Stellar spectra encode detailed information about the stars. However, most machine learning approaches in stellar spectroscopy focus on supervised learning. We introduce Mendis, an unsupervised learning method, which adopts normalizing flows consisting of Neural Spline Flows and GLOW to describe the complex distribution of spectral space. A key advantage of Mendis is that we can describe the conditional distribution of spectra, conditioning on stellar parameters, to unveil the underlying structures of the spectra further. In particular, our study demonstrates that Mendis can robustly capture the pixel correlations in the spectra leading to the possibility of detecting unknown atomic transitions from stellar spectra. The probabilistic nature of Mendis also enables a rigorous determination of outliers in extensive spectroscopic surveys without the need to measure elemental abundances through existing analysis pipelines beforehand.
Stokes inversion techniques are very powerful methods for obtaining information on the thermodynamic and magnetic properties of solar and stellar atmospheres. In recent years, very sophisticated inversion codes have been developed that are now routinely applied to spectro-polarimetric observations. Most of these inversion codes are designed for finding an optimum solution to the nonlinear inverse problem. However, to obtain the location of potentially multimodal cases (ambiguities), the degeneracies, and the uncertainties of each parameter inferred from the inversions, algorithms such as Marko…
We discuss how to achieve mapping from large sets of imperfect simulations and observational data with unsupervised domain adaptation. Under the hypothesis that simulated and observed data distributions share a common underlying representation, we show how it is possible to transfer between simulated and observed domains. Driven by an application to interpret stellar spectroscopic sky surveys, we construct the domain transfer pipeline from two adversarial autoencoders on each domains with a disentangling latent space, and a cycle-consistency constraint. We then construct a differentiable pipel…
With the upcoming plethora of astronomical time-domain datasets and surveys, anomaly detection as a way to discover new types of variable stars and transients has inspired a new wave of research. Yet, the fundamental definition of what constitutes an anomaly and how this depends on the overall properties of the population of light curves studied remains a discussed issue. Building on a previous study focused on Kepler light curves, we present an analysis that uses the Unsupervised Random Forest to search for anomalies in TESS light curves. We provide a catalogue of anomalous light curves, clas…
Chromospheric sensitivity of stellar spectral lines: an unsupervised machine learning classification
Instituto de Astronomía y Física del Espacio (IAFE, CONICET–UBA)（天文与空间物理研究所）；Laboratorio de Física, Universidad Nacional de Tres de Febrero (UNTREF)（特塞德费弗罗国立大学物理实验室）；Ciclo Básico Común, Universidad de Buenos Aires (UBA)（布宜诺斯艾利斯大学基础循环学院）
Metal-poor stars play an import role in the understanding of Galaxy formation and evolution. Evidence of the early mergers that built up the Galaxy might remain in the distributions of abundances, kinematics, and orbital parameters of the stars. In this work, we report on preliminary results of an on-going chemo-kinematic analysis of a sample of metal-poor ([Fe/H] $\leq$ -1.0) stars observed by the GALAH spectroscopic survey. We explored the chemical and orbital data with unsupervised machine learning (hierarchical clustering, k-means cluster analysis and correlation matrices). Our final goal …
Unsupervised learning for variability detection with Gaia DR3 photometry. The main sequence-white dwarf valley
Radboud University（拉波德大学）；KU Leuven（荷语鲁汶大学）；University of Surrey（萨里大学）；Universitat de Barcelona（巴塞罗那大学）；University of Cape Town（开普敦大学）；South African Astronomical Observatory（南非天文台）；The Inter-University Institute for Data Intensive Astronomy, University of Cape Town（开普敦大学数据密集型天文学跨大学研究所）；Hamburger Sternwarte, University of Hamburg（汉堡天文台，汉堡大学）；Texas Tech University（德克萨斯理工大学）
Stars with peculiar element abundances are important markers of chemical enrichment mechanisms. We present a simple method, tangent space projection (TSP), for the detection of $X$-enriched stars, for arbitrary elements $X$, even from blended lines. Our method does not require stellar labels, but instead directly estimates the counterfactual unrenriched spectrum from other unlabelled spectra. As a case study, we apply this method to the $6708~$Å Li doublet in LAMOST DR5, identifying 8,428 Li-enriched stars seamlessly across evolutionary state. We comment on the explanation for Li-enrichement f…
Machine learning in APOGEE: Unsupervised spectral classification with $K$-means
Instituto de Astrofísica de Canarias（加那利天体物理研究所）；Universidad de La Laguna（拉古纳大学）；Instituto de Física de Cantabria (CSIC-UC)（坎塔布里亚物理研究所（西班牙国家研究委员会-坎塔布里亚大学））
Stellar Karaoke: deep blind separation of terrestrial atmospheric effects out of stellar spectra by velocity whitening
Department of Astronomy, University of Washington（华盛顿大学天文系）；The Raw Data Speaks Initiative（原始数据发声计划）；Department of Electrical Engineering, Amirkabir University of Technology（阿米尔卡比尔理工大学电气工程系）
Flow-Based Generative Emulation of Grids of Stellar Evolutionary Models
Kavli Institute for Astrophysics and Space Research, Massachusetts Institute of Technology（麻省理工学院卡弗里天体物理与空间研究所）；Institute for Astronomy, University of Hawai‘i（夏威夷大学天文研究所）
We present our 500 pc distance-limited study of stellar fares using the Dark Energy Camera as part of the Deeper, Wider, Faster Program. The data was collected via continuous 20-second cadence g band imaging and we identify 19,914 sources with precise distances from Gaia DR2 within twelve, ~3 square-degree, fields over a range of Galactic latitudes. An average of ~74 minutes is spent on each field per visit. All light curves were accessed through a novel unsupervised machine learning technique designed for anomaly detection. We identify 96 flare events occurring across 80 stars, the majority o…
