Observation of Multiple Topological Corner States in Thermal Diffusion
Pith reviewed 2026-05-06 19:51 UTC · model claude-opus-4-7
The pith
A kagome-patterned thermal metamaterial hosts multiple corner-localized modes of heat diffusion, identifiable by their decay rates.
A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.
Core claim
The authors build a 2D kagome-lattice thermal metamaterial and argue that its diffusion Hamiltonian — though anti-Hermitian, with purely imaginary eigenvalues that encode decay rates rather than oscillation frequencies — supports multiple higher-order topological corner modes. They identify these modes by directly measuring how fast temperature relaxes at different sites: corner-localized eigenmodes show characteristic decay rates distinct from bulk and edge modes. This is presented as the first observation of multiple corner states in a purely diffusive (no wave propagation) setting.
What carries the argument
The diffusion Hamiltonian of a 2D kagome network of thermally coupled cells. Because heat conduction is dissipative, this operator is anti-Hermitian with purely imaginary spectrum, so eigenmodes correspond to decay rates. The kagome geometry's known higher-order topology is transplanted to this setting, and corner eigenmodes are read off from site-resolved temperature decay curves.
If this is right
- Thermal hotspots or cold spots can be pinned at sample corners by lattice topology rather than by insulating barriers, giving a passive route to localized thermal management.
- Higher-order topological classifications developed for Hermitian wave systems carry over, at least phenomenologically, to anti-Hermitian diffusion operators with purely imaginary spectra.
- Decay-rate spectroscopy — measuring relaxation times at each site — becomes a usable analogue of band-structure measurement for diffusive metamaterials.
- Other lattice geometries known to host corner modes in photonics or acoustics (breathing honeycomb, square SSH-like tilings) become candidates for thermal corner-state realization.
Where Pith is reading between the lines
- The same construction should generalize to mass diffusion or charge diffusion in resistor-capacitor networks, where decay rates play the role of imaginary eigenvalues in exactly the same way.
- Because corner modes here are identified by being slow (or fast) relaxers, their utility for thermal design depends on how cleanly their decay rates separate from the continuum of edge and bulk rates — a gap question that becomes the practical figure of merit.
- A natural next step is to add advection (a drift term), which would tilt the spectrum off the imaginary axis and let one study non-Hermitian skin effects and topological corner states in the same diffusive platform.
- If the bulk-boundary correspondence really survives the anti-Hermitian setting, one expects a quantized topological invariant computable from the static thermal coupling matrix alone, independent of any time-domain measurement.
Load-bearing premise
That the corner-localized temperature decays observed in the experiment are genuine topological corner modes inherited from the kagome lattice's classification, rather than ordinary slow relaxation at sites with fewer thermal connections.
What would settle it
Build the same kagome thermal lattice but tune the inter-cell couplings into the topologically trivial regime (intra-cell stronger than inter-cell, or vice versa, depending on convention). The corner sites should then relax with rates indistinguishable from bulk/edge sites, and the multiple distinct corner-mode decay rates seen in the nontrivial sample should disappear. Persistence of the corner-mode signatures across this tuning would falsify the topological interpretation.
Figures
read the original abstract
Higher-dimensional topological meta-materials have more flexible than one-dimensional topological materials, which are more convenient to apply and solve practical problems. However, in diffusion systems, higher-dimensional topological states have not been well studied. In this work, we experimentally realized the 2D topological structure based on a kagome lattice of thermal metamaterial. Due to the anti-Hermitian properties of the diffusion Hamiltonian, it has purely imaginary eigenvalues corresponding to the decay rate. By theoretical analysis and directly observing the decay rate of temperature through experiments, we present the various corner states in 2D topological diffusive system. Our work constitutes the first realization of multiple corner states with high decay rates in a pure diffusion system, which provides a new idea for the design of topological protected thermal metamaterial in the future.
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. The manuscript reports an experimental realization of a 2D kagome-lattice thermal metamaterial whose diffusion dynamics are described by an anti-Hermitian generator with purely imaginary eigenvalues. The authors claim to observe multiple corner-localized modes with characteristic decay rates and identify them as higher-order topological corner states, framing the work as the first such realization in a pure diffusion system. The central experimental signature is the directly measured temperature decay rate at corner sites compared to edge/bulk sites of the finite kagome flake.
Significance. If the topological character of the observed corner modes is established, the result extends the higher-order topological insulator (HOTI) program from Hermitian wave systems (photonic, phononic, electronic) to genuinely dissipative diffusion systems, where the generator is anti-Hermitian rather than Hermitian. That extension is non-trivial and of interest to the thermal metamaterials community, particularly because it would suggest design principles for thermally protected localization. The direct measurement of decay rates as the topological signature, rather than indirect proxies, is a methodological strength. The significance is, however, contingent on the authors actually distinguishing topology-protected corner modes from generic geometric corner localization in a finite kagome cut — see major comments.
major comments (3)
- [Central claim / abstract] The diffusion generator on a kagome network is, up to a factor of i, a weighted graph Laplacian L = D − W. The slowest-decaying eigenmodes of such a Laplacian are biased toward sites with smallest local degree D_ii, and corners of a finite kagome flake are intrinsically under-coordinated relative to edge and bulk sites. Slow-decay corner localization is therefore generically expected even without any HOTI dimerization. The abstract does not describe a control experiment or theoretical argument that separates this trivial geometric effect from a topology-protected effect. Without that control the 'multiple topological corner states' claim is not established. A standard control would be to demonstrate that the corner modes disappear under a continuous deformation of the intra/inter-cell coupling ratio that crosses the putative topological transition while keeping the lattice geometry (and
- [Topological classification of the anti-Hermitian generator] The kagome HOTI in the Hermitian setting relies on a generalized chiral / sublattice symmetry that pins corner modes to E=0 and on a quantized bulk invariant (bulk polarization, nested Wilson loop, or chiral winding). For the Laplacian L = D − W, the diagonal degree term D generally breaks the analogous chiral structure, and it is not automatic that a quantized invariant survives. The abstract does not state which invariant is computed, on which symmetry it relies, or how bulk-boundary correspondence is formulated for the anti-Hermitian generator. A clear statement and computation of the invariant for the diffusion Hamiltonian — and a demonstration that the observed corner modes track that invariant — is load-bearing for the central claim.
- [Decay-rate spectrum / gap] Identifying corner modes as in-gap requires a demonstrated gap in the decay-rate spectrum between the corner branch and the edge/bulk branches, in both theory and experiment. The abstract does not report such a gap or its size relative to experimental resolution. Please provide the full computed spectrum of decay rates for the experimental geometry, mark the corner-mode branch, and quantify the gap and the experimental uncertainty in extracted decay rates.
minor comments (4)
- [Abstract, sentence 1] 'Higher-dimensional topological meta-materials have more flexible than one-dimensional topological materials' is ungrammatical; please rephrase (e.g., 'are more flexible than').
- [Abstract] 'Purely imaginary eigenvalues corresponding to the decay rate' would be clearer if the sign convention and the relation between eigenvalue and measured decay rate were stated explicitly (e.g., T(t) ∝ exp(−γ t) with γ = |Im λ|).
- [Framing] The phrase 'first realization' should be qualified relative to prior 1D SSH-type and non-Hermitian skin-effect demonstrations in thermal/diffusive systems; a brief contrast with that literature in the introduction would help the reader place the contribution.
- [Terminology] 'Topological protected thermal metamaterial' — clarify what is being protected (the existence of the corner mode, its decay rate, or its spatial profile) and against which class of perturbations, given that the diffusion generator does not enjoy the energy-gap protection familiar from Hermitian systems.
Simulated Author's Rebuttal
We thank the referee for a careful and constructive report. The three major comments identify exactly the points where the present manuscript is underspecified, and we agree that all three must be addressed before the central claim — that the observed slow-decay corner localization is topological rather than a generic consequence of corner under-coordination on a kagome cut — can be considered established. In the revised manuscript we will (1) add a control experiment and accompanying simulations in which the intra/inter-cell coupling ratio is tuned across the topological transition while the lattice geometry and corner coordination are held fixed, isolating the topology-induced contribution from the trivial geometric one; (2) make the symmetry analysis and topological invariant explicit, computing the C3-rotation invariant / 2D Zak phase for the off-diagonal coupling matrix W and explaining how bulk–boundary correspondence carries over to the anti-Hermitian generator L = D − W when D is approximately uniform; and (3) report the full computed decay-rate spectrum for the experimental geometry, with the corner branch identified, the gap to the edge/bulk continuum quantified, and a propagated experimental uncertainty on the measured decay rates so that the in-gap nature of the corner modes can be assessed quantitatively. We believe these additions, all of which are in scope for the existing samples and apparatus, directly answer the referee's substantive concerns and considerabl
read point-by-point responses
-
Referee: Slow-decay corner localization is generically expected on a finite kagome cut because corner sites are under-coordinated (smaller D_ii); the manuscript does not separate this trivial geometric effect from a topology-protected effect. A control experiment varying the intra/inter-cell coupling ratio across the topological transition while keeping geometry fixed is needed.
Authors: We agree this is the load-bearing concern and we will address it directly in the revised manuscript. The point is well taken: for L = D − W on a finite kagome flake, slow-decay localization at three-coordinated corner vertices is indeed expected on purely geometric grounds, and the present manuscript does not explicitly rule it out. In revision we will (i) add a side-by-side comparison of two samples that share the same outer geometry and corner coordination but differ only in the intra-cell vs inter-cell thermal coupling ratio (t1/t2), so that one sample sits in the putative nontrivial regime (t1<t2) and the other in the trivial regime (t1>t2); (ii) report the measured corner-site decay rates for both, demonstrating that the slow-decay corner branch is present only in the nontrivial sample; and (iii) supplement this with numerical sweeps of t1/t2 through the gap-closing point, showing the corner branch detaches from the edge/bulk continuum only on the nontrivial side. We will also subtract, in theory, the diagonal degree term to display the spectrum of the off-diagonal coupling matrix alone, which isolates the topology-induced contribution from the coordination-induced shift. revision: yes
-
Referee: The topological invariant for the anti-Hermitian generator is not specified. The diagonal degree D generically breaks the chiral/sublattice symmetry that pins kagome HOTI corner modes to E=0, so it is not automatic that a quantized invariant survives. State which invariant, which symmetry, and how bulk-boundary correspondence is formulated.
Authors: We accept that the abstract is silent on this and will make the symmetry analysis and invariant explicit in the revised text. Concretely: (a) We work with H = iL where L = D − W. On a uniform kagome lattice all sites have the same coordination, so D is proportional to the identity within each sublattice and the off-diagonal hopping matrix W inherits the C3 and generalized chiral structure of the standard kagome HOTI model; the constant D shift moves the entire spectrum along the imaginary axis but does not lift the sublattice/C3-protected degeneracy among corner modes. (b) The relevant invariant is the bulk polarization (2D Zak phase) computed from the eigenmodes of W, equivalent to the C3-rotation invariant χ^(3) classification of Benalcazar et al. We will include the explicit calculation showing χ^(3) is nontrivial for our chosen t1/t2. (c) Bulk-boundary correspondence is formulated for the spectrum of L: in the nontrivial phase, three corner modes detach from the bulk/edge continuum and pile up near the smallest decay rate, set by D plus a t1/t2-dependent shift. We will add a section and supplementary calculation making these statements quantitative, and discuss the residual symmetry breaking when corner coordination differs from bulk (treated perturbatively). revision: yes
-
Referee: Identifying corner modes as in-gap requires a demonstrated gap in the decay-rate spectrum between corner and edge/bulk branches, in both theory and experiment. Provide the full computed spectrum, mark the corner branch, quantify the gap and the experimental uncertainty in extracted decay rates.
Authors: We agree and will add the requested quantitative material. The revision will include: (i) the full numerically computed eigenvalue spectrum of L for the experimental finite flake, with the corner-mode branch, edge branch, and bulk continuum color-coded, and the decay-rate gap Δγ between the corner branch and the edge band quoted explicitly (in units of s^-1); (ii) a comparison plot of the same spectrum in the trivial coupling regime, where the gap closes; (iii) an experimental error budget for the extracted decay rates, including uncertainty from the exponential-fit window, IR-camera noise floor, ambient drift, and finite contact thermal resistance, propagated to a 1σ uncertainty on each measured γ; and (iv) a statement of the signal-to-noise ratio Δγ/σ_γ, which determines whether the corner branch is experimentally resolvable as in-gap. We acknowledge that in the present manuscript these numbers are not given, and adding them is necessary to support the in-gap claim. revision: yes
- Whether the chiral/sublattice symmetry needed to pin the corner modes survives strictly at the boundary of the finite flake — where corner-site coordination differs from bulk — is only approximate; we treat the resulting symmetry breaking perturbatively but cannot claim an exactly quantized invariant for the open boundary geometry.
- Until the t1/t2 control sample is fabricated and measured, our experimental separation of topological from geometric corner localization rests on simulation; the referee is correct that the existing data alone do not decide the question.
Circularity Check
No significant circularity visible in the abstract; the open concern is a correctness/control issue (geometric vs. topological corner localization), not a self-referential derivation.
full rationale
Only the abstract is available, so a full derivation-chain audit cannot be performed. From what is stated, the paper claims (i) a kagome thermal metamaterial whose diffusion Hamiltonian is anti-Hermitian, (ii) theoretical analysis predicting corner states, (iii) experimental measurement of decay rates that match those modes. The empirical observable (temperature decay rate) is, in principle, an externally measurable quantity not defined in terms of the topological prediction, so the claim is not self-definitional in the sense this analyzer flags. There is no fitted parameter in the abstract that is then re-presented as a prediction, no load-bearing self-citation quoted, and no uniqueness theorem invoked. The skeptic's concern — that corner-localized slow-decay modes on a finite kagome flake are forced by reduced corner coordination of the graph Laplacian, independent of any dimerization-driven HOTI invariant — is a substantive correctness/control concern: it questions whether the observed signature actually distinguishes a topological from a trivial geometric effect. That is a bulk-boundary-correspondence and falsifiability concern, not circularity. It would only become circularity if the paper *defined* "topological corner state" to mean "slow-decaying corner mode" and then claimed to have observed topological corner states by observing slow-decaying corner modes. Without the body text, that reduction cannot be exhibited from quotes. Accordingly, score is 1: a placeholder reflecting that the abstract alone is insufficient to confirm circularity, and what is visible reads as standard prediction-then-measurement structure. The skeptic's attack should be routed to a correctness/controls review (does the paper compute a bulk invariant for the anti-Hermitian generator? does it show the corner modes vanish under a topology-closing deformation that preserves geometry?), not to circularity.
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.