🔬 mesoscale physics

Characterizing universal topological phases beyond spatial embedding

This paper establishes a universal framework for characterizing topological phases using only the Hamiltonian, decoupling topological identification from spatial embedding by introducing a connectivity-based spatial reconstruction that extends topological diagnostics to abstract platforms like complex networks.

Original authors: Sunkyu Yu, Xianji Piao, Namkyoo Park

Published 2026-10-22

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Sunkyu Yu, Xianji Piao, Namkyoo Park

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

In the modern study of materials, scientists have discovered that some substances possess a hidden quality called topology. This quality acts like a mathematical fingerprint that protects certain behaviors, such as the flow of electricity or light, from being disrupted by dirt, defects, or disorder. For decades, researchers have mapped these protected states in physical objects like crystals, where the arrangement of atoms follows a strict grid in space. In these familiar settings, the rules for identifying these topological states rely heavily on the physical distance between parts of the material. If you know where every atom sits, you can calculate whether the material has this special, protected quality.

However, the world of physics is expanding beyond solid crystals. Scientists now create systems using sound waves, light, and even abstract networks of connections that do not exist in a simple physical grid. In these new realms, the "location" of a part of the system is often just a label we assign, not a fixed point in space. This creates a puzzle: if the physical arrangement of parts can be changed without altering the system's fundamental rules, how can we be sure we are still looking at the same protected state? The old methods of measurement, which depend entirely on where things are placed, often fail to give a consistent answer in these abstract environments.

A team of researchers from Seoul National University and the University of Seoul has developed a new way to solve this problem. They propose a method to identify these universal topological phases that relies only on the system's internal rules, ignoring the arbitrary placement of its parts. Instead of asking "where is this part?" they ask "how strongly is this part connected to its neighbors?" By analyzing the strength of the connections between every component, they can reconstruct a map of the system that reveals its true topological nature. This approach works for both traditional materials and complex, abstract networks, showing that the essence of these protected states lives in the connections themselves, not in the space they occupy.

The researchers began by demonstrating why the old methods were insufficient. They took a well-known model of a material that exhibits these topological properties and ran a simulation where they kept the internal rules exactly the same but completely scrambled the physical positions of the parts. In the original, orderly arrangement, the standard tools correctly identified the presence of a protected state. But when the parts were rearranged into a chaotic, inside-out pattern, those same tools declared the state to be gone, even though the underlying physics had not changed at all. The tools were fooled by the change in geometry, proving that they were measuring the shape of the container rather than the content inside.

To fix this, the team created a new diagnostic tool that builds its own map from the system's internal connections. They treated the system as a network where the strength of the link between two parts determined how close they should be considered. Using a mathematical technique that groups together the most strongly connected parts, they generated a new set of coordinates for every point in the system. This new map was not based on where the parts were physically placed, but on how they interacted with one another. When they applied their standard measurement tools to this new, connection-based map, the results were consistent. The protected state reappeared, exactly as it did in the original, orderly version. This proved that the topological phase was a real, intrinsic feature of the system's rules, independent of any external arrangement.

The team then tested this method on a wide variety of complex networks to see if it could find these phases in places where no physical grid exists. They looked at networks that ranged from perfectly regular rings to completely random webs of connections. They found that the ability to support these protected states depended heavily on the network's structure. In a perfectly regular ring, the method showed that no protected state could form because there was no distinction between the center and the edge. In a completely random network, a protected state could form, but it was fragile and easily broken.

The most significant finding emerged when they examined networks that sat in between these two extremes, known as small-world networks. These are structures where most connections are local, but a few long-range shortcuts link distant parts of the system together. In these networks, the researchers discovered a sweet spot. The shortcuts allowed the system to organize itself into a clear core and edge, creating a protected state, while the remaining local connections kept the state robust against disorder. This suggests that nature might favor these specific types of network architectures when it needs to maintain stable, protected flows of energy or information.

By decoupling the identification of topological phases from physical space, this work reshapes the foundation of how scientists understand these materials. It shows that the special properties of these systems are not an artifact of how we build them on a table or in a computer, but are fundamental features of the connections themselves. This insight opens the door to designing new materials and systems based on their connectivity patterns rather than their physical layout, potentially leading to more robust technologies for transmitting light, sound, or data in environments where traditional spatial arrangements are impossible.

Technical Summary: Characterizing Universal Topological Phases Beyond Spatial Embedding

Problem Statement
Conventional characterizations of topological phases of matter rely heavily on the spatial embedding of a system's constituents. Whether utilizing reciprocal-space invariants (e.g., Chern numbers) or real-space tools (e.g., Chern markers, Bott indices, spectral localizers), these diagnostics require a predefined geometric arrangement of sites or a specific reciprocal-space manifold. While the underlying wave dynamics and Hamiltonian eigenstates remain invariant under Hamiltonian-preserving transformations, conventional topological metrics often fail to uniquely identify topological phases in abstract platforms where spatial embedding is not intrinsic. This limitation affects synthetic dimensions (frequency modes, Fock states), complex networks, and non-Euclidean geometries, where the "geometry" of the physical domain is decoupled from the connectivity defined by the Hamiltonian. Consequently, a topological phase that is physically robust may appear trivial if the site coordinates are arbitrarily rearranged, conflating intrinsic topology with the geometry of the embedding.

Methodology
The authors propose an embedding-independent framework grounded in the invariance of topological features under arbitrary Hamiltonian-preserving deformations. The core methodology involves constructing a Hamiltonian-derived spatial embedding that groups system constituents based solely on their connectivity, rather than pre-assigned physical coordinates.

Graph Laplacian Construction: The connectivity structure of the system's Hamiltonian ( $H$ ) is encoded into an adjacency matrix $A$ , where $A_{mn} = |H_{mn}|$ . From this, a graph Laplacian $G = D - A$ is constructed ( $D$ being the degree matrix).
Eigenstate-Based Embedding: The authors utilize the eigenstates of the graph Laplacian. Since the lowest-order nontrivial eigenstates ( $v_m$ ) satisfy the condition that $|v_m(i) - v_m(j)|$ is small where the coupling $A_{ij}$ is large, these eigenstates naturally group strongly coupled sites.
Coordinate Definition: For a $d$ -dimensional topological diagnosis, the spatial coordinates $(x_1, \dots, x_d)$ of the $i$ -th site are defined by the $d$ lowest-order nontrivial Laplacian eigenstates that are locally independent. For 2D systems, the coordinates are set as $(x, y)_i = [v_1(i), v_2(i)]$ .
Topological Diagnosis: This derived embedding is used to construct position operators ( $X, Y$ ) for the spectral localizer ( $L$ ). The local Chern index ( $C_L$ ) and local gap ( $\mu_L$ ) are then calculated using $L$ , which now depends only on the Hamiltonian and the intrinsic connectivity, not on external geometry.

Key Results
The study demonstrates that this Hamiltonian-derived embedding successfully recovers established topological characteristics and extends them to complex networks:

Recovery of Canonical Models: Applied to the Hofstadter model on a square lattice, the method reproduces the quantized bulk Chern index ( $C_L = +1$ ) and the vanishing index at the boundary, despite using no prescribed site positions. The method also successfully reconstructs the Hofstadter butterfly spectrum as a function of flux and energy. Similar results were obtained for Kagome lattices, hyperbolic lattices, stealthy hyperuniform (SHU) lattices, and the Haldane model.
Resolution of Embedding Ambiguity: In the Hofstadter model, an "inside-out" transformation of site coordinates causes conventional diagnostics to yield a trivial result ( $C_L = 0$ ) despite the persistence of chiral edge states. The Hamiltonian-derived embedding correctly identifies the nontrivial topology regardless of this spatial rearrangement.
Topological Phases in Complex Networks: The framework was applied to Watts-Strogatz networks (spanning regular, small-world, and random regimes).
- Regular Networks ( $p=0$ ): The embedding yields a hollow circle with no bulk-edge separation, resulting in a trivial topology ( $C_L = 0$ ).
- Small-World Networks ( $p \approx 0.1$ ): Shortcut links create a distinct bulk-edge separation in the derived embedding. This regime supports a robust nontrivial core ( $C_L \approx +1$ ) with a finite local gap ( $\mu_L$ ), indicating a stable topological phase.
- Random Networks ( $p=1$ ): While a nontrivial bulk can be formed, the local gap diminishes significantly, rendering the topological phase fragile.
- Optimal Regime: The results indicate that small-world networks offer an optimal architecture, balancing the "shortcut-induced bulk" necessary for topology with the "clustering" required for robustness.

Significance and Claims
The paper claims to reshape the conceptual foundations of topological physics by decoupling topological characterization from prescribed physical platforms. By establishing a framework where topological phases are defined as universal features preserved under any Hamiltonian-preserving deformation, the authors provide a unified lens for understanding both canonical topological matter and abstract systems like complex networks.

The authors modestly note that while the Laplacian eigenstate construction successfully generalizes conventional diagnostics for local Hamiltonians, it may not capture every topological phase a complex network can host (e.g., those designed via joint optimization of gauge fields and positions in previous work). They suggest that other constructions, such as normalized graph Laplacians, diffusion maps, or magnetic graph Laplacians, may be necessary to capture a broader spectrum of universal phases. Nevertheless, the proposed method offers a natural route to identifying universal topological phases consistent with constituent-resolved wave transport, highlighting the critical role of network architecture (specifically the small-world regime) in supporting robust topological phenomena.

More like this