Expanders, rank and graphs of groups

Marc Lackenby

TL;DR

Marc Lackenby establishes a powerful "trichotomy" for finitely presented groups. He proves that for any lattice of finite index normal subgroups, the group must either exhibit a specific structural decomposition (amalgamated free product or HNN extension), possess the geometric expansion property known as Property $(\tau)$, or have a "rank gradient" of zero. This work bridges the gap between combinatorial group theory and the geometry of finite Cayley graphs.

What Problem Does This Paper Address?

In group theory, we often want to understand the relationship between a large group $G$ and its finite index subgroups $G_i$. Specifically, how do the algebraic properties of these subgroups (like their "rank"—the minimum number of generators needed) relate to the geometry of the quotient groups $G/G_i$?

A central challenge is understanding the "rank gradient." If $G$ has rank $d(G)$, a standard result (Reidermeister-Schreier) gives an upper bound for the rank of a subgroup $G_i$ based on its index $[G : G_i]$. The rank gradient measures how much smaller the actual rank is compared to this theoretical maximum as the index goes to infinity. Lackenby seeks to categorize groups based on this growth and other structural properties, providing a roadmap for when a group can be decomposed into simpler pieces.

Methods

The paper utilizes Geometric Group Theory, specifically looking at groups through the lens of topology and graph theory:

  • Cayley Graphs: The author views finite quotients $G/G_i$ as geometric objects (graphs) and studies their Cheeger constant, which measures how difficult it is to "cut" the graph into two large, disconnected pieces.
  • 2-Complexes and Covering Spaces: To prove the main theorem, Lackenby represents the group $G$ as the fundamental group of a finite 2-complex $K$. He then analyzes the covering spaces $K_i$ corresponding to the subgroups $G_i$.
  • Seifert-Van Kampen Theorem: By decomposing these covering spaces into overlapping subsets based on the geometry of the Cayley graphs, he uses this classical topological tool to show that the subgroup $G_i$ must be a "pushout" of other groups, leading to the amalgamated free product or HNN extension structure.

Datasets

Not discussed in the paper. As a theoretical mathematics paper, it relies on logical proofs and existing mathematical definitions rather than empirical datasets.

Key Insights and Ideas

  • The Trichotomy (Theorem 1.1): This is the "aha moment." It suggests that if a group's quotients aren't "expanding" (meaning they are easy to cut in half) and the subgroup ranks are growing quickly (non-zero rank gradient), then the subgroups must have a very specific algebraic structure (decomposing as a graph of groups).
  • Rank Gradient as a Metric of Complexity: The paper formalizes "rank gradient" as a way to distinguish "special" groups (like mapping tori or certain arithmetic groups) from more "generic" or complex groups.
  • Connection to 3-Manifolds: Lackenby draws a direct parallel between this group-theoretic result and the Virtually Haken Conjecture in 3-manifold topology. He shows that the algebraic growth of a group is deeply tied to the topological complexity of the manifolds they represent.

Key Results

  • Theorem 1.1 (The Main Result): For a finitely presented group $G$ and a lattice of normal subgroups $\{G_i\}$, at least one of these holds:
  • $G_i$ is an amalgamated free product or HNN extension for infinitely many $i$.
  • $G$ has Property $(\tau)$ (the quotients are expander graphs).
  • The rank gradient is zero.
  • Amenable Groups (Theorem 1.2): Lackenby proves that any finitely presented, residually finite, amenable group must have a rank gradient of zero. This provides a clear algebraic signature for amenability.
  • Free Quotients (Theorem 1.3): He provides a set of equivalent conditions for when a subgroup of $G$ can map onto a non-abelian free group, linking this to the growth of the first Betti number and rank.

Limitations and Future Work

  • Proving Non-Zero Rank Gradient: The author notes that proving a group has a non-zero rank gradient is notoriously difficult because finding lower bounds for the rank of a group is a hard problem in combinatorial group theory.
  • Unresolved Theoretical Possibility: Lackenby points out a gap in current knowledge: it is unknown if there exist finitely presented groups that fail Property $(\tau)$ but do not have a finite index subgroup with infinite abelianization.
  • Constructing Expanders: While the paper relates expansion to rank, it acknowledges that constructing explicit families of expander graphs remains a difficult task in the field.

Summary updated

Abstract

Let G be a finitely presented group, and let {G_i} be a collection of finite index normal subgroups that is closed under intersections. Then, we prove that at least one of the following must hold: 1. G_i is an amalgamated free product or HNN extension, for infinitely many i; 2. the Cayley graphs of G/G_i (with respect to a fixed finite set of generators for G) form an expanding family; 3. inf_i (d(G_i)-1)/[G:G_i] = 0, where d(G_i) is the rank of G_i. The proof involves an analysis of the geometry and topology of finite Cayley graphs. Several applications of this result are given.
BibTeX
@inproceedings{lackenby2005expanders,
  author = {Marc Lackenby},
  title = {{Expanders, rank and graphs of groups}},
  booktitle = {Israel},
  year = {2005},
  doi = {10.1007/bf02773541},
}

Loading related content...

Figure