Abstract:Spectral line observations encode a wealth of information. A key challenge, therefore, lies in the interpretation of these observations in terms of models to derive the physical and chemical properties of the astronomical environments from which they arise. In this paper, we present pomme: an open-source Python package that allows users to retrieve 1D or 3D models of physical properties, such as chemical abundance, velocity, and temperature distributions of (optically thin) astrophysical media, based on spectral line observations. We discuss how prior knowledge, for instance, in the form of a steady-state hydrodynamics model, can be used to guide the retrieval process, and demonstrate our methods both on synthetic and real observations of cool stellar winds.

What problem does this paper attempt to address?

The paper attempts to address the problem of how to reconstruct the physical and chemical properties of astrophysical environments from spectral line observations. Specifically, the authors propose an open-source Python package named `pomme` to derive one-dimensional or three-dimensional physical property models, such as chemical abundance, velocity, and temperature distribution, based on spectral line observation data. ### Main Issues 1. **Information Extraction**: Spectral line observations contain rich information, but extracting the physical and chemical properties of the astrophysical environment from these observations is a key challenge. 2. **Projection Problem**: Most astronomical observations are limited to the plane of the sky, so the observation results are only partial information projections of the target object, especially information along the line of sight. 3. **Reconstruction Uncertainty**: Due to the lack of observational information, there is a certain degree of uncertainty in the reconstruction model, which needs to be captured through probability distributions. ### Solutions 1. **Bayesian Method**: The authors use the Bayesian method, utilizing prior knowledge (such as steady-state hydrodynamic models) to guide the reconstruction process to improve the accuracy of the reconstruction. 2. **Forward and Inverse Problems**: - **Forward Problem**: Describes how the physical properties of the astrophysical medium lead to spectral line observations. This includes the formal solution of the radiative transfer equation and its numerical solution methods. - **Inverse Problem**: Reconstructs the physical property model of the astrophysical medium from spectral line observation data. The authors propose a probabilistic framework to determine the best model by maximizing the posterior probability. 3. **Loss Functions**: - **Reproduction Loss**: Measures the difference between the synthetic observations generated by the model and the real observations. - **Regularization Loss**: Measures the model's adherence to prior assumptions, including smoothness, symmetry, and physical law constraints. 4. **Software Implementation**: Developed the `pomme` package, implementing the above methods and testing them on synthetic and actual observation data. ### Application Examples - **Stellar Winds**: Particularly suitable for explaining the complex stellar winds around evolved stars (such as asymptotic giant branch stars and red supergiants). - **Other Objects**: Can also be applied to other astrophysical objects such as protoplanetary disks and supernova remnants. ### Conclusion Through the Bayesian method and probabilistic framework, the `pomme` package can effectively reconstruct the physical and chemical properties of astrophysical environments from spectral line observations, providing a powerful tool for astronomical research.

Bayesian model reconstruction based on spectral line observations

$\texttt{bayes_spec}$: A Bayesian Spectral Line Modeling Framework for Astrophysics

Spectroscopic analysis of hot, massive stars in large spectroscopic surveys with de-idealised models

A Bayesian direct method implementation to fit emission line spectra: Application to the primordial He abundance determination

Spectroscopic analysis of hot, massive stars in large spectroscopic surveys with de-idealized models

Comprehensive High-resolution Chemical Spectroscopy of Barnard's Star with SPIRou

FlopPITy: Enabling self-consistent exoplanet atmospheric retrievals with machine learning

Realistic Multimedia Tools based on Physical Models: I. The Spectrum Analyzer and Animator (SA2)

Bayesian inference methodology for Primordial Power Spectrum reconstructions from Large Scale Structure

Kalkayotl 2.0 Bayesian phase-space modelling of star-forming regions, stellar associations, and open clusters

Bayesian sampling with BeAtlas, a grid of synthetic Be star spectra I. Recovering the fundamental parameters of αEri and βCMi

BayesLine: Bayesian Inference for Spectral Estimation of Gravitational Wave Detector Noise

Bayesian inference of T Tauri star properties using multi-wavelength survey photometry

Reproducing Bayesian Posterior Distributions for Exoplanet Atmospheric Parameter Retrievals with a Machine Learning Surrogate Model

Bayesian Deconvolution of Astronomical Images with Diffusion Models: Quantifying Prior-Driven Features in Reconstructions

Toward a Spectral Foundation Model: An Attention-Based Approach with Domain-Inspired Fine-Tuning and Wavelength Parameterization

A Bayesian framework for spectral reprojection

Impacts of high-contrast image processing on atmospheric retrievals

Bayesian Multi-line Intensity Mapping