Documentation

dftkit bundles the DFT-code interfaces that connect various DFT packages to DFTTools for DFT+DMFT calculations.

General idea

For each supported DFT code dftkit provides up to two components:

• a Converter, which reads the native output of a DFT run and writes the projected local orbitals / Hamiltonian into the TRIQS-compatible HDF5 archive expected by SumkDFT. This is all that is needed for one-shot DFT+DMFT calculations.

• a Driver, which wraps the DFT executable itself and drives a full run (and, for charge self-consistent DFT+DMFT, the repeated DFT updates inside the DMFT loop). A driver typically calls the matching converter at the end of each DFT step.

The two are organised per code under triqs_dftkit.<code>, e.g. triqs_dftkit.wien2k.Converter and triqs_dftkit.wien2k.Driver. Some codes share a converter: Quantum Espresso and Abinit reuse the Wannier90 converter and only add their own driver.

Converters and drivers

The following interfaces are available. The usage of each converter is described in the corresponding guide page of the DFTTools documentation:

Code	Converter	Driver	Guide
Wien2k	triqs_dftkit.wien2k.Converter	triqs_dftkit.wien2k.Driver	conv_wien2k
VASP	triqs_dftkit.vasp.Converter (PLOVasp) or via Wannier90	triqs_dftkit.vasp.Driver	conv_vasp
Quantum Espresso	via Wannier90	triqs_dftkit.qe.Driver	conv_W90
Abinit	via Wannier90	triqs_dftkit.abinit.Driver	conv_W90
Elk	triqs_dftkit.elk.Converter	✗	conv_elk
Wannier90	triqs_dftkit.wannier90.Converter	✗	conv_W90
H(k)	triqs_dftkit.hk.Converter	✗	conv_generalhk

Resulting hdf5 archive

All converters write into the same, standardized hdf5 layout that is read by SumkDFT. The groups and datasets are documented in standardized hdf5 structure.

• standardized hdf5 structure
  • Overview of the groups
  • Storage conventions
  • Main data: dft_input
  • Code specific data: dft_misc_input
  • Symmetry operations: dft_symmcorr_input and dft_symmpar_input
  • Partial projectors: dft_parproj_input
  • Data on a k-path: dft_bands_input
  • Transport data: dft_transp_input
  • Fermi surface contours: dft_contours_input
  • Charge density correction: dft_update
  • General and simple H(k) converter

Tutorials

End-to-end DFT+DMFT tutorials that exercise the converters live in the DFTTools documentation:

• SrVO3 (VASP, one-shot)

• Sr2MgOsO6 without SOC

• Sr2MgOsO6 with SOC

• Ce-gamma Fermi surfaces (Wien2k)

A full charge self-consistent DFT+DMFT calculation driven end-to-end through the dftkit VASP Driver is documented in the ModEST documentation:

• Charge self-consistent DFT+DMFT with VASP: SrVO3

See the full DFTTools documentation for the DMFT self-consistency loop, post-processing and advanced topics.

Python reference manual

triqs_dftkit

triqs_dftkit: DFT Converters for TRIQS