---
title: Quantum Topology Witness Overview
url: https://www.emergentmind.com/topics/quantum-topology-witness
type: topic
---

# Quantum Topology Witness Overview

Searching arXiv for recent papers explicitly using or closely related to "quantum topology witness".
arxiv_search(query="\"quantum topology witness\" OR topology witness quantum topological witness", max_results=10, sort_by="submittedDate")

arxiv_search(query="quantum topology witness", max_results=10, sort_by="submittedDate")

A quantum topology witness is an operational diagnostic whose value, scaling, or spatial structure distinguishes topological from trivial organization in a quantum system without necessarily relying on conventional momentum-space invariants. Across the literature, the term refers to several non-equivalent constructions: information-theoretic witnesses of abelian anyonic order, projector-based local and nonlocal quantum geometric markers for disordered or amorphous band structures, local Dirac-node witnesses based on a non-Hermitian Zeeman quantum geometric tensor, dynamical witnesses built from out-of-time-order correlations or from strange correlators re-expressed through Kirkwood–Dirac quasiprobabilities, and photonic interferometric or high-dimensional spectral witnesses that extract topology from spatial correlations and phase singularities [1112.2527], [2511.09664], [2604.09725], [2302.00927], [2606.19730], [2503.12540], [2606.11002]. In this sense, the subject is not a single invariant but a family of experimentally oriented criteria tailored to distinct notions of topology.

## 1. Information-theoretic witnesses of abelian topological order

The earliest explicit witness in this corpus is the anyonic topological entropy, introduced for abelian anyonic systems and stable quantum memories. It is defined in the anyon-occupation basis on an annular partition with an inner disk \(A\), a surrounding annulus \(B\), and a coarse-grained variable \(\tilde{A}\) encoding the net anyon content of \(A\). With \(S_B=-\mathrm{tr}(\rho_B\log\rho_B)\) and \(S_{\tilde{A}B}=-\mathrm{tr}(\rho_{\tilde{A}B}\log\rho_{\tilde{A}B})\), the construction is
\[
\Gamma' = S_{\tilde{A}B}-S_B,\qquad
\Gamma = c_D + S_B - S_{\tilde{A}B},
\]
with \(c_D=2\log D\) for a general abelian anyon model and, for the planar or toric code,
\[
\Gamma = 2\log 2 + S_B - S_{\tilde{A}B}.
\]
Because the formalism is written in the anyon-occupation basis, it can be computed from classical probability distributions over anyon variables rather than from full quantum-state tomography [1112.2527].

Its diagnostic content is tied directly to error correction. In the thermodynamic limit, if the anyon occupations in \(B\) suffice to determine the net anyon content inside \(A\), then thickened loop operators can be defined on the annulus and the system is in the topologically ordered or error-correcting phase. In that case \(\Gamma\) attains its maximal value,
\[
\Gamma = 2\log D.
\]
If the net anyon content of \(A\) is random and uncorrelated with \(B\), then
\[
\Gamma = 0.
\]
An intermediate value,
\[
\Gamma = \log D,
\]
is possible when one sector is determined and the other is random, but the paper stresses that this value is ambiguous: it can indicate partial loop order or a state that is not truly topologically ordered despite exhibiting sector-resolved order [1112.2527].

The witness is closely connected to decoding. In the planar code with independent bit and phase errors, one may estimate \(\Gamma\) exactly from empirical distributions \(p(\tilde{A},B)\), or lower-bound it by running a decoder on \(B\) and using the resulting success probabilities \(P_s\) and \(P_p\). The same framework reproduces known finite-temperature instability in the 2D toric or planar code: for thermal states with independent plaquettes, \(I(\tilde{A}:B)=0\), and for any fixed \(\beta>0\) one has \(\Gamma\to 0\) as system size increases [1112.2527].

Conceptually, this witness differs from band-topological markers. It does not diagnose Berry curvature, Chern number, or edge-state localization directly; instead it measures whether topological information is encoded nonlocally and remains decodable from an annular region. Within abelian quantum double models \(D(G)\), it is therefore both a topology witness and a witness of memory stability.

## 2. Real-space quantum geometric markers and nonlocal topology witnesses

A distinct line of work treats topology through occupied-state projectors and position operators, entirely in real space and without assuming long-range translational symmetry. For a gapped single-particle Hamiltonian \(H\), with projector onto occupied states
\[
P=\sum_{n\in\mathrm{occ}} |\psi_n\rangle\langle\psi_n|,\qquad Q=1-P,
\]
the local Chern marker is taken as
\[
C(r)=-4\pi\,\mathrm{Im}\,\langle r|P\hat{x}Q\hat{y}P|r\rangle,
\]
and the local quantum metric marker is
\[
\Omega(r)=4\pi\langle r|P\hat{x}Q\hat{x}P + P\hat{y}Q\hat{y}P|r\rangle.
\]
Their spatial averages reproduce the Chern invariant and the integrated quantum metric. The same projector/position-operator structure extends naturally to arbitrary geometries, disorder, and amorphous matter [2511.09664].

The defining step for the witness is the move from local to nonlocal kernels. The nonlocal Chern marker and nonlocal quantum metric are
\[
C(r,r')=-i\frac{2\pi}{A}\sum_\alpha \langle r,\alpha|P\hat{x}Q\hat{y}P-P\hat{y}Q\hat{x}P|r',\alpha\rangle,
\]
\[
\Omega(r,r')=\frac{4\pi}{A}\sum_\alpha \langle r,\alpha|P\hat{x}Q\hat{x}P+P\hat{y}Q\hat{y}P|r',\alpha\rangle.
\]
These quantities act like correlation functions. The kernel
\[
K_{\mu\nu}(r,r')=\langle r|P\hat{r}_\mu Q\hat{r}_\nu P|r'\rangle
\]
measures the amplitude for an occupied state to virtually hop to an unoccupied sector via the position operator and return to the occupied sector, with resolved spatial endpoints. In gapped phases the kernels decay exponentially or algebraically with \(|r-r'|\), depending on localization properties; at topological criticality the correlation length \(\xi\) diverges and the magnitude of the kernels sharply increases [2511.09664].

This nonlocality yields a refined witness of phase transitions. For a chosen modulation vector \(k\), one defines integrated indicators
\[
C_k=\iint dr_1\,dr_2\, C(r_1,r_2)e^{ik\cdot(r_1-r_2)},\qquad
\Omega_k=\iint dr_1\,dr_2\, \Omega(r_1,r_2)e^{ik\cdot(r_1-r_2)}.
\]
Summing over \(k\) retrieves the coarse bulk information, while the \(k\)-resolved quantities retain the fine structure distinguishing different transitions. In the two-band Chern insulator
\[
H(k)=[M-2t(\cos k_x+\cos k_y-2)]\sigma_z + 2t\sin k_x\,\sigma_x + 2t\sin k_y\,\sigma_y,
\]
three transitions occur at \(M/t\in\{0,-4,-8\}\), with gap closings at \(k=(0,0)\), \(k=(0,\pi)/(\pi,0)\), and \(k=(\pi,\pi)\), respectively. Near these transitions, the nonlocal markers display distinct real-space modulations: no-oscillation radial decay at \(k=0\), cross-square modulation for edge-of-Brillouin-zone closings, and checkerboard modulation at \((\pi,\pi)\). The integrated markers \(C_k\) and \(\Omega_k\) peak selectively at the corresponding transition and continue to track it under disorder \(\delta M\) without mixing between indicators [2511.09664].

A notable consequence is that the witness resolves transitions within a single Altland–Zirnbauer class. In class A, the three Chern transitions are not merely “topology changes”; they have different spatial fingerprints. In fully amorphous lattices with similar local coordination, the same witness also separates a \(C=-1\) phase that persists from a \(C=+1\) phase that collapses into a broad metallic region. The explanation given is pattern compatibility with local geometry: the \(k=0\) transition produces uniform kernels compatible with any local polygon, while edge-of-Brillouin-zone patterns require well-defined second-neighbor structure and are degraded by triangular motifs [2511.09664].

## 3. Local topology from dual quantum geometric tensors

A different witness targets local nodal topology rather than global band topology. The starting point is the contrast between the conventional Hermitian quantum geometric tensor and the Zeeman quantum geometric tensor. For a single isolated Bloch band, the conventional tensor
\[
Q_{ij}(k)=\langle \partial_{k_i}u|(1-|u\rangle\langle u|)|\partial_{k_j}u\rangle
=g_{ij}(k)+ i\,\Omega_{ij}(k)/2
\]
contains the real symmetric quantum metric and the imaginary antisymmetric Berry curvature. The Zeeman construction instead uses mixed interband position and spin matrix elements,
\[
T^{Z,ab}_{nm}=r^a_{nm}\sigma^b_{mn},
\]
and is generically non-Hermitian because \(T^{Z,ab}_{nm}\neq (T^{Z,ba}_{nm})^*\) [2604.09725].

Its decomposition yields four symmetry-resolved sectors. The normal sector contains a real symmetric metric-like tensor \(g^N\) and an imaginary antisymmetric curvature-like tensor \(\Omega^N\). The anomalous sector contains an imaginary symmetric metric-like tensor \(g^A\) and a real antisymmetric curvature-like tensor \(\Omega^A\). In the paper’s notation,
\[
g^{N/A}_{nm}=\frac{1}{2}\,\mathrm{Re/Im}[\,r_{nm}\otimes \sigma_{mn}+\sigma_{mn}\otimes r_{nm}\,],
\]
\[
\Omega^{A/N}_{nm}=-\,\mathrm{Re/Im}(r_{nm}\times \sigma_{mn}).
\]
Replacing \(\sigma\to r\) collapses the anomalous sector and recovers the conventional Hermitian quantum geometry [2604.09725].

For the two-dimensional Dirac model \(H=d(k)\cdot \sigma\) with \(d=(k_x,k_y,m)\), the anomalous Zeeman curvature becomes
\[
\Omega^A(k)=-\frac{1}{2}\frac{(k_x,k_y,2m)}{d^2},
\qquad d^2=k_x^2+k_y^2+m^2,
\]
with divergence
\[
\nabla_k\cdot \Omega^A = -\frac{m^2}{d^4}.
\]
In the massless limit,
\[
\nabla_k\cdot\Omega^A = -\frac{1}{2}\nabla_k\cdot\left(\frac{k}{k^2}\right)= C\,\delta^{(2)}(k),
\qquad C=-\pi.
\]
This gives a radial flux singularity at the Dirac node. The same work proves the Hodge-dual relation
\[
\Omega^A = \frac{1}{2}*w = \frac{1}{2}\,\hat z\times w,
\]
where \(w=\nabla_k\theta_k\) is the winding field. Consequently the local \(\pi_1\) topology of the node can be represented equally as winding or as anomalous-curvature flux,
\[
\oint_\Gamma \Omega^A\cdot \hat n\,dl = -\pi\,w.
\]
The witness is therefore local and gauge-invariant: it identifies isolated Dirac singularities without integrating over the full Brillouin zone [2604.09725].

The same tensor enters measurable transport. The four components of the gyrotropic conductivity and the reciprocal kinetic magnetoelectric response map one-to-one onto the four Zeeman-geometric sectors. In particular,
\[
\mathrm{Im}\,\sigma^- (\omega)\sim \omega\langle \Omega^A\rangle,\qquad
\mathrm{Im}\,\sigma^+ (\omega)\sim \omega\langle g^N\rangle,
\]
\[
\mathrm{Re}\,\sigma^- (\omega)\sim \omega^2\langle \Omega^N\rangle,\qquad
\mathrm{Re}\,\sigma^+ (\omega)\sim \omega^2\langle g^A\rangle,
\]
with complementary selection rules for the reciprocal KME tensor \(\alpha^\pm\). This turns the local geometric witness into an experimentally filtered transport diagnostic, with tensor symmetry and low-frequency scaling isolating the relevant geometric sector [2604.09725].

## 4. Dynamical witnesses: OTOCs, strange correlators, and KDQs

A dynamical witness of topological phase transitions can be built from a fidelity-type out-of-time-order correlation. With Heisenberg evolution \(W(t)=e^{iHt}We^{-iHt}\) and a projector \(V=|\psi_0\rangle\langle\psi_0|\) onto the initial pure state, the witness is
\[
F(t)=\langle \psi_0|W(t)VW(t)V|\psi_0\rangle.
\]
The central result is a zero-to-finite-value transition in the long-time limit: \(F(t\to\infty)\approx 0\) in trivial phases and remains finite in topological phases. The mechanism is “topological locality”: localized edge or corner states suppress complete scrambling, whereas bulk modes in trivial phases allow the echo signal to decay to zero. The paper demonstrates this in the SSH model, Creutz ladder, Haldane model, and the \(\pi\)-flux 2D SSH model, and reports robustness to choices of the initial state, to diverse local or multi-site operators \(W\), and to weak disorder in real space [2302.00927].

The same article gives analytic forms in the clean SSH chain. For example, with \(|\psi_0\rangle=|1,A\rangle\), one obtains explicit large-\(N\) expressions for \(F(t)\) involving \(\lambda_\pm^{(k)}=\pm \epsilon[1+\nu^2+2\nu\cos(k\pi/(N+1))]^{1/2}\). In the topological phase \(\nu<1\), the constant term survives after time averaging and produces finite \(F(\infty)\); in the trivial phase \(\nu>1\), oscillatory contributions dominate and \(F(\infty)\to 0\). The Haldane-model calculations show the same behavior across the Chern transition at \(|\mu|=3\sqrt{3}\eta_2\) for \(\phi=\pi/2\), even though that model lacks chiral symmetry [2302.00927].

A second dynamical witness uses strange correlators between two many-body states of the same system, one with known trivial topology and one under test. For pure states \(|\Psi\rangle\) and \(|\Omega\rangle\), and local or bilinear operator \(\hat o(r)\),
\[
s[\hat o]_{r,r'}=\frac{\langle \Omega|\hat o(r)\hat o(r')|\Psi\rangle}{\langle \Omega|\Psi\rangle}.
\]
In the BHZ example, the momentum-space version is written with
\[
O_k=\frac{1}{2}\big(\eta_{+,k}^\dagger\eta_{-,k}+\eta_{-,k}^\dagger\eta_{+,k}\big),\qquad
|\Upsilon_k\rangle=\eta_{+,k}^\dagger\eta_{-,k}|\Psi\rangle,
\]
so that
\[
s[\hat o_k]=\frac{\langle \Omega|\Upsilon_k\rangle}{\langle \Omega|\Psi\rangle}.
\]
If the two states have different topology classes, then near \(k\to 0\)
\[
|s[\hat o_k]|\sim |k|^{-\alpha},\qquad \alpha=2\eta_E\ge 0;
\]
if they have the same topology,
\[
|s[\hat o_k]|\sim |k|^\beta,\qquad \beta\ge 0.
\]
In real space this translates into algebraic decay for different topology classes and exponential decay otherwise [2606.11002].

The same paper re-expresses the strange correlator in terms of Kirkwood–Dirac quasiprobabilities. With \(\rho=|\Psi\rangle\langle\Psi|\), \(\Xi=|\Omega\rangle\langle\Omega|\), eigenvalues \(\lambda_1=+1\), \(\lambda_2=-1\), and spectral projectors \(\Lambda_{k,j}\) of \(O_k\), the relevant KDQs are
\[
q_{k,j}=\mathrm{Tr}(\rho\,\Xi\,\Lambda_{k,j}),
\]
and with \(p_\Xi=|\langle \Omega|\Psi\rangle|^2\) the witness becomes
\[
W_{\mathrm{top}}(k)\equiv s[\hat o_k]
=\frac{1}{p_\Xi}\sum_j \lambda_j q_{k,j}.
\]
Equivalently,
\[
\mathrm{Re}\,s[\hat o_k]=\frac{1}{2p_\Xi}\mathrm{Tr}(\rho\{\Xi,O_k\}),\qquad
\mathrm{Im}\,s[\hat o_k]=-\frac{i}{2p_\Xi}\mathrm{Tr}(\rho[\Xi,O_k]).
\]
This makes the witness simultaneously a strange correlator, a weak value,
\[
s[\hat o_k]=\frac{\langle \Omega|O_k|\Psi\rangle}{\langle \Omega|\Psi\rangle},
\]
and a sequential-measurement quantity reconstructable interferometrically from the characteristic function
\[
G_k(u)=\mathrm{Tr}(\rho\,e^{-iuO_\Xi}e^{iuO_k}).
\]
The ancilla-assisted protocol measures \(\mathrm{Re}\,G_k(u)\) and \(\mathrm{Im}\,G_k(u)\) via \(\langle \sigma_x^A\rangle\) and \(\langle \sigma_y^A\rangle\), then reconstructs the KDQ distribution by inverse Fourier transform [2606.11002].

## 5. Interferometric and high-dimensional photonic witnesses

Structured-light platforms support another family of quantum topology witnesses in which topology is read from two-photon interference or from a many-invariant “topological spectrum.” In topological quantum interferometry, the central object is the exchange Berry phase,
\[
\Phi_{\mathrm{BPX}}(\theta_c,\theta_d)=2(q_a\theta_c\pm q_b\theta_d),
\]
defined on the joint azimuthal coordinate plane of the output ports \(c\) and \(d\). Its gradient plays the role of a Berry connection,
\[
\mathbf{A}_{\mathrm{BPX}}=\nabla \Phi_{\mathrm{BPX}}=(2q_a,\pm 2q_b),
\]
and the witness is the winding number
\[
W_{\mathrm{topo}}
=\frac{1}{2\pi}\oint_{\partial S}\nabla \Phi_{\mathrm{BPX}}\cdot d\boldsymbol{\ell}
=\sum_j \chi_j,
\]
obtained by contour integration or by counting phase singularities with charge \(\chi_j\) in the coincidence manifold. Because the Bell-channel amplitudes carry the phase factor \(e^{-i\Phi_{\mathrm{BPX}}}\), the interference topology is controlled deterministically by the q-plate charges \(q_a,q_b\) and detunings \(\delta_a,\delta_b\). The same work states the dimensionality estimate
\[
d\gtrsim |W_{\mathrm{topo}}|,
\]
so the witness functions as a non-tomographic lower bound on the Hilbert-space dimension of the joint state [2606.19730].

The q-plate transformation is written as
\[
\hat U_{\text{q-plate}}=\hat R(-\gamma)\hat D(\delta)\hat R(\gamma),\qquad \gamma=2q\theta,
\]
with the corresponding partial spin–orbit conversion in the circular/OAM basis. After the 50:50 beamsplitter, the Bell-channel amplitudes depend on \(C_{a,b}=\cos(\delta_{a,b}/2)\), \(S_{a,b}=\sin(\delta_{a,b}/2)\), and on the BPX phase. For identical detunings, the HOM visibility in the \(\nu\)-channel is
\[
V(\delta)=\frac{\sin^2(\delta/2)}{\cos^2(\delta/2)+1}.
\]
A key claim of the work is that the spatial probability and visibility maps can be decomposed into a small set of fundamental azimuthal modes,
\[
\mathcal{M}(\theta_c,\theta_d)=\sum_{i=1}^{5}Q_i(q_a,q_b)\,A_i(\delta_a,\delta_b)\,\mathcal{O}_i(\theta_c,\theta_d),
\]
so the witness is both topological and geometrically interpretable [2606.19730].

A more algebraic photonic witness is the topological spectrum of high-dimensional quantum states. For a qudit of dimension \(d\), the density matrix is expanded in the \(su(d)\) generators,
\[
\rho(\vec r)=\frac{1}{d}\mathbf{1}_d+\sum_{a=1}^{d^2-1}m_a(\vec r)\,T_a,\qquad
m_a(\vec r)=\mathrm{Tr}(T_a\rho(\vec r)),
\]
and one studies the family of maps defined by triples \((m_i,m_j,m_k)\). The total number of candidate triples is
\[
\binom{d^2-1}{3},
\]
while the number of independent non-trivial invariants for the state class considered is
\[
N_{\rm ind}=\frac{1}{4}d(d-1)(d-2)(d+3)\qquad (d>2).
\]
The resulting viewpoint replaces a single topological number by a spectrum of integer invariants supported by many embedded maps \(S^2\to S^2\) and \(D^2\to D^2\) [2503.12540].

For the usual sphere maps, the index is
\[
N=\frac{1}{4\pi}\int \vec S\cdot\left(\frac{\partial \vec S}{\partial r}\times \frac{\partial \vec S}{\partial \phi}\right)\,dr\,d\phi\in \mathbb Z.
\]
For exotic disk maps glued into sphere maps, a representative qutrit invariant is
\[
\hat N_{124}=(l_0-l_1)\,\mathrm{sgn}(|l_2|-|l_1|).
\]
The same work reports a 48-dimensional topology with a topological spectrum spanning over \(17000\) maps, specifically \(17296\) candidate maps for \(d=7\), and describes emergent signatures in previously trivial sectors as a witness of perturbation-induced modal mixing. The decision rule combines confidence intervals on the individual invariants with spectral comparisons using cosine similarity,
\[
\mathrm{CosSim}=\frac{\sum(A_\alpha E_\alpha)}{\|\vec A\|_2\|\vec E\|_2},
\]
with \(\mathrm{CosSim}>0.9\) used in the qutrit experiments to certify the same topological class [2503.12540].

## 6. Experimental implementations, scope, and open problems

The witness concept is explicitly experimental across all these frameworks, but the required observables differ sharply. Projector-based nonlocal markers require diagonalization of a finite tight-binding Hamiltonian, construction of \(P\) and \(Q\), evaluation of matrix elements such as \(P\hat xQ\hat yP\), and either direct visualization of \(C(r,r_0)\), \(\Omega(r,r_0)\) or integration into \(C_k\) and \(\Omega_k\). Cold atoms, photonic and phononic metamaterials, and electronic platforms with scanning probes or nonlinear geometric responses are proposed as suitable arenas. Practical constraints include finite-size and edge effects, the need for bulk averaging, and the requirement that a mobility gap exist even if a spectral gap does not [2511.09664].

The Zeeman-QGT witness is formulated for transport and magnetoelectric measurements rather than for state reconstruction. Its experimental logic is symmetry filtering and low-frequency scaling: one isolates \(\Omega^A\), \(g^N\), \(\Omega^N\), or \(g^A\) by selecting the appropriate gyrotropic or reciprocal KME channel and reading its \(O(\omega)\) or \(O(\omega^2)\) behavior. The proposed materials include graphene-based heterostructures, 2D Dirac semimetals, transition-metal chalcogenides, topological-insulator surface states, and moiré Dirac materials with tunable inversion breaking and Zeeman coupling [2604.09725].

For dynamical witnesses, the resource bottleneck is coherence time rather than spatial resolution. OTOC protocols require forward evolution, application of a local or few-site operator \(W\), backward evolution, and measurement of the return fidelity \(F(t)\). The KDQ-based witness requires ancilla-assisted interferometry, controlled exponentials \(e^{-iuO_\Xi}\) and \(e^{iuO_k}\), and adequate post-selection overlap \(p_\Xi=|\langle \Omega|\Psi\rangle|^2\). The same paper emphasizes that small \(p_\Xi\) amplifies normalization variance, while decoherence and SPAM errors bias the characteristic function \(G_k(u)\). Superconducting circuits, trapped ions, NV centers, and Floquet-engineered realizations are identified as plausible platforms [2302.00927], [2606.11002].

Photonic witnesses trade many-body control for spatially resolved detection. BPX-based interferometry needs accurate control of q-plate charges and detunings, angular calibration on the \((\theta_c,\theta_d)\) plane, and sufficient signal-to-noise ratio to locate phase singularities reliably. The topological-spectrum approach, by contrast, requires high-dimensional quantum-state tomography. Its measurement count scales as
\[
N_{\rm meas}=\big(4\binom{d}{2}+d\big)^2,
\]
and the number of candidate maps grows combinatorially. The same work therefore relies on \(su(2)\) “nice pairs,” threshold-based denoising, glue constructions for exotic maps, and parallelizable numerical integration [2606.19730], [2503.12540].

Several limitations recur. In metallic phases without a mobility gap, projector-based markers can lose quantization; the single-particle projector framework does not directly cover interacting topological order; the anyonic topological entropy is defined and justified for abelian anyons and depends on the chosen anyon basis; direct generalization to non-abelian anyons is not straightforward; the OTOC witness is established in noninteracting band-topological models; the KDQ witness depends on identifying an operator whose bulk form becomes an edge operator; and the topological-spectrum program is derived analytically for a specific family of OAM states \( |\psi(\vec r)\rangle=\sum_i f_i(r)e^{i\ell_i\phi}|i\rangle \) [1112.2527], [2511.09664], [2302.00927], [2606.11002], [2503.12540].

Taken together, these developments define a modern meaning of quantum topology witness: not a universal invariant, but a family of operational criteria tailored to different topological objects. Depending on context, a witness may certify error-correcting loop order, reproduce a bulk invariant from real-space projectors, identify the momentum character of a transition through nonlocal spatial modulations, convert local nodal winding into measurable curvature flux, distinguish topology classes through anomalous dynamical correlators, or infer high-dimensional topological structure from interference singularities and spectral deconstruction. This diversity is precisely what makes the notion useful: topology is probed through the observable structure most natural to the system under study.

Source: https://www.emergentmind.com/topics/quantum-topology-witness