The Heavy-Tail Topology

Topological protection survives when the network’s degree distribution has a fat tail.

Topological physics lives on lattices. Periodic structures with translational symmetry admit Bloch theory, band structures, and topological invariants — Chern numbers, winding numbers, Z₂ indices. Aperiodic extensions (quasicrystals, fractals, curved spaces) have expanded the playground, but all these structures share one property: their connectivity is bounded. Each node has a small, well-defined number of neighbors.

Palacios, Cassella, and Morfonios (arXiv:2603.10467) break this assumption. They study networks with heavy-tailed degree distributions — where a few hubs connect to an unbounded number of nodes while most nodes connect to few. The degree distribution follows a power law, or worse. In such networks, the standard topological toolkit fails because the adjacency matrix is unbounded: hub nodes dominate any finite-dimensional approximation.

Yet topological protection persists. Edge states remain localized at boundaries. Spectral gaps survive despite the divergent connectivity of hub nodes. The mechanism: topology is a property of the gap structure, not the bandwidth. As long as a gap opens — however irregularly — the topological invariant is defined and the edge states follow.

The network perspective unifies apparently different platforms. A quasicrystal, a fractal, and a heavy-tailed random graph all look different geometrically but may share the same topological class when viewed as networks with specific connectivity patterns. The lattice was always just one way to organize connections. The connections themselves are what topology protects.

The regime this opens is genuinely new: topological phenomena in networks where the variance of the degree distribution diverges.

Palacios, Cassella, and Morfonios, “Topological heavy-tailed networks,” arXiv:2603.10467 (2026).