Sven Buder

We report the discovery, validation, and characterization of 44 new symbiotic stars (SySts) accreting at the lowest detectable levels (L _acc ≲3 L _⊙ ). Candidates were initially selected from the Galah survey among non-pulsating M-type red giants exhibiting H α excess. Between 2021 and 2025, these targets were subjected to intensive multi-wavelength campaigns, including ground-based, high-resolution Echelle spectroscopy over 544 nights, multi-epoch flickering searches down to 1 mmag, and UV/X-ray observations with the Swift satellite. Our high-resolution spectra allowed for a detailed investigation of H α flux and line-profile variability. Crucially, the radial velocities measured for the M-giant components yielded reliable spectroscopic orbits for 22 objects and preliminary orbits for an additional eight, corroborated by binary-induced modulations in ASAS-SN photometry. Measured accretion levels vary by factors of up to 20×, emphasizing the need for high-sensitivity, long-term monitoring. Swift X-ray data exclude neutron star companions in all targets, consistent with predominance of white-dwarf degenerates in known SySts. Combined with the 33 objects from Paper I, this sample of 77 new SySts significantly exceeds the number of previously known, accreting-only Galactic symbiotic stars, constituting the bulk of the low-accretion (LowAcc) variety discovered to date and representing >10% of all known SySts in the Galaxy. As SySts likely spend most of their lifetimes in LowAcc faint states, characterizing this population is essential for estimating the total Galactic SySt population and evaluating the role of SySts as Type Ia supernova progenitors or sources of UV excess in elliptical galaxies.

Stellar abundances are generated by complex chemical enrichment pathways yet show strong correlation structure. We test how well high-quality spectra of low-α disk stars can be reconstructed using low-dimensional models. By validating model performance in flux space, we quantify and locate where residual information persists. Using HARPS spectra (R = 115, 000, SNR >100) and labels derived with korg, we train generative models (the cannon) conditioned on stellar parameters Teff, log (g), [M/H], vsin i, and vmic, and three abundance labels, [Ti/H], [Y/H], and [Nd/H], representing distinct nucleosynthetic families. Evaluating model performance on unseen test stars through flux residuals, the model reproduces the global spectra to a median residual amplitude of 0.3%. Deviations in 15 absorption features for unlabelled elements are typically ≤0.7%. Residuals of ∼1% correspond to abundance differences of 0.01–0.05 dex. Degrading the spectra to medium resolution (R = 28, 000) decreases the median residual amplitude from 0.3% to 0.2%, redistributing the residual signal over the instrumental line profile. Our publicly available model reproduces features from 29 elements to sub-percent precision across −1.13 < [M/H] < 0.29. This demonstrates that the dominant chemical variation encoded in low-α disk spectra can be captured by a low-dimensional model, while weaker chemical information remains encoded in flux residuals. High-resolution, high-SNR spectra (R ∼ 100, 000, SNR >100) are essential to resolve the hierarchical layers of chemical information beyond this low-dimensional structure. The model provides a benchmark for assessing the chemical information accessible to current surveys like 4MOST and GALAH 2, as well as upcoming facilities like VLT/HRMOS and GMT/G-CLEF.

The question of whether genuinely young high- α stars exist has been discussed for more than a decade since their discovery from asteroseismology of giant stars, as it is challenging to break the degeneracy between the binary interaction and the genuinely young scenarios. Young high- α stars are hard to explain with a traditional chemical evolution model, as the high- α disk is typically associated with the early epoch of star formation in the Milky Way. Combined with recent advances of gyrochronology, and that ⁷ Li can serve as an unambiguous indicator for identifying merger products in dwarfs (thanks to its low burning temperature), we identified young high- α dwarf candidate stars through their fast rotation in a previous study. For this paper, we performed high-resolution spectroscopic follow-up on these candidates using the Potsdam Echelle Polarimetric and Spectroscopic Instrument, and confirm three additional stars that are most likely genuinely young. Together with the star from the earlier paper, we find three out of four of them center around [Fe/H] = −0.5 dex, are ∼5 Gyr old, and have a similar amount of elevated Li (∼0.5 dex) and Al (∼0.1 dex) compared to stars with matching log g , T _eff , Mg, and Fe within observational uncertainties, hinting at their common formation pathway.

We present a comprehensive data-driven spectroscopic analysis of 357,415 red giant stars using Gaia DR3 Radial Velocity Spectrometer (RVS) spectra (8460–8700 Å; R ≈ 11, 500), aimed at deriving precise stellar parameters and elemental abundances (collectively referred to as stellar labels). We employ The Cannon, a generative model based on 2747 giants in common with GALAH DR4, adopting GALAH labels (R ≈ 28, 000) for training. The resulting model predicts eleven stellar labels for RVS giants: effective temperature (Teff), surface gravity (log g), projected rotational velocity (vsin i), and abundances of [Fe/H], [Ca/Fe], [Si/Fe], [Ni/Fe], [Ti/Fe], as well as the neutron-capture elements [Zr/Fe], [Ce/Fe], and [Nd/Fe]. Building on these results, we develop a probabilistic framework to chemically identify debris from the Gaia–Sausage–Enceladus (GSE) accretion event. A logistic regression classifier, optimised via Markov Chain Monte Carlo sampling and trained on a small reference sample of GSE members and comparison stars, identifies stars with high GSE membership probabilities based solely on their chemical abundances, with the resulting candidates exhibiting distinctive abundance-ratio patterns, including [Ca/Ti], [Ti/Ce], and [Nd/Zr]. Applying independent kinematic constraints yields a robust sample of GSE candidates, demonstrating that the characteristic chemical signatures remain consistent after applying these constraints. This work demonstrates the power of data-driven analysis techniques to extract detailed chemical information from medium-resolution spectra and establishes a framework for tracing Galactic accretion events using chemical abundances.

Context . Recent studies have revealed that the Milky Way’s stellar halo is a composite of stellar populations of different origins, including multiple accretion events. To better understand how the Milky Way and other spiral galaxies were formed, it is necessary to thoroughly characterize the chemical and kinematic properties of these structures. Aims . We search for kinematic structures of the stellar halo to find any substructures within them (when indeed present) and characterize the chemodynamical properties of the identified groups with the GALAH DR4 and Gaia surveys. Methods . We applied wavelet transforms in the space defined by a square root of radial action ( J _r ) and azimuthal action ( L _z ) to search for kinematic overdensities. Then, we selected stars in the detected structures and investigated their elemental abundances to determine their origin. Additionally, we checked for any contamination from other stellar populations within the detected groups with the unsupervised machine learning algorithm t-Distributed Stochastic Neighbor Embedding (t-SNE), for which we performed chemical tagging in a high-dimensional parameter space using 15 elemental abundances as input. Results . We recovered five kinematic structures in the action space with the wavelet transform. These groups are the Galactic disk, Splash, Gaia -Sausage-Enceladus (GSE), Thamnos1, and Thamnos2. We found that GSE has two peaks with the wavelet transform. One of these peaks is located at J _r ≃ 25 kpc km s ⁻¹ and is a result of contamination from disk stars. The other peak corresponds to the ‘cleanest’ GSE population and is located above J _r ≃ 40 kpc km s ⁻¹ . We also detected three peaks in Thamnos. We linked two of them to Thamnos1, while the peak with the stars on the most retrograde orbits was linked to Thamnos2. The t-SNE algorithm confirmed these findings. We also analyzed individual elemental abundances of each group and found that Thamnos2 has a higher [ α /Fe] ratio than the other groups and that iron-peak elements are more abundant in the Splash than in the halo groups, while the halo structures retain a higher r -process signature than the splashed disk. Conclusions . A multiply peaked substructure we observe in action space in GSE and Thamnos suggests that the splashed disk extends beyond the borders of prograde orbits. Each of the four halo groups studied in this paper have unique chemodynamical properties that confirm their extra-galactic origin.