Wannier Model for Lattice Representations

• Wannier model is an effective lattice representation that converts delocalized Bloch states into spatially localized orbitals, bridging ab initio calculations and tight-binding Hamiltonians.
• Techniques like projection, maximal localization, and symmetry adaptation are employed to handle multi-band, topological, and disordered systems with precision.
• Derived Wannier functions enable the extraction of key physical quantities including Hamiltonian matrix elements, interaction parameters, and topological invariants.

A Wannier model is an effective lattice representation for the quantum states of a crystalline or quasi-periodic system, constructed by transforming delocalized Bloch states (in periodic systems) or suitably chosen energy eigenstates (in more general contexts) into a set of spatially localized orbitals known as Wannier functions. These functions are central for bridging first-principles calculations and low-energy, real-space theoretical descriptions, providing a basis for constructing tight-binding Hamiltonians, evaluating interaction parameters, and explicitly encoding symmetry and topology. Modern Wannier models extend to materials with multiple bands, topological order, disorder, multi-orbital character, and a wide range of condensed-matter systems.

1. Fundamental Definition and Construction

A set of $J$ Bloch eigenstates $\{\psi_{m\mathbf{k}}(\mathbf{r})\}$ of a periodic Hamiltonian $H$ in a crystal lattice provides the starting point. For a target manifold of bands, Wannier functions are constructed by a $(\mathbf{k}$-dependent) unitary rotation $U_{mn}(\mathbf{k})$,

$$W_{n\mathbf{R}}(\mathbf{r}) = \frac{V}{(2\pi)^d}\int_{\mathrm{BZ}} d\mathbf{k} \; e^{-i\mathbf{k}\cdot\mathbf{R}} \sum_{m=1}^J U_{mn}(\mathbf{k})\,\psi_{m\mathbf{k}}(\mathbf{r}),$$

where $\mathbf{R}$ is a Bravais lattice vector, $U(\mathbf{k}) \in \mathrm{U}(J)$ is chosen for maximal real-space localization, and $V$ is the volume of the primitive cell (Marzari et al., 2011).

Key steps in constructing a Wannier model:

1. Choice of Bloch Subspace: Identify the set of bands to be Wannierized (can be entangled).
2. Gauge Selection: Select trial localized orbitals, possibly symmetry-adapted, for initial projection and orthonormalization (Marzari et al., 2011, Gresch et al., 2018).
3. Gauge Optimization: Adjust the unitary $U(\mathbf{k})$ via localization functionals (typically the Marzari–Vanderbilt spread $\{\psi_{m\mathbf{k}}(\mathbf{r})\}$0) (Marzari et al., 2011, Zhang, 12 Feb 2025).
4. Real-Space Tight-Binding Mapping: Project the underlying Hamiltonian and relevant operators (e.g., valley, spin, symmetry) onto the Wannier basis.

In disordered systems, the construction generalizes via mixing energy eigenstates (band matrix) to minimize the spread and preserve real-space localization centers (Zhu et al., 2015).

2. Methodologies: Algorithms, Localization, and Symmetry

Wannier models employ a range of techniques, depending on the system and goals:

• Projection-Based Construction: Project trial orbitals onto the target band manifold and orthonormalize, forming a smooth gauge $\{\psi_{m\mathbf{k}}(\mathbf{r})\}$1 as long as $\{\psi_{m\mathbf{k}}(\mathbf{r})\}$2 everywhere.
• Maximal Localization (MLWFs): Use the spread functional,

$\{\psi_{m\mathbf{k}}(\mathbf{r})\}$3

and apply iterative gradient-based minimization (Marzari–Vanderbilt algorithm) to fix the gauge uniquely (Marzari et al., 2011).

• Handling Topology and Obstruction: For nontrivial topology, topological obstructions prevent the construction of exponentially localized Wannier functions in some subspaces (Soluyanov et al., 2010, Po et al., 2017). For $\{\psi_{m\mathbf{k}}(\mathbf{r})\}$4 topological insulators, the obstruction only exists if time-reversal symmetry is enforced globally.
• Symmetry Adaptation: Post hoc symmetrization via group-averaging can exactly enforce the crystal’s symmetry (unitary and antiunitary), ensuring model consistency under symmetry operations (Gresch et al., 2018).
• Disordered Systems: In non-periodic systems, construct Wannier functions by an initial “phase-setting” transformation to set the correct center, followed by unitary energy-mixing (band matrix) to achieve minimal localization (Zhu et al., 2015).
• Homotopy and Gauge Continuity: For multi-band and higher $\{\psi_{m\mathbf{k}}(\mathbf{r})\}$5 constructions, explicit algorithms on $\{\psi_{m\mathbf{k}}(\mathbf{r})\}$6-valued loops or grids, via parallel transport and homotopy contraction, ensure continuous, smooth gauges for systems with vanishing Chern number (Gontier et al., 2018).

3. Physical Content: Tight-Binding Hamiltonians and Interactions

Once a Wannier basis is established, all relevant physical quantities can be projected:

• Hamiltonian: Real-space matrix elements,

$\{\psi_{m\mathbf{k}}(\mathbf{r})\}$7

define a tight-binding model, capturing hoppings (up to controlled distance), exchange, and more.

• Interactions: Projected Coulomb, exchange, and density–density terms naturally arise, as in extended Hubbard models. Wannier basis also exposes correlated hopping, pair-hopping, and direct exchange terms when constructed from non-orthogonal (overlapping) orbitals (Milner et al., 2022).
• Operator Projections: Crystal symmetries, valley, spin, and other physical operators can be interpolated into the Wannier representation to directly model symmetry breaking, intervalley scattering, or other phenomena (Cao et al., 2020).
• Spectral and Geometric Reproduction: Wannier models can faithfully reproduce bandstructure, Berry curvature, polarization, and orbital magnetization (via Berry-phase and real-space Wannier-center expressions) (Marzari et al., 2011).

4. Topology, Obstructions, and Wannier Representability

Topological band structures fundamentally constrain Wannier construction. Main points:

• Chern Obstruction: Bands with nonzero Chern number lack exponentially localized Wannier representations (Marzari et al., 2011, Po et al., 2017, Zhang, 12 Feb 2025).
• $\{\psi_{m\mathbf{k}}(\mathbf{r})\}$8 Obstruction: In time-reversal-invariant topological insulators, a Wannier representation exists only if the gauge is allowed to break time-reversal symmetry; enforcing Kramers pairing globally obstructs localization (confirmed in the Kane–Mele model) (Soluyanov et al., 2010).
• Fragile Topology: Some band representations ("fragile topology") lack Wannier functions only if considered in isolation, but become representable when trivial additional bands are appended (Po et al., 2017).
• Wannier–Wilson Loop Diagnostics: Wilson-loop eigenphases (Berry/Wannier centers) provide a direct fingerprint for obstructions and topological invariants (Huang et al., 2012, Po et al., 2017).
• Phase-Space and Obstruction Bypass: Recent “phase-space” formalisms sidestep obstruction by embedding real space in momentum-dependent bundles, allowing for obstruction-free model construction by not gluing different $\{\psi_{m\mathbf{k}}(\mathbf{r})\}$9 fibers into single momentum manifolds (Sharma et al., 2024).

5. Advanced Applications and Generalizations

Wannier models are broadly applied in quantum materials modeling:

• Correlated Moiré Materials: Construction of symmetry-adapted, maximally-localized Wannier orbitals for magic-angle twisted bilayer graphene yields emergent honeycomb lattice models with three-centered "peak" orbitals and realistic extended Hubbard parameters (Kang et al., 2018, Koshino et al., 2018, Cao et al., 2020, Carr et al., 2019).
• Supercell and Multivalley Models: Large real-space supercells are employed to address small Fermi pockets in bilayer and rhombohedral stacking, providing mesoscopic scale models capturing both spectral weight and Berry curvature (Fischer et al., 2024).
• Disorder and Photonics: Construction is extended to disordered systems and photonic lattices, permitting the extraction of tight-binding descriptions and the analysis of topological defect states and localization (Zhu et al., 2015, Gupta et al., 2022).
• Non-Abelian and Universal Scaling: Universal one-dimensional scaling of Wannier functions and their moments is described by low-energy Dirac field theories, relating polarization and fluctuation theorems to characteristic length and gap phase (Piasotski et al., 2021).
• Ab Initio Integration: Automated workflows incorporate first-principles electronic structure with Wannier construction and symmetry post-processing, enabling parameter extraction for large material databases and externally-tuned lattices (Gresch et al., 2018).

6. Table: Key Features of Wannier Models

Dimension/System	Construction Strategy	Topological Obstruction
Periodic, Single Band	Parallel transport or potential-theory gauge	Chern number $H$0 blocks
Multi-band, Trivial	Projection + MLWF (spread minimization)	None (if all relevant Chern numbers vanish)
$H$1 Topological Insulator	Modified projection, $H$2-breaking gauge	Only for $H$3-pair gauges (Soluyanov et al., 2010)
Fragile Topology	Add trivial/atomic bands until obstruction lifts	Obstructed only in minimal subspace
Disordered systems	Phase-setting + band-matrix mixing (Zhu et al., 2015)	No obstruction; flexible localization
Moiré/Multi-valley TBG	Multi-band, symmetry-adapted Wannierization	Requires multivalley (not single-valley) models (Cao et al., 2020)

7. Physical Significance and Impact

Wannier models serve as the linchpin for translating microscopic, k-space-based physics into transparent, real-space descriptions:

• Interpolative Bandstructure: Fast, accurate interpolation of DFT bands, Berry curvature, and electron-phonon couplings (Marzari et al., 2011).
• Model Hamiltonians: Direct parameterization of tight-binding, Hubbard, and spin models.
• Materials Design: Essential for downfolding complex ab-initio results to tractable models, exploring phase diagrams, and enabling device modeling (Gresch et al., 2018, Fischer et al., 2024).
• Topological Field Theory: Provide a link between bulk invariants (Chern number, Zak phase, $H$4 index) and real-space physical observables (polarization, boundary charge) (Huang et al., 2012, Piasotski et al., 2021).
• Generalization: Extend beyond electrons to phonons, photons, and cold atom lattices via analogous procedures (Marzari et al., 2011, Gupta et al., 2022).

In summary, the Wannier model formalism enables controlled, symmetry-respecting, and, where possible, topology-aware passage from extended band theory to localized, physically interpretable, and computationally efficient lattice models. This framework is foundational in contemporary quantum materials research, many-body theory, and the topology–real-space interface.