ExoIris is a Python package for modeling exoplanet transmission spectroscopy. ExoIris removes the typical limitations of the two-step workflow by modeling the full two-dimensional spectroscopic transit time series directly. It supports combining transmission spectroscopy datasets from multiple instruments observed in different epochs, yielding self-consistent wavelength-independent and wavelength-dependent parameters, simplifying joint analyses, and delivering results quickly.
Transmission spectroscopy is often done following a two-step workflow: (1) fit a white light curve to infer wavelength-independent parameters; (2) fit each spectroscopic light curve independently, constrained by the white-light solution. This split can introduce approximations and inconsistencies.
ExoIris takes a different approach. It models spectrophotometric time series end-to-end, enabling:
- Self-consistent inference of shared (wavelength-independent) and spectral (wavelength-dependent) parameters.
- Joint modeling of multiple datasets from different instruments and epochs.
- Accounting for transit timing variations and dataset-dependent offsets within a unified framework.
This design is a natural fit for JWST-class data, where correlated noise, multi-epoch observations, and cross-instrument combinations are the norm.
Full documentation and tutorials: https://exoiris.readthedocs.io
Install from PyPI:
pip install exoirisLatest development version:
git clone https://github.com/hpparvi/ExoIris.git
cd ExoIris
pip install -e .ExoIris supports Python 3.9+. See the docs for dependency details and optional extras.
-
Direct modelling of spectroscopic transit time series
Built on PyTransit’sTSModel, optimised for transmission spectroscopy; scales to hundreds–thousands of light curves simultaneously. -
Flexible limb darkening
Use standard analytical laws (quadratic, power-2, non-linear), numerical intensity profiles from stellar atmosphere models, or user-defined radially symmetric functions. -
Robust noise treatment
Choose white noise or time-correlated noise via a Gaussian Process likelihood, without changing the overall workflow. -
Full control of spectral resolution
The transmission spectrum is represented as a cubic spline with user-defined knots, allowing variable resolution across wavelength. -
Reproducible, incremental workflows
Save and reload models to refine a low-resolution run into a high-resolution analysis seamlessly. -
Joint multi-dataset analyses
Combine instruments and epochs in one fit, with support for transit timing variations and dataset-specific systematics and offsets.
ExoIris is designed for speed and stability:
- A transmission spectroscopy analysis of a single JWST/NIRISS dataset at R ≈ 100 typically runs in 3–5 minutes assuming white noise, or 5–15 minutes with a GP noise model, on a standard desktop CPU.
- A high-resolution analysis of the JWST/NIRISS WASP-39 b dataset (~3800 spectroscopic light curves; see Feinstein et al. 2023) can be optimised and sampled in about 1.5 hours on an AMD Ryzen 7 5800X (8 cores, ~3-year-old desktop).
© 2025 Hannu Parviainen