Installation

GASFIR requires Python 3.8 or higher.

Dependencies

Core (installed automatically with pip install gasfir):

NumPy ≥ 1.26.4
SciPy ≥ 1.15.3
Numba ≥ 0.61.2 — JIT compilation for kernel performance
pandas ≥ 1.5.0

Retrieval extras (pip install gasfir[retrieval]):

lmfit ≥ 1.3.3 — parameter fitting and residual construction
emcee ≥ 3.1.0 — MCMC sampling
CMA-ES (cma) ≥ 3.0.0 — global optimisation
tqdm ≥ 4.60.0 — progress bars
corner ≥ 2.2.0 — posterior corner plots

Development extras (pip install gasfir[dev]):

pytest, pytest-cov, black, isort, flake8, mypy, tabulate

Documentation extras (pip install gasfir[docs]):

Sphinx, sphinx-rtd-theme, sphinx-autodoc-typehints

Installation Methods

Using pip

Core only (ionization calculations, unit conversions, parameter store):

pip install gasfir

With fitting and retrieval pipeline:

pip install gasfir[retrieval]

With all extras:

pip install gasfir[retrieval,dev,docs]

Using conda (recommended for development)

conda env create -f environment.yml
conda activate gasfir
pip install -e ".[retrieval,dev,docs]"

To recreate environment.yml after changing dependencies:

conda env export -n gasfir --no-builds | grep -v "^prefix:" > environment.yml

From source

git clone https://gitlab.mpcdf.mpg.de/gaf/gasfir.git
cd gasfir
pip install -e ".[retrieval,dev,docs]"

Verification

After installation, verify the core package:

import gasfir
print(gasfir.__version__)

from gasfir import create_pulse, get_parameters, get_diabatic_ionization_probability
laser  = create_pulse(800, 1e14, 0, 6)
params = get_parameters("Hydrogen_SFA")
prob   = get_diabatic_ionization_probability(pulse=laser, param_dict=params)
print(f"P = {prob:.4e}")

Verify the retrieval extras:

from gasfir.retrieval import RetrievalConfig
from gasfir.fitting   import ret_residual_function
print("Retrieval extras OK")

Troubleshooting

ImportError: No module named ‘lmfit’

Fitting and retrieval require the retrieval extras:

pip install gasfir[retrieval]

ImportError: No module named ‘numba’

pip install --upgrade numba

Slow first run

Numba compiles kernels on the first call. Subsequent calls are fast. Pre-warm with a short trial pulse before your main calculation.

For development setup, see Contributing.