Kaufman–Oppenheim Coset Complexes

• Kaufman–Oppenheim Coset Complexes are highly structured d-dimensional simplicial complexes defined from group cosets, offering explicit bounded-degree high-dimensional expanders.
• They combine rigorous spectral and coboundary expansion properties—validated through group-theoretic and combinatorial methods—to enable robust error-correcting codes and PCP constructions.
• Their efficient, polylogarithmic computability and well-controlled local parameters make them practical for applications in coding theory, probabilistically checkable proofs, and robust testing.

Kaufman–Oppenheim Coset Complexes are a family of highly structured simplicial complexes constructed from group-theoretic data, providing fundamental examples of bounded-degree high-dimensional expanders (HDX) with explicit and elementary descriptions. These complexes unify combinatorial, spectral, and topological notions of high-dimensional expansion and have broad applications in coding theory, probabilistically checkable proofs (PCPs), and theoretical computer science.

1. Definition and Construction

Kaufman–Oppenheim coset complexes arise from a finite group $G$ and a finite family of subgroups $\{H_i\}_{i\in I}$, where $|I|=d+1$ and $d\geq 2$. The $d$-dimensional coset complex $X=CC(G;(H_i))$ is defined as follows:

• Vertices: $X(0) = \bigsqcup_{i=0}^d G/H_i$, the disjoint union of coset spaces.
• Simplices: A set of vertices $\{g_0H_{i_0}, \ldots, g_kH_{i_k}\}$ spans a $k$-simplex if and only if the corresponding cosets have nonempty intersection: $$g_0H_{i_0} \cap \cdots \cap g_kH_{i_k} \neq \varnothing$$.
• Faces and Links: The structure is downward closed: all lower-dimensional faces exist by inclusion. For any face $\sigma$ of type $T\subset I$, the link $\mathrm{lk}_X(\sigma)$ is isomorphic to $CC\left(H_T; (H_{T\cup\{i\}})_{i\notin T}\right)$, where $H_T = \bigcap_{i\in T} H_i$.

A canonical instance is obtained by taking $G = SL_{n+1}(\mathbb{F}_p[t]/(t^s))$ for $p$ prime and $s>3n$, and $H_i$ the subgroups generated by certain “tiny transvection” elementary matrices. The resulting $n$-dimensional complex is $(n+1)$-partite, pure, and of uniformly bounded degree, independent of the field size for fixed $n$ (O'Donnell et al., 13 Nov 2025, Kaufman et al., 5 Nov 2024, Peralta et al., 10 Jan 2024, Harsha et al., 2019).

2. Group-Theoretic Framework and Relation to Kac–Moody–Steinberg Groups

The coset complex construction generalizes via input from Kac–Moody–Steinberg (KMS) groups and their quotients. For a $d$-spherical, purely $d$-spherical generalized Cartan matrix $A$ (of rank $d+1$) and finite field $k=\mathbb{F}_q$, the KMS group $\mathcal{U}_A(k)$ is defined as a direct limit over unipotent subgroups $U_J \leq \mathcal{G}_A(k)$ for proper spherical $J\subset I$. Quotients of $\mathcal{U}_A(k)$ with the intersection property (IP) yield coset complexes $CC(G; (H_i))$ with the desired expansion and bounded-degree structure.

When $A$ is affine, these constructions recover the classical Kaufman–Oppenheim complexes as quotients of Chevalley groups over $\mathbb{F}_q[t]/(t^s)$, and in the appropriate specialization, extend to $G_2$-type root systems (Peralta et al., 10 Jan 2024).

3. Expansion Properties: Spectral and Coboundary

Kaufman–Oppenheim coset complexes are characterized by strong spectral expansion and coboundary expansion properties:

• Spectral Expansion: For links of codimension 2 (i.e., 1-skeletons of links of $(d-2)$-faces), the adjacency spectra are controlled by explicit group-theoretic calculations:

$$\lambda_2 \leq \begin{cases} 1/\sqrt{q} & \text{(type } A_2) \ \sqrt{2/q} & \text{(type } B_2) \ \sqrt{3/q} & \text{(type } G_2) \end{cases}$$

These bounds suffice, via Oppenheim's “trickle-down” or Garland's descent theorem, to imply global spectral expansion of the complex, making $X$ an $\epsilon$-one-sided HDX with $\epsilon = \lambda/(1-(d-1)\lambda)$ (Peralta et al., 10 Jan 2024, Harsha et al., 2019). Recent work gives near-optimal constants using path-complex trace machinery and C-Lorentzian polynomials (Leake et al., 2 Mar 2025).

• Coboundary Expansion: For any finite coefficient group $A$, there exists a constant $\beta>0$ (independent of the size of the underlying field for $p\gg n$) so that

$h^1_{\text{cobound}}(X;A)\ge\beta$

and $H^1(X;A)=0$, making $X$ a uniform $(A,\beta)$-coboundary expander. The proof uses a global covering argument involving successive quotients and Dehn-function bounds in matrix groups over polynomial rings (Kaufman et al., 5 Nov 2024, O'Donnell et al., 13 Nov 2025).

4. Explicitness, Combinatorial Structure, and Links

Kaufman–Oppenheim complexes are strongly explicit: vertices, neighbors, and faces can be listed and sampled in polylogarithmic time in the ambient group size $|G|$ via linear algebraic operations over finite fields (O'Donnell et al., 13 Nov 2025). The complexes are $(n+1)$-partite, $n$-dimensional, and pure. Each link is again a coset complex of lower dimension, typically isomorphic to a high-quality bipartite expander (such as the affine line graph for $d=2$) (Harsha et al., 2019).

Local parameters are tightly controlled: each vertex lies in $|H_i|$ maximal faces (with $|H_i|=q^{\ell_i}$ for explicit $\ell_i$), and the complexes form infinite families of bounded-degree expanders as $s\to\infty$ or $q\to\infty$ (Peralta et al., 10 Jan 2024, Dinur et al., 2023).

5. Applications: Coding Theory, PCPs, and Robust Testing

Kaufman–Oppenheim coset complexes underpin new constructions of locally testable LDPC codes by serving as Tanner graphs built from their triangles and links. Codewords correspond to global sections whose restrictions to edge links are codewords in Reed–Solomon (or similar) codes. For appropriate parameter regimes this enables symmetric codes with constant distance, multiplication property, locally testable structure, and exponential symmetry group (Dinur et al., 2023).

Recent developments exploit these coset complexes to realize sparse direct product testers of arbitrarily low soundness, a key ingredient in the construction of quasilinear-length PCPs with constant soundness and two queries. All components of these constructions are strongly explicit and elementary, bypassing the need for the representation-theoretic machinery involved in earlier HDX-based PCPs (O'Donnell et al., 13 Nov 2025).

6. Topological, Combinatorial, and Markov Properties

The complexes exhibit topological expansion properties: uniform lower bounds on the $1$-coboundary constant over all finite coefficient groups imply Gromov's topological overlapping property and enable robust agreement testing. Markov chain mixing results on the facets (maximal faces) follow from strong local spectral expansion, with near-optimal rates for mixing and concentration in random walks through the higher-dimensional structure (Leake et al., 2 Mar 2025, Kaufman et al., 5 Nov 2024).

7. Broader Impact and Ongoing Directions

Kaufman–Oppenheim coset complexes serve both as the paradigmatic first elementary high-dimensional (co-)boundary expanders of bounded degree and as a flexible new toolkit for constructing sparse testers and PCP systems with unprecedented explicitness. Their description by "tiny" mod-$t^s$ transvections over elementary matrix groups renders the construction accessible to efficient computation and manipulation, and opens potential generalization to further classes of groups and expanders. Future research aims to extend these paradigms to other group families satisfying analogous axioms (2-spherical generation, Ramanujan local expansion, residual finiteness), as well as to further optimize quantitative constants in the expansion and testing regimes (Peralta et al., 10 Jan 2024, O'Donnell et al., 13 Nov 2025).

---

Key References:

• (Peralta et al., 10 Jan 2024) High-dimensional expanders from Kac–Moody–Steinberg groups
• (Kaufman et al., 5 Nov 2024) Coboundary expansion of coset complexes
• (O'Donnell et al., 13 Nov 2025) Low-soundness direct-product testers and PCPs from Kaufman–Oppenheim complexes
• (Dinur et al., 2023) New Codes on High Dimensional Expanders
• (Harsha et al., 2019) A note on the elementary construction of High-Dimensional Expanders of Kaufman and Oppenheim
• (Leake et al., 2 Mar 2025) Optimal Trickle-Down Theorems for Path Complexes via C-Lorentzian Polynomials with Applications to Sampling and Log-Concave Sequences