Skip to content

Serre's Property FA

A group property requiring every action on a simplicial tree to have a global fixed vertex, equivalently excluding the quotient, amalgam, and ascending-union obstructions identified by Serre.

Version
v2 · 2026-08-30 · History
Domain-specific #
2759
Origin domain
geometric group theory
Subdomain
group actions on trees
Aliases
Property FA, FA property, Fixed-point property FA

Core Idea

A group (G) has Serre's Property FA when every action of (G) by automorphisms on a simplicial tree has a global fixed vertex. In the standard Bass–Serre convention the action is without edge inversions: no element exchanges the two endpoints of an edge. If arbitrary simplicial actions are admitted, barycentric subdivision removes inversions, and a fixed midpoint in the original geometric realization becomes a fixed subdivision vertex. Stating the convention prevents the false conclusion that a finite group fails FA merely because an involution flips one edge.[1]

The quantifier “every” is load-bearing. A group does not have FA because its trivial action fixes every vertex, or because one specially chosen action fixes a vertex. It has FA only when no fixed-point-free tree action exists. Bass–Serre theory converts that geometric universal condition into algebraic restrictions on splittings. Serre's criterion says that (G) has FA exactly when all three obstructions are absent: (G) has no quotient isomorphic to (mathbb Z); (G) is not a nontrivial amalgamated free product; and (G) is not the union of a strictly increasing sequence of subgroups. For a countable group, the third condition is equivalent to finite generation.[2][1]

This is a domain-specific abstraction. It is a mature, independently testable group property that translates between action geometry and algebraic decomposition. Yet its exact content depends on groups, simplicial trees, Bass–Serre actions, quotients, and amalgams. The transferable idea of a fixed point is already represented by the Fixed Point prime; FA is not itself a new substrate-neutral prime.

Structural Signature

The defining signature is

group G + every inversion-free simplicial action G -> Aut(T) + one vertex fixed by all of G in each action -> Property FA.

Its components are:

  • a group (G) whose property is being tested;
  • an arbitrary simplicial tree (T), not one preferred Cayley graph or one finite tree;
  • an action \(G\curvearrowright T\) by graph automorphisms, ordinarily without edge inversions;
  • a global fixed vertex (v) satisfying (g v=v) for every \(g\in G\);
  • universal scope, requiring such a vertex for each eligible action;
  • the quotient obstruction, a surjection \(G\twoheadrightarrow\mathbb Z\), which would let (G) act by translations on a line;
  • the amalgam obstruction, a nontrivial expression \(G=A\mathbin{*}_{C}B\), whose Bass–Serre tree has no global fixed vertex;
  • the cofinality obstruction, representation of (G) as a strictly increasing countable union \(G_0<G_1<\cdots\), which supplies the third failure mode in Serre's criterion;
  • for countable (G), finite generation, the equivalent positive form of absence of the third obstruction.

An HNN extension is not an omitted fourth condition. A nontrivial HNN extension has a canonical epimorphism to (mathbb Z) sending the stable letter to (1) and the base group to (0), so it already fails the quotient condition. More generally, the Bass–Serre tree is the interface that turns an algebraic graph-of-groups splitting into a fixed-point-free action and turns a tree action into decomposition information.[1]

What It Is Not

  • Not existence of a fixed point for one action. The property quantifies over all simplicial-tree actions.
  • Not elementwise ellipticity alone. Knowing that each generator fixes some vertex does not by itself produce one vertex fixed by the whole group; the fixed sets must meet globally. Serre's criterion controls the universal conclusion.
  • Not the generic Fixed Point abstraction. FA is a property of a group across an entire class of actions, not a single state left unchanged by one transformation.
  • Not a fixed end or invariant line. A group may preserve an end or line while translating along it and hence have no fixed vertex.
  • Not Property FR. FR requires fixed points for actions on real trees, a broader class than simplicial trees; FR implies FA, while FA alone should not be silently upgraded.
  • Not Property FW or a cubical fixed-point property. Actions on wall spaces, median graphs, or CAT(0) cube complexes introduce different test classes.
  • Not Kazhdan's Property (T). Property (T) implies FA, by Watatani's theorem, but the converse fails.[3]
  • Not indecomposability in every algebraic sense. FA forbids the relevant Bass–Serre splittings; it does not say that the group is simple, directly indecomposable, or free of finite quotients.
  • Not finite generation by definition. Countable FA groups are finitely generated by the criterion, but the general theorem uses the ascending-union condition and must retain that qualification.

Scope of Application

Property FA belongs to combinatorial and geometric group theory, especially Bass–Serre theory, lattices, arithmetic groups, and fixed-point properties. It is used to obstruct actions on trees, rule out graph-of-groups decompositions, compare rigidity properties, and control homomorphisms into groups acting on trees.

The definition applies to finite or infinite groups and to finite or infinite simplicial trees. The eligible action need not be faithful. A quotient action matters precisely because any quotient of (G) can act while the kernel acts trivially. Consequently, one infinite cyclic quotient is enough to disprove FA.

The property also behaves well under several group constructions. Quotients of FA groups have FA: an action of a quotient pulls back to an action of the original group. If \(N\triangleleft G\) and both (N) and (G/N) have FA, then (G) has FA: the (N)-fixed subtree is nonempty and (G/N) fixes a vertex in it. If a finite-index subgroup has FA, its supergroup has FA. The reverse subgroup implication fails; for example, an FA group may contain an infinite cyclic subgroup, and (mathbb Z) does not have FA.[1]

Clarity

The fastest diagnostic is to search for a witness to failure rather than enumerate all actions. A surjection onto (mathbb Z) gives the translation action on the integer line. A nontrivial amalgam or HNN decomposition gives its Bass–Serre tree. A strictly increasing countable exhaustion gives the cofinality obstruction. If none exists, Serre's theorem supplies the universal fixed-point conclusion.

For countable groups, one may replace “not a strictly increasing union” with “finitely generated.” This does not reduce the theorem to “finitely generated and finite abelianization.” Finite abelianization blocks maps onto (mathbb Z), but a finitely generated group with finite abelianization may still split as a nontrivial amalgam and fail FA. Conversely, saying merely “does not split” ignores the line-action and cofinality obstructions.

When an action permits inversions, subdivide every edge before applying the vertex formulation. This convention also distinguishes a global fixed point in the geometric realization from a fixed original vertex.

Manages Complexity

FA compresses an infinite family of geometric tests into three algebraic obstruction classes. Direct verification would require examining every tree and every homomorphism \(G\to\operatorname{Aut}(T)\). Serre's criterion turns that impossible search into tractable questions about generation, quotients, and splittings.

The abstraction also joins two languages. A presentation or amalgam normal form produces an action; orbit geometry and stabilizers reveal a graph-of-groups decomposition. This bidirectional dictionary makes a failure constructive: rather than reporting only “not FA,” one can exhibit a quotient map, a normal form, or an explicit fixed-point-free tree action.

Abstract Reasoning

  1. If \(G\twoheadrightarrow\mathbb Z\), pull back the translation action of (mathbb Z) on its simplicial line; (G) fails FA.
  2. If (G=A*_C B) nontrivially, the corresponding Bass–Serre tree has vertex stabilizers conjugate to (A) and (B), and no vertex is fixed by all of (G); FA fails.
  3. If (G) is a nontrivial HNN extension, its stable-letter exponent map onto (mathbb Z) already disproves FA.
  4. If (G) is countable and not finitely generated, enumerate its elements and form a strictly increasing chain of finitely generated subgroups; the third obstruction fails.
  5. If (G) is finitely generated and torsion, it has no (mathbb Z) quotient, and a nontrivial amalgam would supply a reduced word of infinite order; Serre's criterion yields FA.
  6. If (G) has Property (T), Watatani's theorem yields FA; no converse inference is valid.
  7. If (G) has FA and (Q) is a quotient, every (Q)-action pulls back, so (Q) has FA.
  8. If a subgroup \(H\leq G\) lacks FA, that alone does not disprove FA for (G); restriction of a (G)-action and extension of an arbitrary (H)-action are asymmetric operations.
  9. For finitely generated (G), FA implies finite abelianization: its finitely generated abelianization has no infinite cyclic quotient and is therefore finite.
  10. A bounded orbit in a tree has a center; under the inversion-free convention this yields a fixed vertex. Thus bounded-orbit formulations and fixed-vertex formulations agree after the convention is normalized.

Knowledge Transfer

Literal reuse occurs wherever groups act on simplicial trees: Bass–Serre decompositions, arithmetic groups, rigidity theory, automorphism groups, and homomorphism problems. The same roles—group, tree, action, global stabilizer, quotient, amalgam, and exhaustion—remain intact.

The structural residue transfers through existing abstractions. Fixed Point supplies the unchanged object. Decomposition describes the splittings that a fixed-point-free action exposes. Necessity and Sufficiency organizes Serre's exact obstruction theorem. Constraint captures the restriction imposed on possible actions and quotients. These ideas travel broadly, but calling an organizational hierarchy or software tree “Property FA” would be metaphor, not literal transfer.

Examples

  • Finite groups. For an inversion-free action, a finite orbit spans a finite invariant subtree. Its center is fixed, so every finite group has FA.
  • The infinite cyclic group. (mathbb Z) translates its Cayley line and fixes no vertex; equivalently, the identity map is a quotient onto (mathbb Z). It does not have FA.
  • Free group (F_2). Its action on its Cayley tree is fixed-point-free. Algebraically \(F_2\cong\mathbb Z*\mathbb Z\), a nontrivial amalgam over the trivial group.
  • Finitely generated torsion groups. They meet the criterion and have FA; the conclusion does not require the group to be finite.
  • \(\mathrm{SL}_3(\mathbb Z)\). This arithmetic group has Kazhdan's Property (T), and hence FA by Watatani's implication.[4][3]
  • \(\mathrm{SL}_2(\mathbb Z)\). It fails FA because it has the nontrivial splitting (C_4*_{C_2}C_6), whose Bass–Serre tree has no global fixed vertex.[1]
  • An FA group containing a non-FA subgroup. \(\mathrm{SL}_3(\mathbb Z)\) contains infinite cyclic subgroups, showing that FA is not inherited by arbitrary subgroups.
  • A quotient test. If a proposed FA group has an abelianization with an infinite cyclic factor, the induced quotient and line action immediately refute the proposal.

Structural Tensions

  • universal actions vs. finite certificates — the definition quantifies over all trees, while the theorem reduces failure to a short obstruction witness;
  • geometry vs. algebra — fixed vertices are geometric, whereas quotients and splittings are algebraic descriptions of the same obstruction;
  • elementwise fixed sets vs. global intersection — each element may be elliptic without an immediately evident common fixed vertex;
  • subgroup restriction vs. quotient inheritance — FA descends to quotients but not to arbitrary subgroups;
  • simplicial trees vs. broader tree spaces — enlarging the test class leads toward FR and changes the property;
  • inversion convention vs. apparent counterexample — an edge flip may lack a fixed original vertex while fixing the edge midpoint;
  • countable shorthand vs. general groups — finite generation replaces the chain condition only under the countability qualification.

Structural–Framed Character

Property FA is structural. Its truth is invariant under group isomorphism and follows from formal action and splitting criteria. “Rigidity” may carry evaluative overtones in exposition, but no social, historical, or institutional judgment affects membership.

Structural Core vs. Domain Accent

The structural core is universal family of transformations + required common invariant + finite obstruction catalog + equivalence theorem. The domain accent is essential: group homomorphisms, simplicial tree automorphisms, edge inversions, vertex stabilizers, integer quotients, amalgamated products, HNN extensions, and ascending subgroup unions. Removing those roles leaves a generic fixed-point or constraint pattern, not Serre's Property FA.

  • Fixed Point — every eligible action must admit a vertex invariant under the whole group; this is the proposed minimal dependency.
  • Decomposition — nontrivial graph-of-groups splittings are the geometric failure mode exposed by Bass–Serre trees.
  • Necessity and Sufficiency — Serre's three obstructions form an exact criterion, with the countability qualification preserved.
  • Constraint — FA restricts quotients, splittings, and tree representations.
  • Tree (Graph Theory) — supplies the connected acyclic carrier, but not the action property.

The prospective DAG uses one strict composition/presupposes edge to prime:fixed_point. Property FA is not a strict subtype of a fixed point; it is a group property defined by universally requiring fixed points.

Relationships to Other Abstractions

Local relationship map for Serre's Property FAParents appear above the current abstraction, mutual partners to the right, and children below. Node labels state whether each abstraction is prime or domain-specific; colors identify relation types.Serre's Property FADOMAINPrime abstraction: Fixed Point — presupposesFixed PointPRIME

Current abstraction Serre's Property FA Domain-specific

Parents (1) — more general patterns this builds on

  • Serre's Property FA presupposes Fixed Point Prime

    every eligible action must admit a vertex invariant under the whole group; this is the proposed minimal dependency.

Hierarchy path (1) — routes to 1 parentless root

Neighborhood in Abstraction Space

Serre's Property FA sits in a sparse region of the domain-specific corpus (83rd percentile for distinctiveness): few abstractions share its structure, so a faithful description tends to retrieve it precisely.

Family — Unclustered & Miscellaneous (1565 abstractions)

Nearest neighbors

Computed from structural-signature embeddings · 2026-09-08

Not to Be Confused With

  • one action that happens to have a fixed vertex;
  • a group action with a fixed end, invariant axis, or merely bounded behavior before conventions are normalized;
  • Property FR for actions on real trees;
  • Property FW or fixed-point properties for median graphs and cube complexes;
  • Kazhdan's Property (T), which is sufficient but not necessary;
  • fixed-point properties for affine Hilbert-space actions;
  • finite generation plus finite abelianization without the nonsplitting condition;
  • simple, directly indecomposable, or one-ended group properties;
  • HNN indecomposability treated as an independent fourth condition;
  • an action with edge inversion misclassified without barycentric subdivision.

References

[1] Jean-Pierre Serre, Trees, translated by John Stillwell, Springer Monographs in Mathematics, Springer, 2003, especially Chapter I on groups acting on trees and Property FA, ISBN 978-3-540-44237-0. registry ↩a ↩b ↩c ↩d ↩e

[2] Jean-Pierre Serre, “Amalgames et points fixes,” in Proceedings of the Second International Conference on the Theory of Groups, Lecture Notes in Mathematics 372, Springer, 1974, pp. 633–640, MR0376882. registry

[3] Yasuo Watatani, “Property T of Kazhdan implies property FA of Serre,” Mathematica Japonica 27 (1982), pp. 97–103, MR0649023. registry ↩a ↩b

[4] Bachir Bekka, Pierre de la Harpe, and Alain Valette, Kazhdan's Property (T), New Mathematical Monographs 11, Cambridge University Press, 2008, sections on fixed-point properties and the standard arithmetic examples. registry

[5] “Serre's property FA,” Wikipedia, frozen revision 1370737743, https://en.wikipedia.org/wiki/Serre%27s_property_FA. Discovery provenance only; acceptance does not depend on Wikipedia. registry