Counting topological interface modes using simplicial characteristic classes

Nicholas Bohlsen, I. Y. Dodin, Hong QinarXiv 2025 · DOI · arXiv

TL;DR

This paper introduces a robust, derivative-free numerical algorithm to predict the number of topological interface modes (TIMs) in physical systems like fluids and plasmas. By calculating Chern numbers using discrete "simplicial" characteristic classes on a mesh rather than integrating noisy derivatives, the method provides a gauge-invariant and noise-tolerant way to identify topological protections in both simulation and experimental data.

What Problem Does This Paper Address?

Topological interface modes (TIMs) are waves localized at the boundary between two different media (e.g., equatorial waves in the ocean or edge states in quantum Hall insulators). Predicting how many of these modes exist is crucial for understanding wave propagation in complex systems.

The standard mathematical tool for this prediction is the Bulk-Boundary Correspondence, which relates the number of interface modes to a topological invariant called the Chern number. Traditionally, calculating the Chern number requires computing the "Berry curvature," which involves taking derivatives of eigenvectors. This approach has two major flaws in practice:

  1. Noise Sensitivity: Derivatives amplify noise, making the method unreliable for experimental or low-resolution simulation data.
  2. Integer Ambiguity: While the Chern number must be an exact integer, numerical integration of Berry curvature often yields non-integer results that require "guessing" the nearest whole number.

The authors aim to replace this geometric, derivative-based approach with a purely topological, combinatorial one.

Methods

The authors leverage the theory of discrete vector bundles and simplicial cohomology. Instead of treating the parameter space as a smooth manifold, they discretize it into a triangular mesh (a simplicial complex).

Key steps of their algorithm include:

  • Mesh Generation: Sampling points on a sphere in phase space surrounding a "Weyl point" (a point of degeneracy) and connecting them into a mesh.
  • Discrete Trivialization: Solving for local polarization vectors at each mesh vertex.
  • Cech Cocycle Construction: Using Singular Value Decomposition (SVD) to find the "best fit" rotation matrices (transition functions) between adjacent mesh elements, solving what is known as the Orthogonal Procrustes problem.
  • Combinatorial Chern Class: Computing the "winding" of these rotations around each triangle in the mesh. Because this involves rounding to the nearest integer at the local level, the final sum is guaranteed to be an exact integer.
  • Implementation: The algorithm is implemented as an open-source Julia package, DiscreteVectorBundles.jl.

Datasets

The paper uses several physical and mathematical models to validate the algorithm:

  • Tautological Line Bundle: A standard mathematical test case over the Riemann sphere with a known Chern number of -1.
  • Shallow Water (SW) Equations: A model for geophysical fluid dynamics (e.g., waves at the Earth's equator).
  • Topological Langmuir Cyclotron Wave (TLCW): A complex 9-component field model from plasma physics.
  • 2D Dirac Equation: Used to generate synthetic "experimental" data via Wigner transforms to test the algorithm's ability to extract topology from noisy, sampled signals.

Key Insights and Ideas

  • Topology over Geometry: By shifting from a geometric definition (integrating curvature) to a topological one (pairing cohomology classes with homology cycles), the authors eliminate the need for numerical derivatives entirely.
  • Structure Preservation: The algorithm is "structure-preserving," meaning it respects the underlying discrete symmetries of the mesh, ensuring the topological invariants remain robust even if the mesh is coarse or slightly deformed.
  • Wigner Matrix Extraction: The authors demonstrate a bridge to experimental physics by showing that the necessary polarization vectors can be extracted from the Wigner matrix of a wave field—a common representation in signal processing.
  • Orientation Recovery: The paper provides a specific sub-algorithm (geometricorientation) to ensure the mesh's topological orientation matches the physical "outward-pointing" normal, which is essential for getting the sign of the Chern number right.

Key Results

  • Accuracy: The algorithm perfectly reproduced the analytical Chern numbers for the Shallow Water equations ($Ch_1 = -2$ for the positive band) and the TLCW model ($Ch_1 = \pm 1$ for various bands).
  • Robustness: In the synthetic 2D Dirac example, the algorithm consistently recovered the correct Chern number ($Ch_1 = -1$) even when the sampling was relatively sparse (as few as 40–80 points on the sphere).
  • Convergence: The authors showed that as the frequency ($\omega$) of the sampled wave increases (approaching the geometric optics limit), the error in the polarization estimation drops significantly, making the topological extraction highly reliable.

Limitations and Future Work

  • Sampling Thresholds: The authors noted a "heuristic" limit: the algorithm typically requires at least ~50 samples around a sphere to return consistent results. At very low sampling densities (e.g., 20 points), the topological reconstruction can fail.
  • Computational Scaling: While efficient for 2D surfaces, the paper primarily focuses on the first Chern number. Extending this to higher-order invariants (like the second Chern number) or higher-dimensional manifolds would increase computational complexity.
  • Real-World Data: The "experimental" test was performed on synthetic data. The next logical step is applying this to actual laboratory measurements from fluids or plasma experiments where noise and resolution limits are more unpredictable.

Summary updated

Abstract

A computational approach for predicting the number of topological interface modes (TIMs) in hermitian systems using the spectral flow - monopole (SFM) correspondence is presented. The number of TIMs is determined by calculating the Chern number of a complex line bundle of local polarisation vectors over a phase space sphere surrounding a Weyl point. The Chern number is computed by constructing the simplicial first Chern class of a discrete vector bundle on a simplicial mesh. This approach is gauge invariant, derivative free, structure preserving, and robust to noise. The algorithm is shown to reproduce the expected number of TIMs for the case of equatorial fluid waves and the topological Langmuir cyclotron wave. The possibility of using this algorithm to analyse experimental measurements of bulk wave polarisations and predict the associated number of TIMs is explored in a synthetic example.
BibTeX
@misc{bohlsen2025counting,
  author = {Nicholas Bohlsen and I. Y. Dodin and Hong Qin},
  title = {{Counting topological interface modes using simplicial characteristic classes}},
  year = {2025},
  eprint = {2508.01063},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2508.01063},
  howpublished = {arXiv},
}
Figure