MultiREx - Planetary Transmission Spectra Generator

MultiREx is a Python library designed for generating synthetic exoplanet transmission spectra. This tool extends the functionalities of the TauREx library, reorganizing and enabling the massive generation of spectra and observations with added noise. The package was originally devised for training large machine learning models to identify biosignatures in noisy spectra, but it can also be used for other purposes.

For the science behind the model, please refer to the following paper:

David S. Duque-Castaño, Jorge I. Zuluaga, and Lauren Flor-Torres (2024), Machine-assisted classification of potential biosignatures in earth-like exoplanets using low signal-to-noise ratio transmission spectra, submitted to MNRAS, arXiv:2407.19167.

Contents:

Key Features of MultiREx

Planetary System Assembly: Combine different stars, planets, and atmospheres to explore various system configurations.
Customizable Atmospheres: Configure and add a wide range of atmospheric compositions to planets.
Synthetic Spectrum Generation: Produce realistic spectra based on your system’s attributes.
Astronomical Observation Simulation: Use randinstrument to simulate spectral observations with noise determined by signal-to-noise ratios.
Multiverse Analysis: Automate the generation of multiple spectra with random parameter variations for extensive statistical or ML-based analyses.

Downloading and Installing MultiREx

MultiREx is available on the Python Package Index (PyPI) and can be installed on Linux using:

sudo pip install multirex

This command installs all dependencies and downloads useful data, scripts, and constants.

Note

If you do not have access to sudo, you can install MultiREx in your local environment (usually at ~/.local/). In that case, you need to add the local Python installation path to your PATH environment variable. For instance, add the following line to your ~/.bashrc or ~/.bash_profile:

export PATH=$HOME/.local/bin:$PATH

If you are a developer or want to work directly with the package sources, clone MultiREx from the GitHub repository:

git clone https://github.com/D4san/MultiREx-public

To install the package from the sources, run:

cd MultiREx-public
python3 setup.py install

Running MultiREx in Google Colab

To run MultiREx in Google Colab, execute:

!pip install -Uq multirex

After installing, reset the Colab session before importing the package. This avoids unintended behavior from the package pybtex. Once reset, simply import:

import multirex as mrex

Quickstart

To start using MultiREx, import the package:

import multirex as mrex

MultiREx models a planetary system with three main components: a star, a planet, and a planetary atmosphere.

First, define a star:

star = mrex.Star(temperature=5777, radius=1, mass=1)
# Radius and mass are in solar units.

Then, define a planet:

planet = mrex.Planet(radius=1, mass=1)
# Radius and mass are in Earth units.

Now give the planet an atmosphere. This is a basic example of an N2 atmosphere with 100 ppm of CO2 and 1 ppm of CH4:

atmosphere = mrex.Atmosphere(
    temperature=288,    # in K
    base_pressure=1e5,  # in Pa
    top_pressure=1,     # in Pa
    fill_gas="N2",      # the gas that fills the atmosphere
    composition=dict(
        CO2=-4,         # log10(mix-ratio)
        CH4=-6,
    )
)
planet.set_atmosphere(atmosphere)

Assemble the system:

system = mrex.System(star=star, planet=planet, sma=1)
# sma (semimajor axis) is in AU (astronomical units).

Create a transmission model:

system.make_tm()

Plot the transmission spectrum:

wns = mrex.Physics.wavenumber_grid(wl_min=0.6, wl_max=10, resolution=1000)
fig, ax = system.plot_contributions(wns, xscale='log')

All of these functionalities are also available in Taurex. However, the interface in MultiREx is more intuitive and is best suited for the package’s true strength: creating large ensembles of random planetary systems.

For instance, you can create a random planetary system by specifying parameter ranges:

system = mrex.System(
    star=mrex.Star(
        temperature=(4000, 6000),
        radius=(0.5, 1.5),
        mass=(0.8, 1.2),
    ),
    planet=mrex.Planet(
        radius=(0.5, 1.5),
        mass=(0.8, 1.2),
        atmosphere=mrex.Atmosphere(
            temperature=(290, 310),       # K
            base_pressure=(1e5, 10e5),     # Pa
            top_pressure=(1, 10),          # Pa
            fill_gas="N2",
            composition=dict(
                CO2=(-5, -4),  # log10(mix-ratio)
                CH4=(-6, -5),
            ),
        )
    ),
    sma=(0.5, 1)  # in AU
)

Using this approach, you can generate thousands of spectra for training machine learning models or other statistical analyses.

Further Examples

Several Jupyter notebooks illustrating both basic and advanced functionalities are available in the examples/ folder. Additionally, notebooks used to generate figures for our papers are in examples/papers/.

A Note About TauREx

MultiREx is built on the spectral calculation capabilities of TauREx. If you use MultiREx in your research, please also cite TauREx:

A. F. Al-Refaie, Q. Changeat, I.P. Waldmann, and G. Tinetti, “TauREx III: A fast, dynamic, and extendable framework for retrievals,” arXiv:1912.07759 (2019).

TauREx requires molecular opacities or cross-sections from sources like ExoMol or ExoTransmit. We have pre-downloaded some of these molecules; others can be downloaded with multirex.Util.get_gases().

What’s New

For a detailed list of the latest features, see What’s new.