Random Hyperbolic 3-Orbifolds & Spectral Gaps

• Random hyperbolic 3-orbifolds are quotients of ℍ³ by discrete Coxeter group actions, notably using Apollonian group constructions and random covering techniques.
• Spectral analysis of the Laplace–Beltrami operator reveals eigenvalue distributions and distinct spectral gaps, highlighting key geometric and arithmetic features.
• Explicit methods using random perfect matching covers and doubling procedures provide actionable insights into eigenvalue bounds and cusp behavior in these orbifolds.

Random hyperbolic 3-orbifolds are geometric structures arising as quotients of hyperbolic three-space $\mathbb{H}^3$ by discrete group actions, constructed via Coxeter group symmetries and random covering procedures. Two principal models are defined using the Apollonian group and the super Apollonian group, leading to explicit constructions of geometrically finite orbifolds and their random covers. The spectral properties of these models, particularly the behavior of the Laplace–Beltrami operator, reveal fundamental insights into their eigenvalue distributions and spectral gaps, as well as deep connections to conjectures about the spectrum of hyperbolic 3-orbifolds (Hide et al., 15 Dec 2025).

1. Group-Theoretic Construction of Models

The foundational object is the super Apollonian group $\Gamma_{\mathrm{SA}}$, an abstract right-angled Coxeter group generated by eight involutions $$r_1,r_2,r_3,r_4,r_1^\perp,r_2^\perp,r_3^\perp,r_4^\perp$$ subject to $r_i^2=e$ for all $i$, and commutation relations determined by adjacency in the 1-skeleton of the cube. $\Gamma_{\mathrm{SA}}$ admits a concrete embedding into $$\mathrm{Isom}(\mathbb{H}^3)\simeq \mathrm{PGL}(2,\mathbb{Z}[i])\rtimes\langle\rho\rangle$$ with generators mapped to reflections in faces of a regular right-angled ideal octahedron $O\subset\mathbb{H}^3$. The quotient orbifold $$M_{\mathrm{SA}} := \Gamma_{\mathrm{SA}}\backslash\mathbb{H}^3$$ is isometric to the ideal octahedron with mirror faces.

The Apollonian group $\Gamma_{\mathrm{Ap}}$ is the index-$\infty$ subgroup of $\Gamma_{\mathrm{SA}}$ generated by four reflections in pairwise non-adjacent faces of $O$, abstractly isomorphic to $$(\mathbb{Z}/2\mathbb{Z})*\dots*(\mathbb{Z}/2\mathbb{Z})$$ (four factors). Its quotient orbifold $$M_{\mathrm{Ap}} := \Gamma_{\mathrm{Ap}}\backslash\mathbb{H}^3$$ is geometrically finite, of infinite volume, and its limit set is the classical Apollonian gasket.

Further, a surjection $$\pi\colon\Gamma_{\mathrm{SA}} \to (\mathbb{Z}/2\mathbb{Z})^*{}^4$$, sending the four $\perp$-generators to involutions in the free product and killing the others, defines the infilonian group $\Gamma_\infty$, satisfying $$\Gamma_{\mathrm{Ap}}\triangleleft\Gamma_\infty\triangleleft\Gamma_{\mathrm{SA}}$$.

For each $n$, selecting a homomorphism $$\varphi_n\colon(\mathbb{Z}/2\mathbb{Z})^*{}^4 \to \mathrm{Sym}(2n)$$ by mapping generators to independent uniform perfect matchings, yields two random degree-$2n$ covers: $$S_n = \mathrm{Stab}_{\varphi_n}(1)\backslash\mathbb{H}^3 \to M_{\mathrm{Ap}}$$ and $$M_n = \mathrm{Stab}_{\varphi_n\circ\pi}(1)\backslash\mathbb{H}^3 \to M_{\mathrm{SA}}$$. $S_n$ is an infinite-volume manifold cover of $M_{\mathrm{Ap}}$, while $M_n$ is a geometrically finite orbifold covering $M_{\mathrm{SA}}$, with certain interior faces now realized as ideal triangles. Mirror-doubling along remaining mirrors produces a closed manifold cover $DM_n$ of degree $4n$.

2. Laplace–Beltrami Operator and Spectral Analysis

In the upper-half-space model $$\mathbb{H}^3=\{(z,t)\in\mathbb{C}\times\mathbb{R}_{>0}\}$$, endowed with the metric $ds^2=(|dz|^2+dt^2)/t^2$, the Laplace–Beltrami operator acts as: $$\Delta f = -t^2\Big( \frac{\partial^2}{\partial x^2} + \frac{\partial^2}{\partial y^2} + \frac{\partial^2}{\partial t^2} \Big) + t \frac{\partial}{\partial t}$$ For orbifolds $X=\Gamma\backslash\mathbb{H}^3$, $\Delta$ acts on $L^2(X)$, respecting Neumann or Dirichlet boundary conditions on mirror faces.

The automorphic eigenvalue problem seeks spectral data for functions $f\in L^2(X)$ satisfying $\Delta f = \lambda f$ and prescribed boundary conditions (Dirichlet: $f|_{\text{mirror}}=0$, Neumann: $\partial f/\partial n|_{\text{mirror}}=0$). At each rank-2 cusp, moderate growth is enforced in the Fourier expansion, with the spectral parameter $\lambda=1-s^2$ causing the “outgoing” term $t^s$ to vanish for $s>1$—defining the cusp boundary condition.

The spectrum $\sigma(\Delta)$ decomposes into:

• Residual spectrum: Poles of Eisenstein series in $(1,2]$,
• Discrete (“cuspidal”) eigenvalues in $(0,1)$,
• Continuous spectrum $[1,\infty)$.

3. Explicit Spectral Gap Results

Two main theorems establish spectral gap properties of the random covers.

Theorem A (infinite-volume covers of $\Gamma_{\mathrm{Ap}}$): Fix $\epsilon>0$. As $n\rightarrow\infty$, with probability approaching one,

$$\sigma(\Delta_{S_n}) \cap [0,1-\epsilon] = \{\lambda_0(\Gamma_{\mathrm{Ap}} \backslash \mathbb{H}^3)\}$$

i.e., no new $L^2$-eigenvalues below $1-\epsilon$ except the Patterson–Sullivan eigenvalue of the base orbifold.

Theorem B (finite-volume covers of $\Gamma_{\mathrm{SA}}$): Let $M_n$ and $DM_n$ be as defined; with $\lambda_1(X)$ the first nonzero eigenvalue on $X$, as $n\rightarrow\infty$:

1. $$\lambda_1(M_n) \to \lambda_0(\Gamma_\infty\backslash\mathbb{H}^3)$$ in probability,
2. On $DM_n$, the even (Neumann) spectrum has

$$\lambda_1^N(DM_n) \to \lambda_0(\Gamma_\infty\backslash\mathbb{H}^3)$$

and the odd (Dirichlet) spectrum satisfies $\lambda_1^D(DM_n) > 1-\epsilon$ with high probability.

Explicit lower bounds are established using residual-spectrum input, a doubling trick, strong-convergence of permutation representations, and geometric control of cusp contributions. For large $n$,

$$\lambda_1(M_n) \geq \lambda_0(\Gamma_\infty\backslash\mathbb{H}^3) - o(1)$$

Numerical bounds for $\lambda_0(\Gamma_\infty\backslash\mathbb{H}^3)$ by Brooks–Burger transfer and Coulon’s inequalities are: $$\frac{3-\sqrt{5+2\sqrt{3}}}{67-\sqrt{5+2\sqrt{3}}} \lesssim \lambda_0(\Gamma_\infty) \lesssim \delta(\Gamma_{\mathrm{Ap}})\cdot[1-\delta(\Gamma_{\mathrm{Ap}})/4]$$ with $\delta(\Gamma_{\mathrm{Ap}})=1.3056867\dots$ the critical exponent, yielding $$0.0014\dots \leq \lambda_0(\Gamma_\infty) \leq 0.87948\dots$$

4. Bass Note Spectrum: Definition and Density Results

The bass-note spectrum of finite-volume hyperbolic 3-orbifolds is

$$\mathrm{Bass}(\mathrm{Hyp}^3_{\mathrm{f.v.}}) := \{\lambda_1(X) : X \text{ a finite-volume hyperbolic 3-orbifold}\}$$

with the analogous definition for arithmetic orbifolds. Sarnak’s conjecture posits

$$\text{closure of Bass}(\mathrm{Hyp}^3_{\mathrm{f.v.}}) = [0,1] \cup E$$

where $E\subset(1,\infty)$ is a discrete infinite upper-bounded set.

Theorem C asserts

$$[0,\,\lambda_0(\Gamma_\infty\backslash\mathbb{H}^3)] \subseteq \text{closure of Bass}(\mathrm{Hyp}^3_{\mathrm{arith}}) \subseteq \text{closure of Bass}(\mathrm{Hyp}^3_{\mathrm{f.v.}})$$

The proof utilizes random covers $M_n$ of $M_{\mathrm{SA}}$ with $\lambda_1(M_n)\to\lambda_0(\Gamma_\infty)$. For finite-index subgroups $\Gamma < \Gamma_{\mathrm{SA}}$, all degree-2 covers $Y\to\Gamma\backslash\mathbb{H}^3$ span the full range $[0,\lambda_0(\Gamma_\infty)]$, with small perturbations realised by “simple switchings.” These procedures keep all resulting orbifolds arithmetic.

5. Examples and Numeric Computations

A concrete Apollonian orbifold is generated by $\Gamma_{\mathrm{Ap}}$ acting as reflections in four vertical hemispheres in $\mathbb{H}^3$, boundary circles orthogonal at $0,1,i,1+i$. The fundamental domain is

$$D= \{\Re z\in[0,1],\, \Im z\in[0,1],\, |z-\tfrac12-\tfrac12i|^2+t^2 \geq 1\}$$

Numerical evaluation (Vytnova–Wormell) yields $\delta(\Gamma_{\mathrm{Ap}})=1.305686728\dots$, and thus $$\lambda_0(\Gamma_{\mathrm{Ap}})=\delta(2-\delta) = 0.906555\dots$$

For $M_{\mathrm{Ap}}$ (infinite-volume), the only $L^2$ eigenvalue below $1$ is $\lambda_0 \approx 0.9066$, followed by continuous spectrum $[1,\infty)$. For $M_{\mathrm{SA}}=O$ (all 8 faces mirrored),

• Dirichlet spectrum begins at $\lambda^1_D(O)\approx 1.308\dots$,
• Neumann spectrum begins at $\lambda^1_N(O)=0$, then $\lambda^2_N(O)=1.000\dots$.

For $M_2$ (random degree-4 cover of $M_{\mathrm{SA}}$), numerical approximations via finite element methods (e.g., FreeFEM++) give:

• $\lambda_1(M_2)\approx 0.85\dots$,
• $\lambda_2(M_2)\approx 1.02\dots$, illustrating the near-optimal spectral gap.

As $n$ increases, histograms of eigenvalues for $M_n$ demonstrate concentration: the lowest new eigenvalue clusters near $\lambda_0(\Gamma_\infty)$, while the old spectrum remains bounded below by $1$. For $n\approx50$, numerical evidence suggests $|\lambda_1(M_n)-0.87|<0.02$ with high probability.

6. Techniques and Proof Structure

The spectral gap results rely on several techniques:

• Residual-spectrum vanishing, ensuring old spectrum above the threshold,
• Doubling tricks distinguishing even and odd spectra,
• Strong-convergence of permutation representations (cf. Bordenave–Collins),
• Geometric control of cusp contributions via ball-packing arguments,
• Use of delocalization, IMS-localization, and tangle-freeness to manipulate cover spectra via switchings.

Stepwise, the proof proceeds by controlling the base orbifold spectrum, eliminating new low-lying eigenvalues, and combining estimates to fix the bottom of the spectrum in random covers.

7. Context and Conjectural Implications

This framework elucidates the distribution of Laplacian eigenvalues in random hyperbolic 3-orbifolds related to Apollonian groups, supporting conjectures on the density and closure of the bass note spectrum. The explicit construction of random covers and the quantification of their spectral gaps provide a robust method for probing the distribution of "low-frequency" eigenvalues in finite-volume hyperbolic orbifolds, with implications for the study of arithmetic and geometric properties of 3-manifolds. A plausible implication is the approachability of the full spectral range $[0,\lambda_0(\Gamma_\infty)]$ through arithmetic orbifold covers (Hide et al., 15 Dec 2025).