Kyle A. Rocha

Elizabeth Teng, Ugur Demir, Zoheyr Doctor et al.

Knowledge about the internal physical structure of stars is crucial to understanding their evolution. The novel binary population synthesis code POSYDON includes a module for interpolating the stellar and binary properties of any system at the end of binary MESA evolution based on a pre-computed set of models. In this work, we present a new emulation method for predicting stellar profiles, i.e., the internal stellar structure along the radial axis, using machine learning techniques. We use principal component analysis for dimensionality reduction and fully-connected feed-forward neural networks for making predictions. We find accuracy to be comparable to that of nearest neighbor approximation, with a strong advantage in terms of memory and storage efficiency. By providing a versatile framework for modeling stellar internal structure, the emulation method presented here will enable faster simulations of higher physical fidelity, offering a foundation for a wide range of large-scale population studies of stellar and binary evolution.

Kyle Akira Rocha, Jeff J. Andrews, Christopher P. L. Berry et al.

Binary stars undergo a variety of interactions and evolutionary phases, critical for predicting and explaining observed properties. Binary population synthesis with full stellar-structure and evolution simulations are computationally expensive requiring a large number of mass-transfer sequences. The recently developed binary population synthesis code POSYDON incorporates grids of MESA binary star simulations which are then interpolated to model large-scale populations of massive binaries. The traditional method of computing a high-density rectilinear grid of simulations is not scalable for higher-dimension grids, accounting for a range of metallicities, rotation, and eccentricity. We present a new active learning algorithm, psy-cris, which uses machine learning in the data-gathering process to adaptively and iteratively select targeted simulations to run, resulting in a custom, high-performance training set. We test psy-cris on a toy problem and find the resulting training sets require fewer simulations for accurate classification and regression than either regular or randomly sampled grids. We further apply psy-cris to the target problem of building a dynamic grid of MESA simulations, and we demonstrate that, even without fine tuning, a simulation set of only $\sim 1/4$ the size of a rectilinear grid is sufficient to achieve the same classification accuracy. We anticipate further gains when algorithmic parameters are optimized for the targeted application. We find that optimizing for classification only may lead to performance losses in regression, and vice versa. Lowering the computational cost of producing grids will enable future versions of POSYDON to cover more input parameters while preserving interpolation accuracies.

Abigail J. Lee, Grace E. Chesmore, Kyle A. Rocha et al.

$\textit{The Bachelor}$ is a reality TV dating show in which a single bachelor selects his wife from a pool of approximately 30 female contestants over eight weeks of filming (American Broadcasting Company 2002). We collected the following data on all 422 contestants that participated in seasons 11 through 25: their Age, Hometown, Career, Race, Week they got their first 1-on-1 date, whether they got the first impression rose, and what "place" they ended up getting. We then trained three machine learning models to predict the ideal characteristics of a successful contestant on $\textit{The Bachelor}$. The three algorithms that we tested were: random forest classification, neural networks, and linear regression. We found consistency across all three models, although the neural network performed the best overall. Our models found that a woman has the highest probability of progressing far on $\textit{The Bachelor}$ if she is: 26 years old, white, from the Northwest, works as an dancer, received a 1-on-1 in week 6, and did not receive the First Impression Rose. Our methodology is broadly applicable to all romantic reality television, and our results will inform future $\textit{The Bachelor}$ production and contestant strategies. While our models were relatively successful, we still encountered high misclassification rates. This may be because: (1) Our training dataset had fewer than 400 points or (2) Our models were too simple to parameterize the complex romantic connections contestants forge over the course of a season.