Wannier-Center Approach

• Wannier-center approach is a framework for constructing maximally localized Wannier functions that reveal the real-space distribution of electronic and photonic bands.
• It employs techniques like Wilson loops and selective localization to extract bulk polarization, Chern numbers, and other topological invariants.
• The method has practical applications in identifying first- and higher-order topological phases and in mapping experimental boundary states in quantum materials.

The Wannier-center approach is a framework for analyzing the real-space structure and topological properties of electronic and photonic bands in crystalline and aperiodic solids, based on the explicit construction of Wannier functions and their associated spatial charge (or energy) centers. It plays a central role in modern theories of polarization, higher-order band topology, and localized orbital construction for correlated and condensed matter systems. This approach provides rigorous, gauge-invariant diagnostics for bulk polarization, topological invariants, and localized boundary or corner states by extracting physical information from the real-space centers (Wannier centers) of maximally localized Wannier functions (MLWFs) or from the spectral flow of their “hybrid” variants through the Brillouin zone.

1. Mathematical Foundations of Wannier-Center Formalism

Wannier functions for a group of $N$ Bloch bands are defined as

$$w_n(\mathbf{r}-\mathbf{R}) = \frac{V}{(2\pi)^d} \int_{\mathrm{BZ}} d^d k\,e^{-i\mathbf{k} \cdot \mathbf{R}} \sum_{m=1}^N U_{mn}(\mathbf{k})\,\psi_m(\mathbf{r},\mathbf{k})$$

where $U_{mn}(\mathbf{k})$ is a smooth, $N\times N$ unitary mixing, $\psi_m(\mathbf{r},\mathbf{k})$ are Bloch eigenstates, and $V$ is the unit-cell volume. This decomposition is highly non-unique due to gauge freedom, and the spatial localization is determined by minimizing the quadratic spread functional,

$$\Omega[U] = \sum_{n=1}^N \left[\langle w_n | \mathbf{r}^2 | w_n\rangle - |\langle w_n | \mathbf{r} | w_n\rangle|^2\right]$$

The gauge $U(\mathbf{k})$ that minimizes $\Omega[U]$ defines the maximally localized Wannier functions (MLWFs) (Liu et al., 2022, Li et al., 2023).

Wannier centers are then given by

$$\mathbf{r}_n = \langle w_n | \mathbf{r} | w_n\rangle$$

and play the role of “electric polarization centers” in the modern theory of polarization. In the multiband case, the gauge-invariant content is efficiently captured via “Wilson loops” (non-Abelian Berry phases): $$w_n(\mathbf{r}-\mathbf{R}) = \frac{V}{(2\pi)^d} \int_{\mathrm{BZ}} d^d k\,e^{-i\mathbf{k} \cdot \mathbf{R}} \sum_{m=1}^N U_{mn}(\mathbf{k})\,\psi_m(\mathbf{r},\mathbf{k})$$0 where $$w_n(\mathbf{r}-\mathbf{R}) = \frac{V}{(2\pi)^d} \int_{\mathrm{BZ}} d^d k\,e^{-i\mathbf{k} \cdot \mathbf{R}} \sum_{m=1}^N U_{mn}(\mathbf{k})\,\psi_m(\mathbf{r},\mathbf{k})$$1 is the non-Abelian Berry connection among the occupied bands, and the eigenphases yield the hybrid Wannier centers (WCCs) as functions of momentum (Osada, 10 Aug 2025, Taherinejad et al., 2013).

2. Physical Interpretation and Topological Invariants

In both electronic and photonic crystals, the sum of Wannier centers over all occupied bands within the unit cell yields the macroscopic polarization (modulo a quantum),

$$w_n(\mathbf{r}-\mathbf{R}) = \frac{V}{(2\pi)^d} \int_{\mathrm{BZ}} d^d k\,e^{-i\mathbf{k} \cdot \mathbf{R}} \sum_{m=1}^N U_{mn}(\mathbf{k})\,\psi_m(\mathbf{r},\mathbf{k})$$2

In 1D and higher dimensions, the Zak phase $$w_n(\mathbf{r}-\mathbf{R}) = \frac{V}{(2\pi)^d} \int_{\mathrm{BZ}} d^d k\,e^{-i\mathbf{k} \cdot \mathbf{R}} \sum_{m=1}^N U_{mn}(\mathbf{k})\,\psi_m(\mathbf{r},\mathbf{k})$$3 and its relation to the Wannier center $$w_n(\mathbf{r}-\mathbf{R}) = \frac{V}{(2\pi)^d} \int_{\mathrm{BZ}} d^d k\,e^{-i\mathbf{k} \cdot \mathbf{R}} \sum_{m=1}^N U_{mn}(\mathbf{k})\,\psi_m(\mathbf{r},\mathbf{k})$$4 is given by

$$w_n(\mathbf{r}-\mathbf{R}) = \frac{V}{(2\pi)^d} \int_{\mathrm{BZ}} d^d k\,e^{-i\mathbf{k} \cdot \mathbf{R}} \sum_{m=1}^N U_{mn}(\mathbf{k})\,\psi_m(\mathbf{r},\mathbf{k})$$5

The winding and partner-switching of hybrid Wannier centers (as functions of momentum) reveal the existence of bulk invariants: Chern numbers, $$w_n(\mathbf{r}-\mathbf{R}) = \frac{V}{(2\pi)^d} \int_{\mathrm{BZ}} d^d k\,e^{-i\mathbf{k} \cdot \mathbf{R}} \sum_{m=1}^N U_{mn}(\mathbf{k})\,\psi_m(\mathbf{r},\mathbf{k})$$6 invariants, weak indices, and crystalline/topological obstructions (Ligthart et al., 2024, Lahiri et al., 2023). For instance:

• Integer winding of WCCs indicates quantized charge pumping (Chern/nontrivial topology) (Huang et al., 2012, Taherinejad et al., 2013).
• Odd partner switching within the WCC “sheet” structure signals a nontrivial $$w_n(\mathbf{r}-\mathbf{R}) = \frac{V}{(2\pi)^d} \int_{\mathrm{BZ}} d^d k\,e^{-i\mathbf{k} \cdot \mathbf{R}} \sum_{m=1}^N U_{mn}(\mathbf{k})\,\psi_m(\mathbf{r},\mathbf{k})$$7 index (QSH phases).
• In higher-order insulators (e.g., second-order TIs), flat WCCs at fractional unit-cell positions signal boundary or corner-mode fractionalization (Liu et al., 2022, Liu et al., 2024).

3. Implementation: Computational and Algorithmic Aspects

Practical Wannier-center computation proceeds via:

• Gauge fixing by parallel transport, SCDM or selective localization techniques (Li et al., 2023, Wang et al., 2014, Oikawa et al., 22 Dec 2025).
• Construction of overlap matrices, $$w_n(\mathbf{r}-\mathbf{R}) = \frac{V}{(2\pi)^d} \int_{\mathrm{BZ}} d^d k\,e^{-i\mathbf{k} \cdot \mathbf{R}} \sum_{m=1}^N U_{mn}(\mathbf{k})\,\psi_m(\mathbf{r},\mathbf{k})$$8, along closed paths in BZ.
• Assembly and diagonalization of discretized Wilson loops: $$w_n(\mathbf{r}-\mathbf{R}) = \frac{V}{(2\pi)^d} \int_{\mathrm{BZ}} d^d k\,e^{-i\mathbf{k} \cdot \mathbf{R}} \sum_{m=1}^N U_{mn}(\mathbf{k})\,\psi_m(\mathbf{r},\mathbf{k})$$9.
• Extraction of eigenphases $U_{mn}(\mathbf{k})$0, mapped to Wannier centers via $U_{mn}(\mathbf{k})$1.

Recent variational approaches bypass center ambiguities by defining gauge-invariant density-convolution (DC) spread functionals (Li et al., 2023): $U_{mn}(\mathbf{k})$2 This enables systematically improvable, robust minimization, and is particularly effective for localized (even diffuse) orbitals and under periodic boundary conditions.

For large and/or disordered systems, regional and selective localization is achieved by maximizing a localized population functional (e.g., Pipek–Mezey with real-space weights) over arbitrary fragments, with scaling optimized via block–sequential or stochastic algorithms (Weng et al., 2022, Zhu et al., 2015).

4. Probing Topology: Wannier Centers in Band and Real Space

The essential value of the Wannier-center approach is its ability to distinguish between trivial, boundary-obstructed, and topological phases, even when edge or corner states are not directly accessible. In 1D, STM/STS spectroscopies integrating LDOS across the band identify experimental Wannier centers and their shift under topological transitions (e.g., in SSH, Rice–Mele, and trimer chains) (Ligthart et al., 2024).

For 2D and 3D topological insulators, plotting WCC sheets $U_{mn}(\mathbf{k})$3 across projected BZ planes reveals:

• Winding of WCCs: Chern/topological insulators (e.g. Bi$U_{mn}(\mathbf{k})$4Se$U_{mn}(\mathbf{k})$5).
• Spiral structures: Dirac/Weyl points (spiral acts as monopole/diabolical points) (Osada, 10 Aug 2025).
• Kramers' pair switching: diagnosis of $U_{mn}(\mathbf{k})$6 TIs.
• Partner crossing or “cross-linked” sheets: crystalline TI and higher-order TI phases (Liu et al., 2022, Liu et al., 2024, Taherinejad et al., 2013).

For second-order topological insulators, hybridization enforced by crystalline symmetries (e.g., glide) allows for corner states not attributable to trivial sublattice obstruction or coupled edge states, but emerging from composite MLWFs with non-additive polarization (Liu et al., 2022, Liu et al., 2024). The manifestation of type-III corner states is thus directly linked to nontrivial real-space localization of MLWFs when boundaries truncate the Wannier manifold.

5. Selective and Automated Construction, Generalizations, and Limitations

Modern workflows implement Wannier-center approaches in DFT and beyond-DFT contexts. Initial trial orbitals via automated hydrogenic projections or SCDM schemes, automated clustering of real-space Wannier density, and neural network fitting/inversion facilitate high-throughput construction for complex materials (Oikawa et al., 22 Dec 2025). Selective localization techniques allow for the targeted localization and center-fixing of orbitals relevant to correlated subspaces (e.g., $U_{mn}(\mathbf{k})$7-manifolds in transition-metal oxides) (Wang et al., 2014).

Recent advances address longstanding challenges:

• Gauge- and center-ambiguity under PBCs is removed via DC/TDC spreads (Li et al., 2023).
• Robust localization for fragments (“regionally localized orbitals”) in giant nanoscale systems is achieved with nearly linear scaling in orbital number (Weng et al., 2022).
• Algorithmic extensions to disordered systems proceed via phase-locking to target positions followed by block–bandwidth–limited spread minimization (Zhu et al., 2015).

Limitations remain in the presence of highly entangled bands, metallic regimes, or when orbital shapes deviate strongly from atomic/molecular prototypes (e.g., extremely diffuse “cage” orbitals or Rydberg-like states), but adaptive clusterings, higher-order radial ansätze, and gauge-retraining schemes are under active development (Oikawa et al., 22 Dec 2025).

6. Applications: Topological States, Higher-Order Topology, and Material Characterization

The Wannier-center formalism provides an efficient, bulk-only route to classify and diagnose:

• First-order topological insulators (Chern, quantum spin Hall, $U_{mn}(\mathbf{k})$8) via WCC flow, winding, partner exchange (Taherinejad et al., 2013, Lahiri et al., 2023).
• Weak/strong topology and boundary obstruction in three-dimensional organic conductors and topological crystalline insulators (Osada, 10 Aug 2025, Ligthart et al., 2024).
• Second-order and higher-order topological phases, including mode localization at corners (type-I, II, III), driven by the structure of MLWF hybridization and boundary truncation (Liu et al., 2022, Liu et al., 2024).
• Experimental mapping of boundary Wannier centers via local density-of-states integration in artificial and natural quantum materials (Ligthart et al., 2024).

The formalism is readily combined with DFT, tight-binding, and quantum chemistry codes and is directly compatible with polarization, Hubbard parameter extraction, and modeling of response functions and quantum transport phenomena.

---

Summary Table: Wannier-Center Approach — Core Elements and Applications

Element	Mathematical/Algorithmic Construct	Key Applications
MLWF Center	$U_{mn}(\mathbf{k})$9	Bulk polarization, boundary charge
Wilson Loop/Hybrid WCC	$N\times N$0	Topological invariants, phase transitions
Composite Wannier/Hybridization	Non-Abelian Berry, $N\times N$1 mixing	Corner modes (type-III, higher-order)
Selective/Regional Localization	Partial spread, block optimization	DMFT/DFT+U, impurity/defect states
Experimental Wannier Center Spectroscopy	LDOS integration (STM/STS), $N\times N$2	Boundary/topological phase identification

The Wannier-center approach thus forms the backbone of modern topological and real-space analysis of band structure, enabling both computational and experimental probes of geometric and topological quantum states. Its robust mathematical foundation and algorithmic tractability underpin a wide range of advanced material discoveries and topological phenomena (Liu et al., 2022, Liu et al., 2024, Li et al., 2023, Osada, 10 Aug 2025, Wang et al., 2014).