Semistable Fundamental Group at Infinity

• Semistable fundamental group at infinity means any two proper rays in a 1-ended space are properly homotopic, ensuring controlled stabilization of the fundamental group.
• It is characterized through an inverse system of groups with eventually surjective bonding maps, providing a well-defined pro-fundamental group useful for studying large-scale topology and cohomology.
• The concept connects with simple connectivity at infinity, CAT(0) geometry, and combination theorems, offering insights into the structure of hyperbolic, cubulated, and relatively hyperbolic groups.

Semistable fundamental group at infinity is an asymptotic homotopy property of a 1-ended space or group that records whether the fundamental group seen near infinity stabilizes in a controlled way. For a connected, locally finite CW-complex $X$, the standard formulation requires that any two proper rays converging to the same end be properly homotopic; for a finitely presented group $G$, one applies this to the universal cover of a finite complex with fundamental group $G$. Equivalently, after fixing an exhaustion by compacta $K_1 \subset K_2 \subset \cdots$, semistability is the condition that the inverse system

$$\pi_1(X \setminus K_1) \leftarrow \pi_1(X \setminus K_2) \leftarrow \pi_1(X \setminus K_3) \leftarrow \cdots$$

is pro-epimorphic, so the fundamental pro-group at infinity is well defined up to pro-isomorphism (Mihalik, 22 Jul 2025).

1. Definition and equivalent formulations

The basic objects are ends, proper rays, and complements of compacta. In a connected, locally finite CW-complex, a proper ray is a proper map $r:[0,\infty)\to X$, and two rays determine the same end when sufficiently far tails lie in the same unbounded component of $X-C$ for every compact $C$. For a 1-ended complex $X$, semistability at infinity means that any two proper rays are properly homotopic; more generally, in the non-1-ended setting, the condition is imposed endwise, for rays converging to the same end (Mihalik, 2014).

The inverse-system description is equivalent and is central in applications. Fix a proper base ray $r$ and a nested exhaustion $G$0 with union $G$1. The resulting inverse system of groups $G$2 is semistable precisely when it is pro-isomorphic to one with eventually surjective bonding maps. In the 1-ended case, this is equivalent to several geometric “pushing” statements: loops far out can be pushed arbitrarily farther out by homotopies avoiding a prescribed compact set, and rays sufficiently far out are properly homotopic rel basepoint outside a smaller compact set (Mihalik, 2022).

For finitely presented groups, the definition is transported to the universal cover of a finite $G$3. For finitely generated groups, the same idea can be encoded using a Cayley graph together with finitely many attached 2-cells corresponding to a finite set of relators; this extension was introduced to study asymptotic behavior outside the finitely presented category (Conner et al., 2012).

2. Relation to simple connectivity at infinity and cohomology

Simple connectivity at infinity is the stronger condition that for every compact $G$4 there exists a compact $G$5 such that every loop in $G$6 is null-homotopic in $G$7. In pro-$G$8 language, this means that the inverse system at infinity is eventually trivial. Consequently, simple connectivity at infinity implies semistability (Mihalik, 23 Jun 2025).

This distinction is reflected in the range of geometric phenomena. Semistability controls proper homotopy classes of rays and gives a well-defined pro-$G$9 at infinity, whereas simple connectivity at infinity annihilates the entire fundamental pro-group. The literature repeatedly treats semistability as the asymptotic threshold needed to define “the” fundamental group at infinity of a 1-ended group, while simple connectivity at infinity is a much stronger tameness property (Mihalik, 2022).

There is also a cohomological connection. For finitely presented semistable groups, $G$0 is free abelian, and if the group is simply connected at infinity then $G$1 (Mihalik, 2014). The manual literature further organizes this through pro-homology at infinity, where semistability of first homology at infinity is equivalent to the freeness of $G$2 modulo the standard reductions to the 1-ended case (Mihalik, 22 Jul 2025). This places semistability at the center of the long-standing program relating end behavior, proper homotopy, and group cohomology.

3. Structural criteria and combination theorems

A substantial part of the theory consists of criteria that force semistability under algebraic decompositions. One of the basic tools is the Mihalik–Tschantz combination theorem: finite graphs of groups with finitely presented semistable vertex groups and finitely generated edge groups are semistable at infinity. This principle recurs across amalgams, HNN extensions, graph products, and accessibility arguments (Mihalik, 2020).

Near ascending HNN extensions provide a sharp example. If $G$3 is an infinite finitely presented group, $G$4 has finite index, and $G$5 is a monomorphism, then the HNN extension $G$6 is 1-ended and semistable at infinity; if $G$7 is also 1-ended, then $G$8 is simply connected at infinity (Mihalik, 2022). A related bounded-depth theorem shows that finitely presented ascending HNN extensions of finitely generated groups are semistable whenever the associated ascending HNN presentation has bounded depth, focusing attention on unbounded-depth cases as possible sources of non-semistability (Mihalik, 2017).

Commensurated-subgroup methods form another major branch. If a finitely generated group contains an infinite finitely generated commensurated subgroup of infinite index, then the group is 1-ended and semistable at infinity; if the subgroup is also finitely presented and 1-ended, then the ambient group is simply connected at infinity (Conner et al., 2012). This was extended to finite series of commensurated subgroups, or “subcommensurated” chains, again yielding semistability and, under stronger hypotheses, simple connectivity at infinity (Mihalik, 2014).

A complementary viewpoint uses non-cocompact subgroup actions. In that framework, semistability of a simply connected 1-ended complex $G$9 is decomposed into a $K_1 \subset K_2 \subset \cdots$0-part and a “perpendicular to $K_1 \subset K_2 \subset \cdots$1” part for a properly acting subgroup $K_1 \subset K_2 \subset \cdots$2: $K_1 \subset K_2 \subset \cdots$3-semistability controls motion along $K_1 \subset K_2 \subset \cdots$4-directions, while $K_1 \subset K_2 \subset \cdots$5-co-semistability controls pushing loops to infinity inside $K_1 \subset K_2 \subset \cdots$6-unbounded components. If both conditions hold, then $K_1 \subset K_2 \subset \cdots$7 is semistable at infinity (Geoghegan et al., 2017).

4. Boundary methods, CAT(0) geometry, and cubulations

Boundary topology provides one of the most effective geometric approaches. For proper one-ended CAT(0) spaces, semistability at infinity is equivalent to proper homotopy of geodesic rays, and also to semistability of the boundary’s fundamental pro-group (Geoghegan et al., 2017). This reduces the semistability question for a one-ended CAT(0) group to a geodesic-ray homotopy problem in a CAT(0) model. The same work shows that if a counterexample exists, then its visual boundary must have a weak cut point, its Tits boundary must exhibit a separating $K_1 \subset K_2 \subset \cdots$8-ball, and the group must be rank $K_1 \subset K_2 \subset \cdots$9 (Geoghegan et al., 2017).

There are also local geometric criteria. For proper geodesically complete CAT(0) spaces with connected boundary, concrete infinitesimal conditions on spaces of directions and local cones imply that the boundary is weakly chained, hence pointed $$\pi_1(X \setminus K_1) \leftarrow \pi_1(X \setminus K_2) \leftarrow \pi_1(X \setminus K_3) \leftarrow \cdots$$0-movable, and therefore that the space is semistable at infinity (Plaut, 2021). In Euclidean CAT(0) polyhedral complexes, these conditions become link criteria involving no free faces, connected complements of $$\pi_1(X \setminus K_1) \leftarrow \pi_1(X \setminus K_2) \leftarrow \pi_1(X \setminus K_3) \leftarrow \cdots$$1-balls, and, in one corollary, Moussong’s condition (Plaut, 2021).

Cubulated groups now form a fully settled subclass: every cubulated group is semistable at infinity (Shepherd, 2022). The proof modifies a cocompact cubulation until all halfspaces are one-ended and all quarterspaces are deep, and then uses these two properties to push loops arbitrarily far out while avoiding prescribed compacta (Shepherd, 2022). This resolves semistability for groups acting properly and cocompactly on CAT(0) cube complexes, including many groups that were previously accessible only through case-by-case arguments.

Relatively hyperbolic groups admit a parallel boundary-based and peripheral-based theory. If $$\pi_1(X \setminus K_1) \leftarrow \pi_1(X \setminus K_2) \leftarrow \pi_1(X \setminus K_3) \leftarrow \cdots$$2 is 1-ended, hyperbolic relative to 1-ended finitely generated peripherals, and the Bowditch boundary has no cut point, then $$\pi_1(X \setminus K_1) \leftarrow \pi_1(X \setminus K_2) \leftarrow \pi_1(X \setminus K_3) \leftarrow \cdots$$3 is semistable at infinity (Mihalik et al., 2017). More generally, if $$\pi_1(X \setminus K_1) \leftarrow \pi_1(X \setminus K_2) \leftarrow \pi_1(X \setminus K_3) \leftarrow \cdots$$4 is finitely presented and hyperbolic relative to finitely generated peripheral subgroups that are themselves semistable at infinity, then $$\pi_1(X \setminus K_1) \leftarrow \pi_1(X \setminus K_2) \leftarrow \pi_1(X \setminus K_3) \leftarrow \cdots$$5 is semistable at infinity even when the Bowditch boundary has cut points (Haulmark et al., 2021).

5. Established families and illustrative examples

Several broad classes are known to be semistable. Hyperbolic groups, many CAT(0) groups, all cubulated groups, Artin groups, Coxeter groups, one-relator groups, many solvable groups, and large families of relatively hyperbolic groups all appear in the current positive catalog (Hruska et al., 2019). The modern manual further organizes these results together with invariance principles, reduction theorems, and stronger connectivity-at-infinity statements for particular families such as $$\pi_1(X \setminus K_1) \leftarrow \pi_1(X \setminus K_2) \leftarrow \pi_1(X \setminus K_3) \leftarrow \cdots$$6, mapping class groups, Thompson’s group $$\pi_1(X \setminus K_1) \leftarrow \pi_1(X \setminus K_2) \leftarrow \pi_1(X \setminus K_3) \leftarrow \cdots$$7, and certain arithmetic groups (Mihalik, 22 Jul 2025).

Graph products supply a sharp local criterion. For a graph product on a finite connected graph with finitely presented vertex groups, non-semistability occurs exactly when some vertex group is not semistable at infinity and the subgroup generated by the adjacent vertex groups is finite; equivalently, the link is complete and all adjacent vertex groups are finite (Mihalik, 2020). In particular, right-angled Artin groups and right-angled Coxeter groups are semistable at infinity (Mihalik, 2020).

The Bieri–Stallings groups $$\pi_1(X \setminus K_1) \leftarrow \pi_1(X \setminus K_2) \leftarrow \pi_1(X \setminus K_3) \leftarrow \cdots$$8 provide a useful family for comparing finiteness properties and connectivity at infinity. They are defined as kernels

$$\pi_1(X \setminus K_1) \leftarrow \pi_1(X \setminus K_2) \leftarrow \pi_1(X \setminus K_3) \leftarrow \cdots$$9

where every standard generator maps to $r:[0,\infty)\to X$0. Bieri proved that $r:[0,\infty)\to X$1 is of type $r:[0,\infty)\to X$2 but not $r:[0,\infty)\to X$3, and the exact sequences

$r:[0,\infty)\to X$4

yield that $r:[0,\infty)\to X$5 is $r:[0,\infty)\to X$6-connected at infinity for $r:[0,\infty)\to X$7 (Mihalik, 23 Jun 2025). The conjectural optimum is $r:[0,\infty)\to X$8-connectivity at infinity. The verified low-dimensional cases are $r:[0,\infty)\to X$9, which is 1-ended, and $X-C$0 (Stallings’ group), which is simply connected at infinity; therefore $X-C$1 has semistable fundamental group at infinity (Mihalik, 23 Jun 2025).

6. Counterexamples, limitations, and open problems

The general finitely presented case remains open. The standard formulation—whether every finitely presented group has semistable fundamental group at infinity—has been emphasized for over forty years and is still unresolved (Hruska et al., 2019). The same status persists in major geometric subclasses: for example, it is still unknown whether every one-ended CAT(0) group is semistable at infinity (Geoghegan et al., 2017).

A major recent development is that the lamplighter group is not semistable at infinity (Mihalik, 20 May 2025). This gives a confirmed finitely generated non-semistable example and validates a long-standing suspicion from the finitely generated theory. The proof uses van Kampen band arguments in truncated Cayley 2-complexes to contradict the loop-pushing property that semistability would require (Mihalik, 20 May 2025).

At the same time, this counterexample does not immediately transfer to the finitely presented setting. The ascending HNN extension of the lamplighter group known as the Extended Lamplighter group is finitely presented, yet it is semistable at infinity and in fact simply connected at infinity (Mihalik, 20 May 2025). This sharply illustrates that non-semistability of a finitely generated base group does not force non-semistability of a finitely presented ascending HNN extension.

Current search strategies therefore remain indirect. Bounded-depth ascending HNN extensions are already known to be semistable (Mihalik, 2017), so unbounded-depth ascending HNN extensions continue to be treated as a natural frontier. More broadly, the modern picture suggests that semistability sits at the intersection of boundary topology, large-scale proper homotopy, decomposition theory, and subgroup geometry. A plausible implication is that progress on the finitely presented open problem will continue to come from interactions among these methods rather than from a single universal criterion (Mihalik, 22 Jul 2025).