Parameters for Magnetic Interaction Calculations

Introduction

TB2J provides several command-line tools to calculate magnetic interactions from different electronic structure codes. Each tool shares a common set of parameters for controlling the magnetic interaction calculations, while also having some tool-specific parameters.

To view the complete list of parameters for any tool, use the –help option:

wann2J.py --help    # For Wannier90-based calculations
siesta2J.py --help  # For SIESTA-based calculations
abacus2J.py --help  # For ABACUS-based calculations

Common Parameters

These parameters are available across all TB2J tools and control the core aspects of magnetic interaction calculations.

System Definition

  • elements: List of magnetic elements to include in the Heisenberg model. Example: --elements Fe Ni

  • index_magnetic_atoms: Explicitly specify magnetic atoms by their indices (starting from 1) in the unit cell. Overrides element-based selection. Example: --index_magnetic_atoms 1 2 4

  • exclude_orbs: Zero-based indices of orbitals to exclude from magnetic site calculations. Useful for fine-tuning the magnetic model.

K-space and Real Space Parameters

  • kmesh: Three integers defining the Monkhorst-Pack k-point mesh (e.g., --kmesh 7 7 7). This mesh determines:

    • The k-points used for Green’s function calculations

    • The real-space supercell size for magnetic interactions

    • For example, a \(7 \times 7 \times 7\) k-mesh creates a supercell where magnetic interactions are calculated between the central atom and all atoms within this supercell

  • rcut: Maximum distance (in Å) for calculating magnetic interactions between atom pairs. If not specified, includes all pairs within the supercell defined by kmesh.

Energy Integration Parameters

  • efermi: Fermi energy in eV. Handling varies by tool:

    • siesta2J.py/abacus2J.py: Automatically read from output

    • wann2J.py: Must be provided from DFT calculation

    • For insulators: Can be within band gap

    • For metals: Should match the DFT value

  • smearing: Fermi smearing width for density computation and CFR. Controls the temperature used in numerical integration. Accepts values like “600K” or “0.2eV”. Default: 600K.

  • emin, emax: Energy range for integration, relative to efermi:

    • emin: Should be low enough to include all magnetically relevant states

    • Check local density of states: spin up/down should be nearly identical below emin

    • emax: Usually 0.0 (Fermi level). Can be adjusted for charge doping studies

    • Note: Direct DFT doping is preferred over emax adjustment

  • nz: Number of energy points for numerical integration of \(\int_{emin}^{emax} d\epsilon\)

Performance and Output Controls

  • np/nproc: Number of CPU cores for parallel processing. Default: 1

  • use_cache: Store wavefunctions and Hamiltonian on disk to reduce memory usage. Useful for large systems. Default: False

  • cutoff: Minimum magnitude of exchange coupling (J) to write to output (in eV). Helps filter numerical noise.

  • output_path: Directory for output files. Default: TB2J_results

Output Customization

  • description: Add essential calculation details to the XML output:

    • Exchange-correlation functional used

    • Hubbard U values if applicable

    • Magnetic state description

    • Any other relevant parameters

Advanced Options

  • orb_decomposition: Analyze orbital contributions in non-collinear calculations. Default: False

  • orth: Apply Löwdin orthogonalization before diagonalization. Usually for testing purposes. Default: False

Tool-Specific Parameters

SIESTA Interface (siesta2J.py)

  • fdf_fname: Path to SIESTA’s input fdf file. Default: ./

  • fname: Output exchange parameters XML filename. Default: exchange.xml

  • split_soc: Enable reading of spin-orbit coupling from SIESTA output. Default: False

ABACUS Interface (abacus2J.py)

  • path: Location of ABACUS calculation files. Default: ./

  • suffix: ABACUS calculation label (used in OUT.suffix). Default: abacus