TL;DR
This paper generalizes the classical Schur orthogonality relations—originally designed for finite groups—to the infinite setting of Gromov hyperbolic groups. By averaging matrix coefficients over expanding annuli in the group, the authors prove that "boundary representations" of free groups are monotonous, providing a major piece of evidence for a long-standing conjecture in the classification of unitary representations.
What Problem Does This Paper Address?
In the representation theory of finite groups, Schur orthogonality is a fundamental tool. it tells us that the matrix coefficients of irreducible representations are orthogonal to each other. However, for infinite groups like the free group or fundamental groups of negatively curved manifolds, the standard sum used in Schur orthogonality diverges.
The challenge is to find a way to "average" these coefficients over the group as it grows toward infinity. Specifically, the authors look at Gromov hyperbolic groups, which are a broad class of groups that geometrically resemble trees or negatively curved spaces. They aim to prove an asymptotic version of orthogonality and use it to classify representations of free groups according to their "monotony"—a property describing the uniqueness of how a representation can be realized on the group's boundary.
Methods
The authors employ a mix of geometric group theory and functional analysis:
- Asymptotic Averaging: Instead of a global sum, they consider the limit of averages over an annulus $A_R$ (elements at a distance $R$ from the identity) as $R \to \infty$.
- Patterson-Sullivan Measures: They construct "boundary representations" using these measures, which are naturally supported on the Gromov boundary of the group.
- Property RD (Rapid Decay): They utilize a strengthened version of Property RD for hyperbolic groups. This property allows them to control the $L^2$-norm of group algebra elements by their operator norm, which is crucial for proving the convergence of their asymptotic sums.
- Equidistribution Results: The authors prove a new equidistribution theorem for the product of Patterson-Sullivan measures on the boundary, extending classical results by Margulis and Roblin to the more general setting of roughly geodesic metrics.
Datasets
Not discussed in the paper. This is a theoretical mathematics paper focused on proving theorems within the framework of group theory and operator algebras.
Key Insights and Ideas
- Normalization via Harish-Chandra Functions: The "aha moment" involves normalizing the matrix coefficients by the Harish-Chandra function $\Xi(g)$ (the matrix coefficient of the constant function 1). This normalization ensures that the limit as $R \to \infty$ is well-behaved and converges to a meaningful inner product on the boundary.
- Metric Generality: The results are not restricted to the standard word metric. They apply to a wide class of invariant metrics $d \in \mathcal{D}(\Gamma)$, which includes the "Green metric" associated with random walks.
- Monotony as a Rigidity Phenomenon: The paper shows that for free groups, boundary representations are "perfect" and unique. This means there is only one way to realize these representations as $L^2$ spaces on the boundary, which is a form of representation-theoretic rigidity.
Key Results
- Asymptotic Orthogonality Theorem: The authors establish a precise limit formula (Theorem 3.3) showing that the weighted average of matrix coefficients over an annulus converges to the inner product of the vectors involved, scaled by boundary functions.
- Proof of Monotony: They prove that boundary representations of non-abelian free groups are monotonous (Theorem 6.2). This confirms that these representations satisfy the "Good Vector Bound" (GVB) criteria established by Kuhn and Steger.
- Application to Random Walks: They extend these results to harmonic measures of a wide class of random walks (Corollary 6.4), showing that representations arising from random walk dynamics also exhibit monotony.
Limitations and Future Work
- Non-Uniform Lattices: The authors note that while their results likely hold for non-uniform lattices (like $SL(2, \mathbb{Z})$), the current proof relies on Property RD. In the non-uniform case, Property RD might not hold relative to the Riemannian metric, requiring a different analytical approach.
- The Full Kuhn-Steger Conjecture: While this paper proves monotony for boundary representations, the full conjecture—that every irreducible representation weakly contained in the regular representation is either monotonous, duplicitous, or odd—remains an open challenge.
- Higher Rank Groups: The current work is firmly rooted in the "rank one" world of hyperbolic groups. Extending these asymptotic relations to higher-rank groups (like $SL(3, \mathbb{Z})$) would require entirely different geometric tools.
Summary updated