wannier90-tight-binding skill

Overview

This skill provides hands-on guidance for generating maximally localized Wannier functions and building tight-binding models with Wannier90. It helps translate DFT outputs into compact Wannier representations, enabling fast band interpolation, topology analysis, and transport modeling. The guidance focuses on reproducible workflows and practical settings for common DFT codes.

How this skill works

The skill inspects DFT outputs (projections, bandstructures, and k-point data) and constructs Wannier90 input files tuned for disentanglement and localization. It walks through setting frozen/disentanglement windows, selecting initial projections, and running iterative minimization to produce MLWFs. After wannierization, it shows how to export tight-binding Hamiltonians, compute interpolated bands, Berry phases, and topological invariants, and connect results to transport and WannierTools workflows.

When to use it

• Converting DFT wavefunctions into compact tight-binding models for large-scale or high-throughput studies.
• Interpolating band structures at high resolution without rerunning costly DFT calculations.
• Computing Berry phases, Chern numbers, Z2 invariants, or other topological diagnostics.
• Preparing Hamiltonians for transport simulations or coupling to quantum transport solvers.
• Refining projection choices and disentanglement windows when initial wannierization fails to converge.

Best practices

• Validate MLWFs by comparing interpolated bands to original DFT bands across the energy window of interest.
• Choose chemically motivated initial projections and test several sets to improve localization and symmetry preservation.
• Set clear frozen and disentanglement windows to avoid spurious band mixing and ensure physical subspace selection.
• Use dense k-grids for final interpolation and sparser grids during initial testing to save time.
• Document convergence parameters (num_iter, convergence tolerance, spread thresholds) and retain input/output files for reproducibility.

Example use cases

• Generate a tight-binding Hamiltonian for a 2D semiconductor to enable large-scale defect or heterostructure modeling.
• Interpolate high-resolution band structures and Fermi surfaces for comparison with ARPES experiments.
• Compute Berry curvature and Chern numbers for candidate topological materials using WannierTools or Z2Pack.
• Export MLWF-based Hamiltonians to a transport solver to model conductance through a nanoscale junction.
• Automate a pipeline from VASP/Quantum ESPRESSO outputs to Wannier90 inputs for high-throughput screening.

FAQ