---
title: 'Infinity Class: Structures at Infinity'
url: https://www.emergentmind.com/topics/infinity-class
type: topic
---

# Infinity Class: Structures at Infinity

“Infinity class” does not denote a single canonical construction. In the works considered here, it names several distinct but structurally related ways of organizing mathematical objects by their behavior “at infinity”: coarse filling growth in mapping class groups, higher connectivity at infinity for groups, class-level infinity axioms in set theory, topological conjugacy classes of polynomial basin dynamics, function classes with controlled escape geometry in complex analysis, rational surfaces classified by semigroups at infinity, and operator classes built around the infinity Laplacian or curved \(A_\infty/L_\infty\)-structures [1109.6048], [2602.12191], [2101.01698], [1107.1091], [2105.05452], [2408.15931], [2511.00931], [1009.6203].

## 1. Quantitative infinity classes in mapping class groups

For an orientable surface \(S=S_{g,b}\), the mapping class group is written
\[
\mathrm{Mod}(S)=\mathrm{MCG}(S).
\]
In the work on filling loops at infinity, the relevant invariant is the Dehn function at infinity, equivalently the first divergence function \(\operatorname{Div}^1\), defined by requiring filling disks to stay outside a large ball in the Cayley \(2\)-complex. A subset is \(r\)-avoidant if it is disjoint from the ball \(B_r\), and the problem is to fill \(r\)-avoidant loops by disks remaining outside a slightly smaller ball [1109.6048].

For surfaces of genus \(g\ge 5\) and any number of punctures or boundary components, the central theorem states that there is a constant \(c>0\) such that any \(r\)-avoidant loop of length \(l\) has an \(\frac rc\)-avoidant filling of area at most \(c\,r^2l^2\). When \(l\asymp r\), this yields
\[
\operatorname{Div}^1(\mathrm{MCG}(S))\preceq r^4.
\]
The same paper also records the Euclidean lower bound
\[
r^2\preceq \operatorname{Div}^1(\mathrm{MCG}(S)),
\]
so for \(g\ge 5\),
\[
r^2\preceq \operatorname{Div}^1(\mathrm{MCG}(S))\preceq r^4.
\]
It further remarks that there is evidence that the true growth might be \(r^3\) [1109.6048].

The proof uses the Gervais presentation, in which generators are Dehn twists about explicit \(\alpha\)-, \(\beta\)-, and \(\gamma\)-curves. The key combinatorial fact is that for every relation in the presentation there exists a common commuting Dehn twist, drawn from an abelian subgroup \(H\) generated by \(g\) mutually disjoint \(\beta\)-curves. Together with the theorem of Farb–Lubotzky–Minsky that abelian subgroups of mapping class groups are undistorted, this makes it possible to push a quadratic van Kampen filling outward through rank-\(\ge 3\) flats, paying an additional quadratic area factor. The resulting heuristic is precisely the quartic estimate: ordinary quadratic filling, multiplied by quadratic Euclidean avoidance inside undistorted abelian flats [1109.6048].

In this usage, an “infinity class” is a coarse growth class for avoidant fillings. Mapping class groups of genus at least \(5\) and arbitrary right-angled Artin groups share the same quartic upper bound,
\[
\operatorname{Div}^1\preceq r^4,
\]
and are therefore placed in the same infinity class at the level of first divergence [1109.6048].

## 2. Connectivity-at-infinity classes of mapping class groups

A second geometric-group-theoretic usage classifies mapping class groups by connectivity at infinity. For a locally finite connected CW-complex \(X\), simple connectivity at infinity means that for every \(n>0\) there exists \(M(n)\) such that every loop in \(X-St^{M(n)}(*)\) is null-homotopic in \(X-St^n(*)\). For finitely presented groups, this is defined on the universal cover of a finite \(2\)-complex or equivalently on the Cayley \(2\)-complex [2602.12191].

The main theorem of the 2026 paper gives a general criterion: if a finitely presented group has a free abelian subgroup \(A\cong \mathbb Z^p\) of rank at least \(3\) with generators \(T\) satisfying two commutation conditions with respect to every relator, then the group is simply connected at infinity. Applying this to Gervais presentations yields that the mapping class groups of closed surfaces of genus \(g\ge 3\) are simply connected at infinity. Since these groups are duality groups of dimension \(4g-5\), the Proper Hurewicz Theorem implies that they are \((4g-7)\)-connected at infinity [2602.12191].

The same work combines this result with earlier classifications to obtain a complete case-by-case list of all mapping class groups \(\Gamma_{g,r}^s\) by their connectivity at infinity. With
\[
d(g,0,0)=4g-5,\qquad
d(g,r,s)=4g+2r+s-4 \text{ when } g>0 \text{ and } r+s>0,\qquad
d(0,r,s)=2r+s-3,
\]
the generic case is that \(\Gamma_{g,r}^s\) is \(d(g,r,s)-2\)-connected at infinity [2602.12191].

| Infinity class | Parameters | Infinity behavior |
|---|---|---|
| Finite | \( (0,0,s),\, s\le 3 \); \( (0,1,0) \) | finite |
| Virtually free | \( (0,0,4), (0,1,1), (0,1,2), (0,2,0), (1,0,0), (1,0,1) \) | infinite, virtually free, simply connected at infinity |
| 1-ended, not simply connected | \( (0,0,5), (0,1,3), (0,2,1), (1,1,0), (1,0,2) \) | semistable at infinity, not simply connected at infinity |
| Highly connected at infinity | all remaining \( (g,r,s) \) with \(2g+r+s>2\) | \(d(g,r,s)-2\)-connected at infinity |

This classification makes “infinity class” literal: every mapping class group belongs to one of four connectivity-at-infinity types, and the previously unsettled closed genus \(3\) and \(4\) cases are placed in the highly connected class [2602.12191].

## 3. Broad Infinity as a class-level infinity principle

In set theory, “Infinity Class” is realized by Broad Infinity, a class-level strengthening of the usual Axiom of Infinity. The basic data are a broad arity
\[
F:E\to \mathrm{Set}
\]
and the associated endofunctor
\[
\mathrm{Maybe}_F(X)=\{\text{Begin}\}\cup\{\mathrm{Make}(x,y)\mid x\in X,\ y\in X^{F(x)}\}.
\]
A class \(X\) is \(\mathrm{Maybe}_F\)-prefixed if \(\text{Begin}\in X\) and whenever \(x\in X\) and \(y\in X^{F(x)}\), one has \(\mathrm{Make}(x,y)\in X\). The least such class is \(\mathrm{SimpleBroad}(F)\), and the Simple Broad Infinity scheme asserts that \(\mathrm{SimpleBroad}(F)\) is a set for every broad arity \(F\) [2101.01698].

The slogan given in the paper is: “Every time we construct a new element, we gain a new arity.” In the broad setting, the next arity depends on the previously constructed node \(x\) through the set \(F(x)\). The resulting objects are three-dimensional well-founded trees built from \(\text{Begin}\) and \(\mathrm{Make}\), with one “vertical” parent component and a family of “horizontal” children indexed by \(F(x)\) [2101.01698].

This principle is much stronger than ordinary Infinity. It quantifies over all class functions \(F:E\to\mathrm{Set}\), not just a fixed successor-like operation. Under AC or WISC, Broad Infinity is equivalent to Broad Set Generation, to Broad Supgeneration, and ultimately to Mahlo’s principle, namely that the class of regular limit ordinals is stationary. Under AC or Global WISC and Powerset, the equivalence chain is
\[
\text{Simple Broad Infinity}\Longleftrightarrow
\text{Full Broad Infinity}\Longleftrightarrow
\text{Broad Set Generation}\Longleftrightarrow
\text{Broad Supgeneration}\Longleftrightarrow
\text{Mahlo's principle}.
\]
Thus Broad Infinity packages Mahlo strength as a concrete closure principle for class-controlled trees rather than as an ordinal reflection statement [2101.01698].

The same framework yields generation principles. Any broad rubric on a class generates a set of derivations; under AC or WISC it also generates a set of elements. In particular, Broad Set Generation implies the Axiom of Universes: for every set \(X\), there is a least Grothendieck universe containing \(X\). Without choice, Broad Infinity still implies Broad Derivation Set, which is sufficient to construct Tarski-style universes [2101.01698].

In this set-theoretic usage, “Infinity Class” is not an asymptotic invariant but a class-parametrized infinitary closure principle.

## 4. Basin dynamics and analytic escape classes

In complex dynamics, an infinity class can mean a topological conjugacy class of basin dynamics. For a polynomial \(f:\mathbb C\to\mathbb C\) of degree \(d\ge 2\), the escape-rate function is
\[
G_f(z)=\lim_{n\to\infty}\frac{1}{d^n}\log^+|f^n(z)|,
\]
and the basin of infinity is
\[
X(f)=\{z\in\mathbb C:f^n(z)\to\infty\}=\{z:G_f(z)>0\}.
\]
When all critical points escape to infinity, the basin \(X(f)\) is a rigid Riemann surface, and the restriction \(f:X(f)\to X(f)\) determines the conformal conjugacy class of the polynomial up to affine transformations. DeMarco–Pilgrim classify the dynamics on \(X(f)\) by combinatorial invariants built from trees, laminations, tableaux, and a new invariant called the pictograph [1107.1091].

The pictograph is a topological-conjugacy invariant. For a fixed pictograph \(D\), the number \(\mathrm{Top}(D)\) of topological conjugacy classes of basins with pictograph \(D\) is algorithmically computable from the discrete data of \(D\). In the cubic case, explicit formulas are given in terms of marked levels, relative moduli, and the Branner–Hubbard tableau or Yoccoz \(T\)-sequence. Thus, in this setting, an infinity class is literally a topological conjugacy class of the dynamical system \((f,X(f))\) [1107.1091].

A different analytic usage appears in the study of entire functions and flows. For a transcendental entire function \(f\), the holomorphic flow
\[
\dot z=f(z)
\]
always has infinitely many pairwise disjoint trajectories tending to infinity in finite increasing time. Nevertheless, for any small transversal segment through a point \(z_0\) with \(f(z_0)\neq 0\), the set of initial conditions whose trajectories escape to infinity in finite time has Lebesgue measure zero. The paper also constructs an example with uncountably many such trajectories [2105.05452].

For the antiholomorphic flow
\[
\dot z=\overline{f(z)},
\]
finite-time escape can fail completely for some transcendental entire functions, but it must occur when \(f\) belongs to the Eremenko–Lyubich class \(\mathcal B\). Here \(\mathcal B\) is the class of transcendental entire functions whose finite critical and asymptotic values form a bounded set. In that class, logarithmic tracts over infinity provide enough geometric control to produce pairwise disjoint trajectories escaping to infinity in finite time. The same tract geometry yields a path to infinity along which \(f\) and all its derivatives tend to infinity, giving a strong affirmative answer to Rubel’s question for \(f\in\mathcal B\) [2105.05452].

This analytic usage makes \(\mathcal B\) an infinity class in the sense of controlled behavior near infinity, while the basin-dynamical usage makes infinity classes into topological conjugacy strata.

## 5. Surfaces at infinity and \(\delta\)-semigroup classes

In algebraic geometry, a surface at infinity is a rational surface obtained from \(\mathbb P^2\) by a simple chain of blow-ups along the branch at infinity of a plane curve with only one place at infinity. The starting curve \(C\subset \mathbb P^2\) satisfies \(C\cap L=\{p\}\) for the line at infinity \(L\), and has only one rational branch at \(p\). Its semigroup at infinity
\[
S_{C,\infty}=\{-\nu_{C,p}(h):h\in \mathcal O_C(C\setminus\{p\})\}\subset \mathbb N
\]
is generated, under the stated characteristic hypothesis, by a \(\delta\)-sequence \(\Delta=(\delta_0,\dots,\delta_g)\) in the sense of Abhyankar–Moh [2408.15931].

A divisorial valuation \(\nu\) of \(\mathbb P^2\) is non-positive at infinity if
\[
\nu(f)\le 0\qquad \text{for all } f\in k[x,y].
\]
For the blow-up surface \(X\) attached to such a valuation, Galindo–Monserrat’s criterion gives an equivalent geometric formulation: the cone of curves \(\mathrm{NE}(X)\) is generated by the strict transform \(\tilde L\) of the line at infinity together with the exceptional divisors \(E_1,\dots,E_n\). Hence \(\mathrm{NE}(X)\) is a finite rational polyhedral cone generated by natural curve classes [2408.15931].

A surface at infinity is then such an \(X\) with \(n\ge m\), where \(m\) is the number of blow-ups in the minimal embedded resolution of the branch at infinity, and with \(\mathrm{NE}(X)\) polyhedral and minimally generated. Its associated valuation semigroup at infinity is generated by a \(\delta\)-sequence of type A,
\[
\Lambda=(\delta_0,\dots,\delta_g,\delta_{g+1}),\qquad
\delta_{g+1}=n_g\delta_g-n+m.
\]
The underlying \(\delta\)-semigroup of the surface is
\[
S_\Delta=\langle \delta_0,\dots,\delta_g\rangle.
\]
Fixing \(S_\Delta\), the paper studies the class \(\mathcal S\) of all surfaces at infinity with that same \(\delta\)-semigroup [2408.15931].

Within \(\mathcal S\), different \(\delta\)-sequences can generate the same semigroup. The paper develops operations that shorten \(\delta\)-sequences to primitive ones and refinements that lengthen them while preserving the same semigroup. It proves finiteness statements: for fixed \(\delta_0\), there are finitely many \(\delta\)-sequences whose first element is \(\delta_0\), and for a fixed \(\delta\)-semigroup \(S\), there are finitely many relevant \(\delta\)-sequences generating \(S\), up to the explicitly excluded trivial infinite process of multiplying the first generator. Consequently, the class \(\mathcal S\) is a combinatorially finite family, and the generators of \(\mathrm{NE}(X)\) vary in a controlled way inside that family [2408.15931].

Here the “class at infinity” is a semigroup-classified family of rational surfaces whose global birational geometry is governed by data encoded at infinity.

## 6. Infinity-class equations in nonlinear PDE

A further usage arises for second-order PDE of the form
\[
M(x,Du,D^2u)+g(u)\,N(x,Du,D^2u)+f(x,u)=0 \quad \text{in }\Omega\subset\mathbb R^n,
\]
where the gradient term is not ad hoc but canonically derived from the operator:
\[
N(x,p,X)=\langle \partial_X M(x,p,X)\,p,p\rangle.
\]
The structural assumptions are a gradient homogeneity condition
\[
M(x,\lambda p,X)=|\lambda|^\alpha M(x,p,X)
\]
and a rank-one perturbation identity
\[
M(x,p,\gamma X+\sigma(p\otimes p))
=
\gamma^{\beta+1}M(x,p,X)+\sigma\,\gamma^\beta N(x,p,X).
\]
Under these hypotheses, the change of variables
\[
G(s)=\int_0^s g(\tau)\,d\tau,\qquad
\Phi_g(t)=\int_0^t e^{G(s)}\,ds,\qquad
v=\Phi_g(u)
\]
transforms the equation into one without the quadratic gradient term,
\[
M(x,Dv,D^2v)+h(x,v)=0,
\]
both for \(C^2\) solutions and for viscosity solutions [2511.00931].

The central example is the infinity Laplacian. Taking
\[
M(x,p,X)=\langle Xp,p\rangle=\Delta_\infty u,\qquad
N(x,p,X)=|p|^4,
\]
one obtains the infinity-class equation
\[
\Delta_\infty u+g(u)|Du|^4+f(x,u)=0.
\]
The same change of variables reduces it to an inhomogeneous infinity-Laplace equation with no explicit \(|Du|^4\)-term [2511.00931].

This yields structural consequences. For the homogeneous equation
\[
\Delta_\infty u+g(u)|Du|^4=0
\]
on a connected domain, the paper proves an Aronsson-type result: either \(Du\equiv 0\) in \(\Omega\) or \(u\) is constant. For the Dirichlet problem
\[
\begin{cases}
\Delta_\infty u + g(u)|Du|^4 + f(x,u)=0 & \text{in }\Omega,\\
u=b & \text{on }\partial\Omega,
\end{cases}
\]
it derives existence, uniqueness under a monotonicity assumption on \(t\mapsto \bar f(t)e^{3G(t)}\) for constant boundary data, and a nonexistence criterion based on the in-ball radius of \(\Omega\) [2511.00931].

In this setting, “infinity class” means a structurally stable class of equations centered on the infinity Laplacian and its natural gradient term, unified with the Laplacian, the \(m\)-Laplacian, and the \(k\)-Hessian operators by the same change-of-variables mechanism.

## 7. Curved \(A_\infty/L_\infty\)-algebras and characteristic classes

A final usage comes from homotopy algebra. A curved \(A_\infty\)-algebra on a finite-dimensional \(\mathbb Z/2\)-graded vector space \(V\) is an odd continuous derivation
\[
m\in \operatorname{Der}(T\Pi V^*)
\]
with \(m^2=0\); a curved \(L_\infty\)-algebra is defined similarly with \(S\Pi V^*\). Curvature is the constant term
\[
m_0=c\in V^{\bar 0},
\]
and a nontrivially curved algebra is one with \(m_0\neq 0\). In the cyclic case, the higher operations are compatible with a nondegenerate graded symmetric bilinear form [1009.6203].

Lazarev–Schedler extend Kontsevich’s characteristic-class construction from uncurved cyclic \(A_\infty\)- and \(L_\infty\)-algebras to curved ones by enlarging the graph complexes to allow vertices of valence \(1\). The resulting curved graph homology is explicitly computable. In the ribbon-graph case, its homology is governed by connected star-shaped graphs in which every vertex has valence \(1\) except possibly one vertex of odd valence; in the ordinary graph case, it is governed by the unique connected graph with two vertices and one edge [1009.6203].

The same paper also classifies nontrivially curved cyclic structures up to gauge equivalence. In the noncyclic case, a nonzero curvature makes all higher operations gauge trivial. In the cyclic \(L_\infty\) case, the same conclusion holds. In the cyclic \(A_\infty\) case, every nontrivially curved structure is gauge equivalent to one with only even-ary operations, determined by scalar invariants \((m_{2i}(c,\dots,c),c)\), and is essentially reduced to a model of dimension at most two. This is why the abstract states that nontrivially curved cyclic \(A_\infty\)- and \(L_\infty\)-algebras are essentially reduced to algebras of dimension at most two with only even-ary operations [1009.6203].

This algebraic usage is not a classification “at spatial infinity.” Rather, it concerns infinity-algebras themselves and the characteristic classes they define. Even so, it preserves the recurrent theme seen in the other usages: a complicated infinite hierarchy is organized by a rigid classifying mechanism, and in the curved case the surviving homology is concentrated in a remarkably small set of canonical models [1009.6203].

Source: https://www.emergentmind.com/topics/infinity-class