Installation

The Python package installs with pip, but the calculation backends (DFTB+, modes, xtb, tblite) are compiled programs distributed via conda-forge and cannot be installed with pip.

Package only

python -m pip install thermoscreening

Full environment (recommended)

Create an environment containing every backend, then install the released Python package:

conda create -n thermoscreening -c conda-forge \
    python=3.12 dftbplus xtb tblite-python pip
conda activate thermoscreening
python -m pip install thermoscreening

For development from a checkout, conda env create -f environment.yml installs the same backends and an editable package with the test tools.

After installation, thermo doctor reports every backend and lists the usable engines. Use thermo doctor --engine dftb+ (or xtb/xtb-cli) when a specific backend is required.

Backends

Backend	Needed for	Install
dftb+, modes	the DFTB+ engine	conda install -c conda-forge dftbplus
xtb	--engine xtb-cli (native xtb)	conda install -c conda-forge xtb
tblite	--engine xtb (in-process GFN-xTB)	conda install -c conda-forge tblite-python

DFTB+ and Slater-Koster setup

The DFTB+ engine needs Slater-Koster (.skf) parameter files. ThermoScreening downloads supported sets to an automatically discovered user-local directory:

thermo setup-dftb --parameter-set 3ob      # or: --parameter-set mio
thermo setup-dftb --solvent water          # a GBSA solvent parameter file

Use DFTB_PREFIX only to override that directory with a custom installation. thermo doctor verifies that executables can start, checks the selected parameter set, and reports the optional xTB backends.