Engineering topology in waveguide arrays

The topological classification of a system depends on the discrete symmetries of its Hamiltonian. In Floquet photonic waveguide arrays, the abstract symmetries of the Altland–Zirnbauer (AZ) scheme – chiral, particle-hole, and time-reversal (for photonics, $z$-reversal) – arise from structural properties of the lattice, yet a systematic correspondence has not been established. Here, we illustrate this correspondence for a simpler system of one-dimensional waveguide arrays with real coupling coefficients, showing how bipartite structure and $z$-reflection symmetry alone determine the whole AZ class. We further demonstrate that non-bipartite networks – lacking conventional particle-hole symmetry, chiral symmetry, and $z$-reversal symmetry – can nonetheless support topologically protected boundary states at quasienergy $\varepsilon = π$, even in one dimension. The protecting symmetry – \textit{shifted}-particle-hole symmetry – applies equally to higher-dimensional Floquet waveguides.

💡 Research Summary

The paper investigates the relationship between lattice structural properties and the Altland‑Zirnbauer (AZ) symmetry classification in one‑dimensional photonic waveguide arrays with real (evanescent) coupling. By treating the paraxial wave equation as a Schrödinger‑type equation, the propagation coordinate z plays the role of time, allowing both static and periodically driven (Floquet) configurations. The authors focus on two structural ingredients: (i) bipartite (or sublattice) structure, where sites can be divided into two families A and B with couplings only between families, and (ii) z‑reflection symmetry, i.e., the existence of an axis z₀ about which the coupling pattern is mirror‑symmetric.

For real couplings, z‑reflection automatically yields a unitary operation that, combined with complex conjugation, becomes an effective “z‑reversal” symmetry (the photonic analogue of time‑reversal). The bipartite condition forces the Hamiltonian into an off‑diagonal block form, which anticommutes with the sublattice operator Σ_z. This anticommutation is precisely the definition of chiral symmetry (CS). Consequently, when both bipartite structure and z‑reflection are present, the system simultaneously possesses chiral symmetry, z‑reversal symmetry, and particle‑hole symmetry (the latter being the product of the first two). The full set of AZ symmetries places the system in class BDI, which in one dimension supports an integer‑valued winding number.

The paper then turns to non‑bipartite lattices, where conventional particle‑hole and chiral symmetries are absent. The authors introduce “shifted particle‑hole symmetry” (s‑PHS), in which the particle‑hole operation is accompanied by a fixed momentum shift k₀. This symmetry can exist even without bipartite structure and protects boundary states at quasienergy ε = π (the Floquet zone edge). A concrete three‑waveguide network is designed to illustrate s‑PHS: the three sites are coupled in a cyclic fashion with real couplings that break sublattice balance but retain the shifted particle‑hole relation. Numerical simulations of the Floquet evolution operator show localized edge modes at ε = π, confirming the topological protection afforded by s‑PHS.

Beyond the qualitative description, the authors derive explicit constraints on the Hamiltonian imposed by each symmetry. Chiral symmetry forces the determinant to change sign under z → −z, leading to det H(z₀)=0 for odd band numbers, and requires a vanishing trace at chiral‑symmetric points. Particle‑hole symmetry imposes similar determinant and trace conditions, but they must hold for all z, not only at special points. These algebraic criteria provide practical checks for experimental realizations.

The manuscript also discusses extensions to complex couplings (synthetic gauge fields) and to systems where polarization acts as a pseudospin. In such cases, z‑reflection and z‑reversal are no longer equivalent, and the anti‑unitary nature of T_z must be retained. The squared value T_z² can be +1 (spinless) or –1 (spin‑½), opening the possibility of Kramers degeneracy in photonic platforms.

In summary, the work establishes a clear, systematic mapping: bipartite lattice + z‑reflection ⇒ full AZ class BDI; absence of bipartiteness ⇒ possible shifted particle‑hole symmetry protecting ε = π edge states. This mapping clarifies how structural design choices directly dictate topological classification in Floquet photonic waveguide arrays, and it opens pathways to engineer higher‑dimensional topological phases using similar symmetry‑based principles.