pyatb

Official Resources

• Repository: https://github.com/pyatb/pyatb
• Documentation: https://pyatb.github.io/ (or repo specific)
• License: GNU General Public License v3.0
• Organization: DeepModeling Community / ABACUS Developers

Overview

pyatb (Python Ab-initio Tight-Binding) is a Python-based simulation package designed for the efficient calculation of electronic structures and physical properties using ab-initio tight-binding Hamiltonians. It is particularly tailored to work with Numerical Atomic Orbitals (NAO) and enables post-processing of Hamiltonians generated by first-principles codes like ABACUS.

Scientific domain: Computational condensed matter physics, electronic structure, topology, optics. Target user community: Users of ABACUS and other NAO-based DFT codes needing advanced property analysis.

Theoretical Methods

• Tight-Binding (TB):
  • Construction of TB Hamiltonians from First-Principles inputs.
  • Slater-Koster interpolation (implicitly via NAO matrix elements).
• Berry Physics:
  • Berry Phase, Berry Curvature, and Chern Numbers.
  • Wilson Loops for topological classification.
• Response Theory:
  • Linear and Nonlinear optical responses (Shift current, Berry curvature dipole).
  • Kubo formula for conductivity.

Capabilities

• Electronic Structure:
  • Band structures (k-line, k-mesh).
  • Density of States (DOS) and Partial DOS (PDOS).
  • Fermi Surface visualization.
  • Weyl/Dirac point search.
• Topological Analysis:
  • Anomalous Hall Conductivity (AHC).
  • $Z_2$ invariants and topological indices.
  • Geometric phase calculations.
• Optical Properties:
  • Linear optical conductivity / Dielectric function.
  • Nonlinear responses for photovoltaics (Shift current).
• Unfolding:
  • Band unfolding for supercells back to primitive representations.

Key Strengths

• Modular Design: Structured into Bands, Geometric, and Optical modules, making it easy to focus on specific properties.
• ABACUS Integration: First-class citizen support for the ABACUS DFT code, filling the gap for a native post-processing tool.
• Advanced Optics: Includes nonlinear optical response calculations often missing in standard TB tools.
• Efficiency: Core routines optimized (often C/C++ backend) for handling dense k-grids.

Inputs & Outputs

• Inputs:
  • Hamiltonian matrices (sparse/dense).
  • Geometry and orbital information.
  • From ABACUS: SR.dat (overlaps), structural files.
• Outputs:
  • Data files for bands, DOS, optical spectra.
  • Visualization-ready formats.

Interfaces & Ecosystem

• ABACUS: Primary upstream DFT code.
• Wannier90: Can interface via standard Hamiltonian formats.
• DeepModeling: Part of the broader DeepModeling ecosystem (related to DP, ABACUS).

Performance Characteristics

• Speed: Efficient sparse matrix operations allow handling relatively large systems (hundreds of atoms).
• Parallelism: MPI/OpenMP support for k-point parallelization in massive grids.

Comparison with Other Codes

• vs [WannierBerri](file:///home/niel/git/Indranil2020.github.io/scientific_tools_consolidated/TightBinding/4.1_Wannier_Ecosystem/WannierBerri.md): Both are excellent for Berry physics/optics. WannierBerri is extremely optimized for Wannier90 specifically (using FFTs); pyatb is designed for NAO/ABACUS workflows directly.
• vs [WannierTools](file:///home/niel/git/Indranil2020.github.io/scientific_tools_consolidated/TightBinding/4.1_Wannier_Ecosystem/WannierTools.md): WannierTools focuses heavily on surface states and topological surface physics (Green's functions); pyatb generally focuses on bulk properties and bulk topology/optics.

Application Areas

• Topological Materials: Screening for WSMs, TIs via Wilson loops and Chern numbers.
• Photovoltaics: Calculating shift currents in non-centrosymmetric materials (BaTiO3, TMDCs).
• Disordered Systems: Using band unfolding to analyze alloy effective bands.

Verification & Sources

• Primary Source: GitHub Repository
• Citation: (See repository for specific ArXiv preprint or publication).
• Verification Status: ✅ VERIFIED (Active development).