi(c)\pmod 2
$$
satisfies
$$
\Delta_\phi(fg,c)=\Delta_\phi(f,g(c))+\Delta_\phi(g,c),
$$
and descends to
$$
\overline{\Delta_\phi}: \PMod(\Sigma_g, Z) \to H^1(\Sigma_g; \mathbb{Z}/2).
$$
Moreover, $\overline{\Delta_\phi}$ factors through the relative homological representation via a crossed homomorphism
$$
\Theta_\phi: \PAut(H_1(\Sigma_g, Z; \mathbb{Z})) \to H^1(\Sigma_g; \mathbb{Z}/2).
$$
The main theorem states that for $g\ge 2$,
$$
\Psi^{rel}(\PMod(\Sigma_g, Z)[\phi]) = \ker(\Theta_\phi),
$$
and for boundary framings,
$$
\Psi^{rel}(\Mod(\Sigma_{g,n})[\phi]) = \ker(\Theta_\phi \circ p_*).
$$
Thus the image of a framed mapping class group in relative homology is cut out by a single crossed homomorphism [2002.02471].

The parity of the zero-order vector $\kappa=(\kappa_1,\dots,\kappa_n)$ controls the form of $\Theta_\phi$. If all $\kappa_i$ are even, equivalently $r=\gcd(\kappa_1,\dots,\kappa_n)$ is even, then the mod $2$ winding number descends to a classical spin structure
$$
q(x)=\phi(x)+1 \pmod 2
$$
on absolute homology, and
$$
\ker(\Theta_\phi)=\PAut(H_1(\Sigma_g,Z;\mathbb{Z}))[q]
\cong \Sp(2g,\mathbb{Z})[q]\ltimes \RelAut(H_1(\Sigma_g,Z;\mathbb{Z})).
$$
If some $\kappa_i$ is odd, then $\Theta_\phi$ is nontrivial on the relative piece. Writing
$$
v_\kappa=\sum_{i=1}^n \kappa_i\,p_i \in H_0(Z;\mathbb{Z}),
$$
one has
$$
\Theta_\phi|_{\RelAut}=v_\kappa^*,
$$
and a short exact sequence
$$
1 \to \ker(v_\kappa^*) \to \ker(\Theta_\phi) \to \Sp(2g,\mathbb{Z}) \to 1.
$$
This dichotomy isolates the even case as a spin-stabilizer problem and the odd case as a purely relative constraint.

Generator computations make the same picture explicit. Squared Dehn twists satisfy $\Delta_\phi(T_a^2,c)=0$, strict bounding pair maps have vanishing $\Delta_\phi$, and point-push maps detect the $\kappa_i$ through
$$
\Delta_\phi(T_{a_i}T_{a'_i}^{-1}, c) = \kappa_i \,\langle [a_i], c \rangle \pmod 2.
$$
The crossed homomorphism therefore packages the exact obstruction to preserving the framing at the relative homology level.

## 4. Abelian differentials, vanishing cycles, and geometric monodromy

Strata of abelian differentials supply canonical framings. If $(X,\omega)$ lies in the stratum $\mathcal{H}(\kappa_1,\dots,\kappa_n)$, the horizontal vector field
$$
H_\omega = 1/\omega
$$
is nonvanishing on $X\setminus Zeros(\omega)$. After real oriented blow-up of the zeros, one obtains a compact surface $X^*$ with boundary components $\Delta_i$ and a relative framing $\phi$ of signature
$$
sig(\phi)=(\phi(\Delta_1),\dots,\phi(\Delta_n))=(-1-\kappa_1,\dots,-1-\kappa_n).
$$
Relative framings with all boundary winding numbers negative are of holomorphic type. In genus $g\ge 5$, for a non-hyperelliptic connected component $\mathcal{H}$, the topological monodromy image is exactly the framed stabilizer:
$$
\rho(\pi_1^{orb}(\mathcal{H})) = Mod_g^n[\bar\phi],
$$
and on the blown-up or pronged covers one has
$$
\rho(\pi_1^{orb}(\mathcal{O}\mathcal{T}_g^{pr}(\kappa))) = Mod_{g,n}[\phi], \qquad
\rho(\pi_1^{orb}(\mathcal{H}^{lab})) = Mod_{g,n}^*[\phi].
$$
The same work proves that these framed stabilizers are finitely generated by explicit admissible Dehn twists associated to E-arboreal spanning configurations and more general assemblages [2002.02472].

Admissibility is defined by winding number. A nonseparating simple closed curve $a$ is admissible if $\phi(a)=0$; then $T_a$ preserves the framing by twist-linearity. For $g\ge 5$, the admissible subgroup
$$
\mathcal{T}_\phi=\langle T_a \mid a \text{ admissible}\rangle
$$
coincides with $Mod_{g,n}[\phi]$. This gives a finite, curve-theoretic generating theory for framed mapping class groups in the holomorphic and stabilized settings [2002.02472].

Plane curve singularities provide a parallel but independent source of canonical framings. If $f:(\mathbb{C}^2,0)\to (\mathbb{C},0)$ has an isolated critical point and Milnor fiber $F$, the Hamiltonian vector field $X_f$, defined by
$$
df=\omega_0(X_f,\cdot), \qquad X_f = J\nabla(\operatorname{Re}f) = -\nabla(\operatorname{Im}f),
$$
is tangent to the level sets of $f$ and nonvanishing on $F$. It therefore determines a canonical relative framing $\phi$ on $F$. If $g(F)\ge 5$ and $f$ is not of type $A_n$ or $D_n$, then the geometric monodromy group is exactly the framed mapping class group,
$$
\Gamma(f)=MCG(F)[\phi],
$$
and a nonseparating simple closed curve $\gamma\subset F$ is a vanishing cycle if and only if
$$
\phi(\gamma)=0.
$$
For the hyperelliptic types $A_n$ and $D_n$, the framing does not determine vanishing cycles; the criterion is instead symmetry under the hyperelliptic involution or its capped-off analogue. In genus at least $7$, the same framework yields non-injectivity of the geometric monodromy representation for nonhyperelliptic singularities [2004.01208].

This monodromy picture ties directly back to relative homology. In the singularity setting, the relative homological monodromy group is described as $\ker(\Theta_\phi)$, so the same crossed-homomorphism formalism that controls framed stabilizers on punctured or bordered surfaces also controls relative periods and homological monodromy.

## 5. Central extensions, braid models, and representation theory

Quantum Teichmüller theory yields a different but closely related appearance of framed mapping class groups. For a punctured surface $\Sigma_{g,r}^s$, Funar and Kashaev construct a central extension
$$
1 \to A \to \widetilde{\Gamma}_{g,r}^s \to \Gamma_{g,r}^s \to 1
$$
whose cohomology class is
$$
[c]=12\mu + \sum_{p\in P} e_p \in H^2(\Gamma_{g,r}^s;A).
$$
Here $\mu$ is the Meyer class and the $e_p$ are the puncture Euler classes. Chain relations lift to $z^{12}$, puncture relations lift to $z$, and lantern relations can be normalized to lift trivially. Passing from $Mod(S,P)$ to the framed mapping class group $FMod(S,P)$, or equivalently blowing up punctures to boundary components and remembering the framing rotations, absorbs the Euler terms: the pullback of $[c]$ becomes $12\mu$. In this sense framed mapping class groups are the natural receptacle in which puncture-framing anomalies disappear while the Meyer anomaly remains [1003.5365].

Braid-theoretic models produce concrete homological representations. The framed braid group
$$
FB_n=\mathbb{Z}^n \rtimes B_n
$$
is isomorphic to the mapping class group $M(S_n^{(r)})$ of a punctured disk with $r$ marked arcs on each inner boundary. From the relative homology of a configuration-space covering
$$
H_{n,m}^{(r)} := H_m(\widetilde{C}_{n,m}^{(r)}, \widetilde{A}^{(r)}; \mathbb{Z}),
$$
one obtains an $R=\mathbb{Z}[q^{\pm1},t^{\pm1}]$-linear representation
$$
\rho_{n,m}^{(r)}: FB_n \to Aut_R(H_{n,m}^{(r)}).
$$
Standard multifork classes span a free submodule of rank
$$
\binom{rn+n+m-2}{m},
$$
and over $\mathbb{Q}(q,t)$ one has
$$
\dim_{\mathbb{Q}}(H_m(\widetilde{C}_{n,m}^{(r)},\widetilde{A}^{(r)})\otimes_R \mathbb{Q})=
\binom{rn+n+m-2}{m},
$$
with all other homology groups vanishing. A quotient recovers Lawrence’s representation of $B_n$; the cases $m=1$ and $m=2$ recover the reduced Burau and Lawrence–Krammer–Bigelow representations. For $m\ge 2$, the framed representation is faithful [1702.03918].

The same paper constructs a monodromy representation from the confluent KZ equation with irregular singularities. In the symmetric case this gives
$$
\theta_{\lambda,\kappa}^{(r)}: FB_n \to Aut_\mathbb{C}(S_\Gamma^{(r^n)}[n-2m]),
$$
where
$$
\dim_\mathbb{C} S_\Gamma^{(r^n)}[n-2m]=\binom{rn+n+m-2}{m}.
$$
The conjecture is that, on an open dense subset of parameters, $\theta_{\lambda,\kappa}^{(r)}$ is equivalent to $\rho_{n,m}^{(r)}$ after the specialization
$$
q=\exp(2\pi i \lambda/\kappa), \qquad t=-\exp(-2\pi i/\kappa).
$$
This extends the classical Lawrence–KZ correspondence to a framed, irregular setting [1702.03918].

Weakly framed configuration spaces furnish yet another representation-theoretic construction. The weakly framed braid group of a closed surface maps onto a discrete Heisenberg group
$$
\phi:\ \mathbb{B}^{\mathtt{f}}_n(\Sigma_g)\longrightarrow H_g,
$$
with
$$
H_g=\mathbb{Z}\nu\times H_1(\Sigma_g;\mathbb{Z}), \qquad
(k,x)\cdot (l,y)=(k+l+x\cdot y,\ x+y).
$$
The corresponding $\mathtt{f}$-based mapping class group acts through oriented automorphisms
$$
f_H:(k,x)\longmapsto \bigl(k+\delta_f(x),\, f_*(x)\bigr),
$$
where
$$
\delta:\mathfrak{M}^{\mathtt{f}}(\Sigma_g,*^{\mathtt{f}})\to H^1(\Sigma_g;\mathbb{Z}\nu)
$$
is a crossed homomorphism. This produces twisted, and for the linearized regular representation $L$ untwisted, actions on Heisenberg homology
$$
\rho:\ \mathfrak{M}^{\mathtt{f}}(\Sigma_g,*^{\mathtt{f}})\to
\mathrm{Aut}\Bigl(H_*\bigl(\mathcal{C}_n^{\mathtt{f}}(\Sigma_g);L\bigr)\Bigr).
$$
The decomposition
$$
Aut^+(H_g)\cong Sp(H_1(\Sigma_g;\mathbb{Z}))\ltimes H^1(\Sigma_g;\mathbb{Z}\nu)
$$
shows that the Heisenberg lift refines the ordinary symplectic action by a framing-dependent cohomological term [2206.11475].

## 6. Moduli spaces, homological stability, and stable group homology

Framed mapping class groups admit a moduli-space description within the general theory of tangential structures. For a surface $F$ with boundary and a fixed boundary framing $\delta$, the moduli space of framed surfaces is
$$
M^{fr}(F;\delta):=
(Bun_\partial(TF,\theta^*\gamma_2;\delta)\times EDiff_\partial(F))/Diff_\partial(F),
$$
where the framed tangential structure is the trivialization of $TF$. When $\partial F\neq \varnothing$, there is a fibration
$$
map_*(F/\partial F,SO(2)) \to M^{fr}(F;\delta)\to M^+(F;\delta^+),
$$
and an exact sequence
$$
0 \to \pi_1(M^{fr}(F;\delta),\xi)\to \Gamma^+(F)\to H^1(F,\partial F;\mathbb{Z})\to \pi_0(M^{fr}(F;\delta))\to *.
$$
Hence each path component of $M^{fr}(F;\delta)$ is a $K(\pi,1)$, and its fundamental group is a framed mapping class group [1001.5366].

The elementary stabilization maps are denoted $\alpha$, $\beta$, and $\gamma$, obtained by gluing a pair of pants along its two legs, gluing a pair of pants along its waist, and gluing a disc. In the framed case, the stability ranges are explicit. For $\alpha(g):M^{fr}(\Sigma_{g,b})\to M^{fr}(\Sigma_{g+1,b-1})$, one has homology epimorphism in degrees $6k\le 2g-2$ and isomorphism in degrees $6k\le 2g-8$. For $\beta(g):M^{fr}(\Sigma_{g,b})\to M^{fr}(\Sigma_{g,b+1})$, one has epimorphism in degrees $6k\le 2g-4$ and isomorphism in degrees $6k\le 2g-10$, with split homology monomorphism in all degrees if one created boundary condition is trivial. For $\gamma(g):M^{fr}(\Sigma_{g,b})\to M^{fr}(\Sigma_{g,b-1})$, one has isomorphism in degrees $6k\le 2g-4$; if $b\ge 2$ it is a split homology epimorphism in all degrees, and if $b=1$ it is a homology epimorphism in degrees $6k\le 2g+2$ [1001.5366].

The stable target is the framed Madsen–Tillmann spectrum. Since the tangent bundle is trivial in the framed case,
$$
MT\theta_{fr}\simeq S^{-2},
$$
and therefore
$$
\Omega^\infty MT\theta_{fr}\simeq \Omega^\infty S^{-2}\simeq \Omega^2Q(S^0).
$$
After group completion,
$$
M^{fr}(\Sigma_\infty)\to \Omega^\infty S^{-2}
$$
is a homology equivalence. Consequently, in the stable range,
$$
H_*(M^{fr}(\Sigma_{g,b};\delta);\mathbb{Z})\cong H_*(\Omega^2Q(S^0);\mathbb{Z}),
$$
and the same holds for the stable homology of framed mapping class groups [1001.5366].

Two structural consequences are particularly sharp. First, the stable rational homology vanishes:
$$
H_*(\Gamma^{fr}(\Sigma_{g,b};\xi);\mathbb{Q})=0
$$
in the stable range. Second, for $g\ge 7$,
$$
H_1(\Gamma^{fr}(\Sigma_{g,b};\xi);\mathbb{Z})\cong \mathbb{Z}/24.
$$
This identifies the stable abelianization of the framed mapping class group with the third stable stem $\pi_3^S$. In the framed mapping-torus picture, the resulting class is computed by the $e$-invariant of the associated framed $3$-manifold [1001.5366].

Taken together, these results place framed mapping class groups at the intersection of low-dimensional topology, spin and tangential-structure theory, relative homological representation theory, and geometric monodromy. In one direction they are stabilizers of concrete winding-number data; in another they are extensions encoding puncture or phase anomalies; and in the stable regime they are governed by the homotopy theory of $\Omega^2Q(S^0)$.

Source: https://www.emergentmind.com/topics/framed-mapping-class-groups