kmaltsev/stellar-evolution-emulators: Stellar-evolution-forecasting

This release contains two Jupyter Notebooks, stellar-evolution-emulator-fitted-models.ipynb and HNNI.ipynb. These demonstrate how the stellar evolutionary track interpolation methods based the machine-learning based surrogate models and on the HNNI algorithm work. Both are developed in Maltsev et al. 2024.

The stellar-evolution-emulator-fitted-models.ipynb Notebook shows how to use the pre-trained machine-learning based surrogate models to make fast predictions of the classical photometric variables bolometric luminosity $\log L/L_\odot$, effective temperature $\log T_\mathrm{eff}/\mathrm{K}$ and surface gravity $\log g/\mathrm{[cm/s^2]}$ of stars during their evolution from the zero-age-main-sequence (ZAMS) up to the end of core-helium burning over a ZAMS mass range $M_\mathrm{ZAMS}/M_\odot \in (0.7, 300)$. The pre-trained models are trained and tested on the MIST catalog pre-computed by Choi et al. 2016. The fitted models can be used for a variety of purposes, including rapid population synthesis and iterative-optimization based stellar parameter inference.

The HNNI.ipynb Notebook contains the Hierarchical Nearest-Neighbor Interpolation (HNNI) algorithm, which is an alternative method for automated interpolation of stellar evolution tracks that does not require their segmetation into separate evolutionary phases. HNNI is more accurate but slower than the machine-learning based surrogate modeling, and predicts any stellar evolution variable of interest. While HNNI has been demonstrated to work on the MIST catalog, the method itself is general.