Topological band insulators are classified using momentum-space topological invariants, such as Chern or winding numbers, when they feature translational symmetry. The lack of translation symmetry in disordered, quasicrystalline, or amorphous topological systems has motivated alternative, real-space definitions of topological invariants, including the local Chern marker and the spectral localizer invariant. However, the equivalence between these invariants is so far implicit. Here, we explicitly demonstrate their equivalence from a systematic perturbative expansion in powers of the spectral localizer’s parameter $\kappa$. By leveraging only the Clifford algebra of the spectral localizer, we prove that Chern and winding markers emerge as leading-order terms in the expansion. It bypasses abstract topological machinery, offering a simple approach accessible to a broader physics audience.

Topological band insulators display a bulk band gap populated by gapless boundary states?[1]. The number of gapless boundary states is a topological invariant, remaining unchanged as long as the bulk remains gapped. The bulk-boundary correspondence?[2, 3, 4, 5] directly connects the number of boundary modes with a topological index defined in the bulk[6].

It is common to use translational invariance to express topological indices as integrals over the Brillouin zone of a function of the momentum-space wave function and its derivatives. However, systems that lack translational symmetry, for example amorphous systems?[7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34], or quasicrystals?[35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51], can also display topological properties. To characterize topology in these systems, it is necessary to express topological indices as an integral over a local quantity, known as a local marker, in real space rather than in momentum space.

Within all the topological classes it is the class of two-dimensional systems displaying a quantized Hall effect?[52, 53] that can be identified with the largest variety of local topological markers. To calculate the real-space invariant for this class, the Chern number, we can choose to Fourier transform the momentum-space formula to arrive to the so-called local Chern marker formula?[54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67]. Alternatively, the spectral localizer index [68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79, 80, 81] and the Bott index [69], rely on quantifying how close the position and Hamiltonian operators are from commuting, and on defining a trivial system as that where these operators commute. An earlier invariant defined by Kitaev is similar to the local Chern maker but based on spatial tripartition [60]. Additionally, it is also possible to calculate the topological index using the rank difference of projectors [82]. Lastly, it is also possible to define a scattering invariant, which can be shown to be equivalent to counting edge modes?[83].

The existence of this variety of markers for two-dimensional Chern insulator systems also poses the challenge of demonstrating their equivalence. It is possible to show that the Kitaev invariant, the Bott Index, the scattering invariant, and the local Chern makers are all equivalent?[84, 85] ¹¹1For example, the Kitaev invariant is an integer exponentially quantized far away from edges?[60] which reduces to the local Chern marker with periodic boundary conditions?[84]. We thank Peru d’Ornellas for this comment.. However an explicit connection between the Localizer index and the rest is still obscure. Using algebraic topology, including K-theory [86, 71] and spectral flow analysis [70, 87], it is possible to show that the localizer index must be implicitly related to the local Chern marker. Ideally, it would be desirable to find an explicit derivation that relies on the simplest possible mathematics and connects to physical intuition. Finding such a derivation can be advantageous to propose other local markers for other topological classes. So far, only the spectral localizer and the scattering invariants apply all topological classes.

Moreover, explicitly connecting the local Chern maker to the localizer index can help to better understand the regime of validity of the latter. Specifically, the spectral localizer requires choosing the magnitude of a scalar parameter $\kappa$. This parameter determines the relative importance of the position and Hamiltonian operators within the spectral localizer. The criteria to fix $\kappa$ is not well understood. It is usually fixed with guidance from rigorous mathematical bounds?[88, 71], yet numerical simulations deliver correct results even when $\kappa$ violates these bounds?[89, 51, 90]. Numerical evidence also shows that the phase diagrams computed for a Chern insulator quasicrystal using both the localizer index and the local Chern marker can be made to coincide for small values of $\kappa$?[51]. This suggest that a perturbation theory on $\kappa$ could be used to show the equivalence between these two methods.

In this work, we reveal how the equivalence between the localizer index and other local markers arises naturally from a perturbative expansion of the spectral localizer in powers of $\kappa$. By exploiting only the fundamental symmetries of the Clifford matrices used by the spectral localizer, we demonstrate that the Chern and winding markers emerge as the leading-order contribution in this expansion. Our derivation applies to those $\mathds{Z}$ invariants in classes A and AIII of the Altland-Zirnbauer ten-fold classification of topological phases. Our derivation not only circumvents heavy topological machinery, but also establishes a direct, intuitive connection between the localizer index and the local Chern marker.

To map the spectral localizer index, Eqs.?(8), to the Chern or winding markers in Eqs.?(2) we use a Taylor perturbative expansion of $(\hat{L}^{2})^{-1/2}$ in powers of $\kappa$. We use the fact that, when $\kappa$ is small, the third term in both Eqs.?(6) is small compared to the gap imposed by the first two terms.

Without changing the spectral localizer, it is possible to replace $\hat{H}$ by a flattened Hamiltonian $\hat{H}_{F}=f(H)$. Here, $f$ is a smooth function that rescales the eigenstate energy $E$ above and below the gap to $f(E)=1$ and $f(E)=-1$, respectively, smoothly interpolating between these values for in-gap gapless modes. In particular $\hat{H}_{F}$ verifies $\hat{H}_{F}^{2}=1$ in the bulk. These properties will simplify our computations.

For clarity, we present the detailed calculation for even dimensions in the main text while the almost identical odd-dimensional case is deferred to Appendix B. In the even-dimensional case, the Taylor perturbative expansion of $(\hat{L}^{2})^{-1/2}$ in Eq.?(8) takes the form

where $c_{n}=\frac{(-1)^{n}(2n)!}{4^{n}(n!)^{2}}=\frac{(-1)^{n}\Gamma(n+1/2)}{n!\sqrt{\pi}}$ and $\hat{g}=(\hat{H}_{F}^{2}+\kappa^{2}\hat{r}^{2})^{-1}$ with $\hat{r}^{2}=\sum_{k}\hat{x}_{k}^{2}$.

When $r\gg 1/\kappa$, $\hat{g}$ becomes small, effectively restricting the real-space trace to a ball of radius $r\lesssim 1/\kappa$ of volume $\sim 1/\kappa^{d}$. So at most the trace involves a sum over $O(1/\kappa^{d})$ terms of order $\kappa^{n}$ giving a global scaling in $O(\kappa^{n-d})$. Therefore, in the $\kappa\to 0$ limit, the expansion can be truncated at $n=d$ since higher-order terms become negligible. While many terms remain for $n\leq d$, most vanish under the trace due to the Clifford algebra structure. Specifically, for any operator $A\otimes\prod_{i}\hat{\sigma}_{k_{i}}$, the trace is nonzero only if $\prod_{i}\hat{\sigma}_{k_{i}}$ is proportional to the identity. In such cases, the trace simplifies to $\Tr^{\prime}(A\otimes\mathds{1})=2^{\lfloor d/2\rfloor}\Tr(A)$.

In the perturbative expansion, each product contains an odd number of $\sigma$ matrices. For any term to be nonzero every $\sigma$ matrix must appear at least once in the product. This is so because the only way that the product is equal to the identity matrix is by virtue of the mathematical identity $\prod_{i=1}^{d+1}\hat{\sigma}_{i}=i^{\lfloor d/2\rfloor}$ . In particular, note that the identity, $\hat{\sigma}_{i}^{2}=1$ cannot change the parity of the number of Clifford matrices in the product. These observations imply that, after tracing over the Clifford degrees of freedom, only the following terms remain:

The term on the second line can be shown to vanish (see appendix A) so we focus on the first term. This term is proportional to the Chern marker as it can be rearranged into

where we have used that $\hat{H}_{F}^{2}=\mathds{1}$ and the fact that any commutator with $\hat{g}=(\hat{H}_{F}^{2}+\kappa^{2}\hat{r}^{2})^{-1}$ would create a higher order term in $\kappa$ that can be neglected. Moreover, we have used the fact that $(\hat{H}_{F}^{2}+\kappa^{2}\hat{r}^{2})^{-(d+1/2)}$ is small when $r\gg 1/\kappa$ making the integral convergent. Therefore, the largest contribution to the trace comes from the bulk, per our arguments below Eq.?(9). This in turn justifies using $\hat{H}_{F}^{2}=\mathds{1}$, which is only true if we neglect boundary contributions. We can rewrite Eq.?(11) in terms of the weight function

which adds the overall prefactor

Combining such coefficient with the rest of the prefactors, one can check that the proportionality constant is one and we reach the desired equality between Eq.?(11) and the averaged Chern marker Eq.?(2a)

Following similar steps for the odd dimensional case (see Appendix B) we obtain

Eqs.?(14) and (15) are the main result of this work.

We have demonstrated a direct equivalence between the spectral localizer index $I_{\text{SL}}$ and the Chern and winding markers of topological classes A and AIII through a systematic perturbative expansion in the small-parameter regime, $\kappa\ll\frac{\Delta^{2}}{\sum_{k}\|[\hat{H},\hat{x}_{k}]\|}$, where $\Delta$ is the bulk gap of the Hamiltonian. By leveraging the Clifford algebra symmetries inherent to the spectral localizer’s construction, we showed that for even dimensions, $I_{\text{SL}}$ reduces to the Chern marker $C_{d/2}$ as the leading-order contribution in $\kappa$, while in odd dimensions it analogously reduces the winding marker $W_{\lceil d/2\rceil}$.

Our direct approach circumvents the need for abstract algebraic topology (e.g., K-theory or spectral flow), instead relying on transparent asymptotic analysis of the spectral localizer’s resolvent. Our result provides an explicit proof of the equivalence. Additionally, this equivalence explains the numerically observed consistency between the topological phase diagrams calculated using the local Chern marker with those calculated with the spectral localizer index at small $\kappa$?[51].

Here we focused on $\mathbb{Z}$-classified topological phases via Chern and winding markers. A natural open question is to adapt the methodology to $\mathbb{Z}_{2}$ topological insulators (e.g., time-reversal symmetric or particle-hole symmetric systems), where a spectral localizer index can be defined [78] but local topological markers are still lacking.

We thank A. Akhmerov, I. Araya Day, P. Delplace, P. d’Ornellas, C. Fulga, and T. Loring for insightful discussions and related collaborations.