Skip to content

README.md

Latest commit

 

History

History
102 lines (63 loc) · 3.86 KB

README.md

File metadata and controls

102 lines (63 loc) · 3.86 KB
Raw

TREXIO tools

PyPI version

Set of tools for TREXIO files.

Requirements

  • python3 (>=3.6)
  • trexio (>=1.0.0) [Python API]
  • numpy (>=1.17.3)
  • resultsFile [optional, for GAMESS/GAU$$IAN conversion]
  • VeloxChem [optional, for VLX/VeloxChem HDF5 conversion]
  • docopt [for CLI] Note: the pyscf<--->trexio converter has been moved to pyscf-forge. Please install the pyscf-forge plugin for the pyscf<--->trexio interface (including multi-reference wave function I/O).

Installation

Installation via PyPI, periodically updated

pip install trexio-tools

Installation from source code

pip install git+https://github.com/TREX-CoE/trexio_tools

Optional converter back ends rely on additional packages in the active Python environment. In particular, the VLX converter requires veloxchem, while the Gaussian/GAMESS converters require resultsFile.

Instructions for users

After installation, trexio-tools provides an entry point, which can be accessed via CLI:

trexio --help

This will list all currently supported command line arguments. For example,

trexio convert-from -t gamess -i data/GAMESS_CAS.log -b hdf5 trexio_cas.hdf5

converts data from the GAMESS_CAS.log output file of the GAMESS code (note also -t gamess argument) into the TREXIO file called trexio_cas.hdf5 using -b hdf5 back end of TREXIO.

For ORCA,

trexio convert-from -t orca -i data/h2o.json -b hdf5 trexio_orca.hdf5

converts data from the h2o.json output file of the ORCA code into the TREXIO file called trexio_orca.hdf5 using the hdf5 back end of TREXIO.

Note that ORCA AOs and MOs are written in spherical coordinates. If you need a cartesian TREXIO file, run the additional conversion step below.

trexio convert-to -t cartesian -o trexio_orca_cart.hdf5 trexio_orca.hdf5

This converts the spherical AOs and MOs from trexio_orca.hdf5 into cartesian coordinates and writes the result to trexio_orca_cart.hdf5.

For a Gaussian formatted checkpoint file,

trexio convert-from -t fchk -i data/h2.fchk -b hdf5 trexio_fchk.hdf5

converts the geometry, Gaussian basis set, and molecular orbitals from the .fchk file into the TREXIO file trexio_fchk.hdf5. Both restricted (RHF/ROHF) and unrestricted (UHF) checkpoints are supported, with either cartesian (6D/10F) or pure spherical (5D/7F) shells, up to g angular momentum. SP shells are automatically split into the separate s and p shells required by TREXIO. This converter has no external dependencies (it does not require resultsFile).

Because TREXIO uses a single global cartesian/spherical flag, fchk files that mix cartesian and pure shells within one calculation (e.g. Gaussian's 6D 7F) cannot be represented and are rejected with an explicit error.

For VeloxChem,

trexio convert-from -w -t vlx -i /path/to/biphenyl-scf.h5 -b hdf5 biphenyl-scf-trexio.hdf5

converts data from a VeloxChem HDF5 file into the TREXIO file called biphenyl-scf-trexio.hdf5 using the hdf5 back end. The -w flag is useful when rerunning the converter against an existing output file. The active Python environment must provide the veloxchem package.

VeloxChem validation

The VeloxChem converter writes and validates the following quantities when they are present in the input HDF5 file: molecular geometry, electron counts, Gaussian basis data, AO shell map, MO coefficients, MO energies, MO occupations, AO overlap matrix, and SCF energy.

For local regression checks, use:

python utilities/run_vlx_regression.py --case biphenyl=/path/to/biphenyl-scf.h5 --case tempo_roscf=/path/to/tempo-roscf.h5 --case trityl_uscf=/path/to/trityl-uscf.h5

This script converts each VeloxChem file to TREXIO and then compares the written TREXIO contents numerically against the source VeloxChem data using utilities/compare_vlx_trexio.py.