Search

We report on a data-driven spectral model that we have developed for the identification of double-lined spectroscopic binary stars (SB2s) in the LAMOST low-resolution survey (R$\sim$1800). Employing simultaneous fitting with both single-star and binary-star models, we detected over 4800 SB2 candidates, where both components are detectably contributing to the spectrum, from an initial pool of 2.6 million objects. Tests show that our model favors FGK-type main-sequence binaries with high mass ratio ($q\geq$ 0.7) and large radial velocity separation ($\Delta \rm RV \geq$ 100~km$\,$s$^{-1}$). Almost all of these candidates are positioned above the main sequence in the color-magnitude diagram, indicating their binary nature. Additionally, we utilized various observational data, including spectroscopy, photometry, parallax, and extinction, to determine multiple physical parameters such as the effective temperature, age, metallicity, radial velocity, mass, mass ratio, stellar radius, along with their associated uncertainties for these SB2 candidates. For the 44 candidates with seven or more observational epochs, we provide complete orbital solutions. We make available catalogs containing various stellar parameters for identified SB2 systems., Comment: 29 pages, 23 figures and 5 tables; Accepted for publication in ApJS

Continual pretraining of large language models on domain-specific data has been proposed to enhance performance on downstream tasks. In astronomy, the previous absence of astronomy-focused benchmarks has hindered objective evaluation of these specialized LLM models. Leveraging a recent initiative to curate high-quality astronomical MCQs, this study aims to quantitatively assess specialized LLMs in astronomy. We find that the previously released AstroLLaMA series, based on LLaMA-2-7B, underperforms compared to the base model. We demonstrate that this performance degradation can be partially mitigated by utilizing high-quality data for continual pretraining, such as summarized text from arXiv. Despite the observed catastrophic forgetting in smaller models, our results indicate that continual pretraining on the 70B model can yield significant improvements. However, the current supervised fine-tuning dataset still constrains the performance of instruct models. In conjunction with this study, we introduce a new set of models, AstroLLaMA-3-8B and AstroLLaMA-2-70B, building upon the previous AstroLLaMA series., Comment: 10 pages, 1 figure, 1 table, accepted to AI4S: The 5th Workshop on Artificial Intelligence and Machine Learning for Scientific Applications at the International Conference for High Performance Computing, Networking, Storage, and Analysis (SC24). Models will be released at https://huggingface.co/AstroMLab. AstroMLab homepage: https://astromlab.org/

Astronomical research traditionally relies on extensive domain knowledge to interpret observations and narrow down hypotheses. We demonstrate that this process can be emulated using large language model-based agents to accelerate research workflows. We propose mephisto, a multi-agent collaboration framework that mimics human reasoning to interpret multi-band galaxy observations. mephisto interacts with the CIGALE codebase, which includes spectral energy distribution (SED) models to explain observations. In this open-world setting, mephisto learns from its self-play experience, performs tree search, and accumulates knowledge in a dynamically updated base. As a proof of concept, we apply mephisto to the latest data from the James Webb Space Telescope. mephisto attains near-human proficiency in reasoning about galaxies' physical scenarios, even when dealing with a recently discovered population of "Little Red Dot" galaxies. This represents the first demonstration of agentic research in astronomy, advancing towards end-to-end research via LLM agents and potentially expediting astronomical discoveries., Comment: Submitted. Comments are welcome!

Our understanding of the assembly timeline of the Milky Way has been transforming along with the dramatic increase in astrometric and spectroscopic data available over the past several years. Many substructures in chemo-dynamical space have been discovered and identified as the remnants of various galactic mergers. To investigate the timeline of these mergers we select main sequence turn off & subgiant stars (MSTOs) from the H3 survey, finding members in seven metal poor components of the halo: GSE, the Helmi Streams, Thamnos, Sequoia, Wukong/LMS-1, Arjuna, and I'itoi. We also select out the metal poor in situ disk to facilitate comparison to the evolution of the Milky Way itself at these early epochs. We fit individual isochrone ages to the MSTOs in each of these substructures and use the resulting age distributions to infer simple star formation histories. For GSE we resolve an extended star formation history that truncates $\approx10$ Gyr ago, as well as a clear age -- metallicity relation. From this age distribution and measured star formation history we infer that GSE merged with the Milky Way at a time $9.5-10.2$ Gyr ago, in agreement with previous estimates. We infer that the other mergers occurred at various times ranging from $9-13$ Gyr ago, and that the metal poor component of the disk built up within only a few billion years. These results reinforce the emerging picture that both the disk and halo of the Milky Way experienced a rapid assembly., Comment: 27 pages, 17 figures

We present a proof-of-concept simulation-based inference on $\Omega_{\rm m}$ and $\sigma_{8}$ from the SDSS BOSS LOWZ NGC catalog using neural networks and domain generalization techniques without the need of summary statistics. Using rapid lightcone simulations, ${\rm L{\scriptsize -PICOLA}}$, mock galaxy catalogs are produced that fully incorporate the observational effects. The collection of galaxies is fed as input to a point cloud-based network, ${\texttt{Minkowski-PointNet}}$. We also add relatively more accurate ${\rm G{\scriptsize ADGET}}$ mocks to obtain robust and generalizable neural networks. By explicitly learning the representations which reduces the discrepancies between the two different datasets via the semantic alignment loss term, we show that the latent space configuration aligns into a single plane in which the two cosmological parameters form clear axes. Consequently, during inference, the SDSS BOSS LOWZ NGC catalog maps onto the plane, demonstrating effective generalization and improving prediction accuracy compared to non-generalized models. Results from the ensemble of 25 independently trained machines find $\Omega_{\rm m}=0.339 \pm 0.056$ and $\sigma_{8}=0.801 \pm 0.061$, inferred only from the distribution of galaxies in the lightcone slices without relying on any indirect summary statistics. A single machine that best adapts to the ${\rm G{\scriptsize ADGET}}$ mocks yields a tighter prediction of $\Omega_{\rm m}=0.282 \pm 0.014$ and $\sigma_{8}=0.786 \pm 0.036$. We emphasize that adaptation across multiple domains can enhance the robustness of the neural networks in observational data., Comment: 28 pages, 14 figures; Accepted for publication in ApJ

Dwarf galaxy star formation histories are theoretically expected to be bursty, potentially leaving distinct imprints on their chemical evolution. We propose that episodic starbursts with quiescent periods longer than $\sim$100 Myr should lead to discontinuous tracks in a dwarf galaxy's [$\alpha$/Fe]-[Fe/H] chemical abundance plane, with metallicity gaps as large as 0.3-0.5 dex at [Fe/H] = -2. This occurs due to continued Fe production by Type Ia supernovae during quiescent periods. We demonstrate that Gaussian mixture models can statistically distinguish discontinuous and continuous tracks based on the Akaike Information Criterion. Applying this method to APOGEE observations of the Sculptor dSph galaxy suggests an episodic star formation history with $\sim$300 Myr quiescent periods. While current dwarf galaxy datasets are limited by small spectroscopic sample sizes, future surveys and extremely large telescopes will enable determining large numbers of precise chemical abundances, opening up the investigation of very short timescales in early dwarf galaxy formation. This unprecedentedly high time resolution of dwarf galaxy formation in the early Universe has important implications for understanding both reionization in the early Universe and the episodic star formation cycle of dwarf galaxies., Comment: 16 pages, 5 figures, submitted to OJA

The exponential growth of astronomical literature poses significant challenges for researchers navigating and synthesizing general insights or even domain-specific knowledge. We present Pathfinder, a machine learning framework designed to enable literature review and knowledge discovery in astronomy, focusing on semantic searching with natural language instead of syntactic searches with keywords. Utilizing state-of-the-art large language models (LLMs) and a corpus of 350,000 peer-reviewed papers from the Astrophysics Data System (ADS), Pathfinder offers an innovative approach to scientific inquiry and literature exploration. Our framework couples advanced retrieval techniques with LLM-based synthesis to search astronomical literature by semantic context as a complement to currently existing methods that use keywords or citation graphs. It addresses complexities of jargon, named entities, and temporal aspects through time-based and citation-based weighting schemes. We demonstrate the tool's versatility through case studies, showcasing its application in various research scenarios. The system's performance is evaluated using custom benchmarks, including single-paper and multi-paper tasks. Beyond literature review, Pathfinder offers unique capabilities for reformatting answers in ways that are accessible to various audiences (e.g. in a different language or as simplified text), visualizing research landscapes, and tracking the impact of observatories and methodologies. This tool represents a significant advancement in applying AI to astronomical research, aiding researchers at all career stages in navigating modern astronomy literature., Comment: 25 pages, 9 figures, submitted to AAS jorunals. Comments are welcome, and the tools mentioned are available online at https://pfdr.app

We present a comprehensive evaluation of proprietary and open-weights large language models using the first astronomy-specific benchmarking dataset. This dataset comprises 4,425 multiple-choice questions curated from the Annual Review of Astronomy and Astrophysics, covering a broad range of astrophysical topics. Our analysis examines model performance across various astronomical subfields and assesses response calibration, crucial for potential deployment in research environments. Claude-3.5-Sonnet outperforms competitors by up to 4.6 percentage points, achieving 85.0% accuracy. For proprietary models, we observed a universal reduction in cost every 3-to-12 months to achieve similar score in this particular astronomy benchmark. Open-source models have rapidly improved, with LLaMA-3-70b (80.6%) and Qwen-2-72b (77.7%) now competing with some of the best proprietary models. We identify performance variations across topics, with non-English-focused models generally struggling more in exoplanet-related fields, stellar astrophysics, and instrumentation related questions. These challenges likely stem from less abundant training data, limited historical context, and rapid recent developments in these areas. This pattern is observed across both open-weights and proprietary models, with regional dependencies evident, highlighting the impact of training data diversity on model performance in specialized scientific domains. Top-performing models demonstrate well-calibrated confidence, with correlations above 0.9 between confidence and correctness, though they tend to be slightly underconfident. The development for fast, low-cost inference of open-weights models presents new opportunities for affordable deployment in astronomy. The rapid progress observed suggests that LLM-driven research in astronomy may become feasible in the near future., Comment: 45 pages, 12 figures, 7 tables. Submitted to ApJ. Comments welcome. AstroMLab homepage: https://astromlab.org/

We undertake a comprehensive investigation into the distribution of insitu stars within Milky Way-like galaxies, leveraging TNG50 simulations and comparing their predictions with data from the H3 survey. Our analysis reveals that 28% of galaxies demonstrate reasonable agreement with H3, while only 12% exhibit excellent alignment in their profiles, regardless of the specific spatial cut employed to define insitu stars. To uncover the underlying factors contributing to deviations between TNG50 and H3 distributions, we scrutinize correlation coefficients among internal drivers(e.g., virial radius, star formation rate [SFR]) and merger-related parameters (such as the effective mass-ratio, mean distance, average redshift, total number of mergers, average spin-ratio and maximum spin alignment between merging galaxies). Notably, we identify significant correlations between deviations from observational data and key parameters such as the median slope of virial radius, mean SFR values, and the rate of SFR change across different redshift scans. Furthermore, positive correlations emerge between deviations from observational data and parameters related to galaxy mergers. We validate these correlations using the Random Forest Regression method. Our findings underscore the invaluable insights provided by the H3 survey in unraveling the cosmic history of galaxies akin to the Milky Way, thereby advancing our understanding of galactic evolution and shedding light on the formation and evolution of Milky Way-like galaxies in cosmological simulations., Comment: 20 pages, 13 figures

Stellar spectra emulators often rely on large grids and tend to reach a plateau in emulation accuracy, leading to significant systematic errors when inferring stellar properties. Our study explores the use of Transformer models to capture long-range information in spectra, comparing their performance to The Payne emulator (a fully connected multilayer perceptron), an expanded version of The Payne, and a convolutional-based emulator. We tested these models on synthetic spectra grids, evaluating their performance by analyzing emulation residuals and assessing the quality of spectral parameter inference. The newly introduced TransformerPayne emulator outperformed all other tested models, achieving a mean absolute error (MAE) of approximately 0.15% when trained on the full grid. The most significant improvements were observed in grids containing between 1000 and 10,000 spectra, with TransformerPayne showing 2 to 5 times better performance than the scaled-up version of The Payne. Additionally, TransformerPayne demonstrated superior fine-tuning capabilities, allowing for pretraining on one spectral model grid before transferring to another. This fine-tuning approach enabled up to a tenfold reduction in training grid size compared to models trained from scratch. Analysis of TransformerPayne's attention maps revealed that they encode interpretable features common across many spectral lines of chosen elements. While scaling up The Payne to a larger network reduced its MAE from 1.2% to 0.3% when trained on the full dataset, TransformerPayne consistently achieved the lowest MAE across all tests. The inductive biases of the TransformerPayne emulator enhance accuracy, data efficiency, and interpretability for spectral emulation compared to existing methods., Comment: 25 pages, 9 figures

The kinematics of the stellar halo hold important clues to the assembly history and mass distribution of the Galaxy. In this study, we map the kinematics of stars across the Galactic halo with the H3 Survey. We find a complex distribution that breaks both azimuthal symmetry about the $Z$-axis and mirror symmetry about the Galactic plane. This asymmetry manifests as large variations in the radial velocity dispersion $\sigma_r$ from as ``cold'' as 70 $\text{km}\text{ s}^{-1}$ to as ``hot'' as 160 $\text{km}\text{ s}^{-1}$. We use stellar chemistry to distinguish accreted stars from in-situ stars in the halo, and find that the accreted population has higher $\sigma_r$ and radially biased orbits, while the in-situ population has lower $\sigma_r$ and isotropic orbits. As a result, the Galactic halo kinematics are highly heterogeneous and poorly approximated as being spherical or axisymmetric. We measure radial profiles of $\sigma_r$ and the anisotropy parameter $\beta$ over Galactocentric radii $10-80\text{ kpc}$, and find that discrepancies in the literature are due to the nonspherical geometry and heterogeneous nature of the halo. Investigating the effect of strongly asymmetric $\sigma_r$ and $\beta$ on equilibrium models is a path forward to accurately constraining the Galactic gravitational field, including its total mass., Comment: submitted to ApJ; comments welcome

Identifying and predicting the factors that contribute to the success of interdisciplinary research is crucial for advancing scientific discovery. However, there is a lack of methods to quantify the integration of new ideas and technological advancements in astronomical research and how these new technologies drive further scientific breakthroughs. Large language models, with their ability to extract key concepts from vast literature beyond keyword searches, provide a new tool to quantify such processes. In this study, we extracted concepts in astronomical research from 297,807 publications between 1993 and 2024 using large language models, resulting in a set of 24,939 concepts. These concepts were then used to form a knowledge graph, where the link strength between any two concepts was determined by their relevance through the citation-reference relationships. By calculating this relevance across different time periods, we quantified the impact of numerical simulations and machine learning on astronomical research. The knowledge graph demonstrates two phases of development: a phase where the technology was integrated and another where the technology was explored in scientific discovery. The knowledge graph reveals that despite machine learning has made much inroad in astronomy, there is currently a lack of new concept development at the intersection of AI and Astronomy, which may be the current bottleneck preventing machine learning from further transforming the field of astronomy., Comment: An interactive version of the knowledge graph is made publicly available at https://astrokg.github.io/. Accepted to IJCAI 2024 AI4Research Workshop. Comments are welcome

The infall of the Large Magellanic Cloud (LMC) is predicted to displace the inner Milky Way (MW), imprinting an apparent 'reflex motion' on the observed velocities of distant halo stars. We construct the largest all-sky spectroscopic dataset of luminous red giant stars from $50-160$ kpc, including a new survey of the southern celestial hemisphere. We fit the full 6D kinematics of our data to measure the amplitude and direction of the inner MW's motion towards the outer halo. The observed velocity grows with distance such that, relative to halo stars at $100$ kpc, the inner MW is lurching at $\approx 40$ km s$^{-1}$ towards a recent location along the LMC's past orbit. Our measurements align with N-body simulations of the halo's response to a $1.8 \times 10^{11} M_\odot$ LMC on first infall, suggesting that the LMC is at least 15% as massive as the MW. Our findings highlight the dramatic disequilibrium of the MW outskirts, and will enable more accurate measurements of the total mass of our Galaxy., Comment: 25 pages, 15 figures. Submitted to ApJ

In recent years, the study of the Milky Way has significantly advanced due to extensive spectroscopic surveys of its stars, complemented by astroseismic and astrometric data. However, it remains disjoint from recent advancements in understanding the physics of the Galactic interstellar medium (ISM). This paper introduces a new model for the chemical evolution of the Milky Way that can be constrained on stellar data, because it combines a state-of-the-art ISM model with a Milky Way stellar disc model. Utilizing a dataset of red clump stars from APOGEE, known for their precise ages and metallicities, we concentrate on the last 6 billion years -- a period marked by Milky Way's secular evolution. We examine the oxygen abundance in the low-$\alpha$ disc stars relative to their ages and birth radii, validating or constraining critical ISM parameters that remain largely unexplored in extragalactic observations. The models that successfully reproduce the radius -- metallicity distribution and the age -- metallicity distribution of stars without violating existing ISM observations indicate a need for modest differential oxygen enrichment in Galactic outflows, meaning that the oxygen abundance of outflows is higher than the local ISM abundance, irrespective of outflow mass loading. The models also suggest somewhat elevated ISM gas velocity dispersion levels over the past 6 billion years compared to galaxies of similar mass. The extra turbulence necessary could result from energy from gas accretion onto the Galaxy, supernovae clustering in the ISM, or increased star formation efficiency per freefall time. This work provides a novel approach to constraining the Galactic ISM and outflows, leveraging the detailed insights available from contemporary Milky Way surveys., Comment: 20 pages, 9 figures. Accepted by MNRAS

Analyzing time series of fluxes from stars, known as stellar light curves, can reveal valuable information about stellar properties. However, most current methods rely on extracting summary statistics, and studies using deep learning have been limited to supervised approaches. In this research, we investigate the scaling law properties that emerge when learning from astronomical time series data using self-supervised techniques. By employing the GPT-2 architecture, we show the learned representation improves as the number of parameters increases from $10^4$ to $10^9$, with no signs of performance plateauing. We demonstrate that a self-supervised Transformer model achieves 3-10 times the sample efficiency compared to the state-of-the-art supervised learning model when inferring the surface gravity of stars as a downstream task. Our research lays the groundwork for analyzing stellar light curves by examining them through large-scale auto-regressive generative models., Comment: 11 pages, 5 figures, ICML 2024 AI4Science workshop

Recent studies using four-point correlations suggest a parity violation in the galaxy distribution, though the significance of these detections is sensitive to the choice of simulation used to model the noise properties of the galaxy distribution. In a recent paper, we introduce an unsupervised learning approach which offers an alternative method that avoids the dependence on mock catalogs, by learning parity violation directly from observational data. However, the Convolutional Neural Network (CNN) model utilized by our previous unsupervised approach struggles to extend to more realistic scenarios where data is limited. We propose a novel method, the Neural Field Scattering Transform (NFST), which enhances the Wavelet Scattering Transform (WST) technique by adding trainable filters, parameterized as a neural field. We first tune the NFST model to detect parity violation in a simplified dataset, then compare its performance against WST and CNN benchmarks across varied training set sizes. We find the NFST can detect parity violation with $4\times$ less data than the CNN and $32\times$ less than the WST. Furthermore, in cases with limited data the NFST can detect parity violation with up to $6\sigma$ confidence, where the WST and CNN fail to make any detection. We identify that the added flexibility of the NFST, and particularly the ability to learn asymmetric filters, as well as the specific symmetries built into the NFST architecture, contribute to its improved performance over the benchmark models. We further demonstrate that the NFST is readily interpretable, which is valuable for physical applications such as the detection of parity violation.

Young {\alpha}-rich (YAR) stars within the old Galactic thick disk exhibit a dual characteristic of relative youth determined with asteroseismology and abundance enhancement in {\alpha} elements measured from high-resolution spectroscopy. The youth origin of YAR stars has been proposed to be binary evolution via mass transfer or stellar mergers. If that is the case, YAR stars should spin rapidly and thus be magnetically active, because they are mass and angular momentum gainers. In this study, to seek this binary footprint we select YAR stars on the main-sequence turn-off or the subgiant branch (MSTO-SGB) from APOGEE DR17, whose ages and projected rotation velocities (vsini) can be precisely measured. With APOGEE vsini and LAMOST spectra, we find that YAR stars are indeed fast rotators and magnetically active. In addition, we observe low [C/N] ratios and high Gaia RUWE in some YAR stars, suggesting that these MSTO-SGB stars probably have experienced mass transfer from red-giant companions. Our findings underscore that magnetic activity can serve as a valuable tool for probing the binary evolution for other chemically peculiar stars, such as red giants with lithium anomalies and carbon-enhanced metal-poor stars., Comment: 7 pages, 6 figures. Accepted for publication in MNRAS

The gravitational interactions between the LMC and the Milky Way cause dynamical perturbations in the MW halo, leading to biased distributions of stellar density and kinematics. We run 50 high-resolution N-body simulations exploring varying masses and halo shapes of the MW and LMC to study the evolution of LMC-induced perturbations. By measuring mean velocities of simulated halo stars, we identify a discontinuity between the first-infall and second-passage scenarios of the LMC's orbital history. In the first infall, the Galactocentric latitudinal velocity hovers around 16 km/s for stars at 50-100 kpc, while it subsides to about 8 km/s in the second-passage scenario. We demonstrate that this reduced perturbation magnitude in the second-passage scenario is mainly due to the short dynamical times of the Galactic inner halo and the lower velocity of the LMC during its second infall into the MW. Using a subset of $\sim 1100$ RR Lyrae stars located in the outer halo ($50 \leq R_{\mathrm{GC}} < 100$ kpc) with precise distance estimates from Gaia, we find the mean latitudinal velocity ($v_{b}$) in the Galactocentric frame to be $\langle v_{b} \rangle = 18.1 \pm 4.1$ km/s. The observation supports the first-infall scenario with a massive LMC ($\sim 2.1 \times 10^{11} \mathrm{M}_{\odot}$) at infall, an oblate MW halo with a virial mass $M_{200} < 1.4 \times 10^{12} \mathrm{M}_{\odot}$ and a flattening parameter $q > 0.7$. Our study indicates that LMC-induced kinematic disturbances can reveal its orbital history and key characteristics, as well as those of the MW. This approach shows promise in helping determine fundamental parameters of both galaxies., Comment: 21 pages, 14 figures, accepted for publication in MNRA

Reconstructing the intrinsic Ly$\alpha$ line flux from high-$z$ QSOs can place constraints on the neutral hydrogen content of the intergalactic medium during reionisation. There are now $\gtrsim10$ different Ly$\alpha$ reconstruction pipelines using different methodologies to predict the Ly$\alpha$ line flux from correlations with the spectral information redward of Ly$\alpha$. However, there have been few attempts to directly compare the performance of these pipelines. Therefore, we devised a blind QSO challenge to compare these reconstruction pipelines on a uniform set of objects. Each author was provided de-identified, observed rest-frame QSO spectra with spectral information only redward of 1260\AA\ rest-frame to ensure unbiased reconstruction. We constructed two samples of 30 QSOs, from X-Shooter and SDSS both spanning $3.5

The detection of the 21-cm signal at $z\gtrsim6$ will reveal insights into the properties of the first galaxies responsible for driving reionisation. To extract this information, we perform parameter inference which requires embedding 3D simulations of the 21-cm signal within a Bayesian inference pipeline. Presently, when performing inference we must choose which sources of uncertainty to sample and which to hold fixed. Since the astrophysics of galaxies are much more uncertain than those of the underlying halo-mass function (HMF), we usually parameterise and model the former while fixing the latter. However, in doing so we may bias our inference of the properties of these first galaxies. In this work, we explore the consequences of assuming an incorrect choice of HMF and quantify the relative biases in our inferred astrophysical model parameters when considering the wrong HMF. We then relax this assumption by constructing a generalised five parameter model for the HMF and simultaneously recover these parameters along with our underlying astrophysical model. For this analysis, we use 21cmFAST and perform Simulation-Based Inference by applying marginal neural ratio estimation to learn the likelihood-to-evidence ratio using Swyft. Using a mock 1000 hour observation of the 21-cm power spectrum from the forthcoming Square Kilometre Array, conservatively assuming foreground wedge avoidance, we find assuming the incorrect HMF can bias the recovered astrophysical parameters by up to $\sim3-4\sigma$ even when including independent information from observed luminosity functions. When considering our generalised HMF model, we recover constraints on our astrophysical parameters with a factor of $\sim2-4$ larger marginalised uncertainties. Importantly, these constraints are unbiased, agnostic to the underlying HMF and therefore more conservative., Comment: 27 pages, 14 figures, 3 tables and 3 appendices. Submitted to MNRAS, comments welcome

Enlightening our understanding of the first galaxies responsible for driving reionisation requires detecting the 21-cm signal from neutral hydrogen. Interpreting the wealth of information embedded in this signal requires Bayesian inference. Parameter inference from the 21-cm signal is primarily restricted to the spherically averaged power spectrum (1D PS) owing to its relatively straightforward derivation of an analytic likelihood function enabling traditional Monte-Carlo Markov-Chain (MCMC) approaches. However, in recent years, simulation-based inference (SBI) has become feasible which removes the necessity of having an analytic likelihood, enabling more complex summary statistics of the 21-cm signal to be used for Bayesian inference. In this work, we use SBI, specifically marginal neural ratio estimation to learn the likelihood-to-evidence ratio with Swyft, to explore parameter inference using the cylindrically averaged 2D PS. Since the 21-cm signal is anisotropic, the 2D PS should yield more constraining information compared to the 1D PS which isotropically averages the signal. For this, we consider a mock 1000 hr observation of the 21-cm signal using the SKA and compare the performance of the 2D PS relative to the 1D PS. Additionally, we explore two separate foreground mitigation strategies, perfect foreground removal and wedge avoidance. We find the 2D PS outperforms the 1D PS by improving the marginalised uncertainties on individual astrophysical parameters by up to $\sim30-40$ per cent irrespective of the foreground mitigation strategy. Primarily, these improvements stem from how the 2D PS distinguishes between the transverse, $k_{\perp}$, and redshift dependent, $k_{\parallel}$ information which enables greater sensitivity to the complex reionisation morphology., Comment: 16 pages, 6 figures and 1 table. Submitted to MNRAS, comments welcome

Stellar chemical compositions can be altered by ingestion of planetary material and/or planet formation which removes refractory material from the proto-stellar disc. These "planet signatures" appear as correlations between elemental abundance differences and the dust condensation temperature. Detecting these planet signatures, however, is challenging due to unknown occurrence rates, small amplitudes, and heterogeneous star samples with large differences in stellar ages, and therefore stars born together (i.e., co-natal) with identical compositions can facilitate such detections. While previous spectroscopic studies were limited to small number of binary stars, the Gaia satellite provides new opportunities for detecting stellar chemical signatures of planets among co-moving pairs of stars confirmed to be co-natal. Here we report high-precision chemical abundances for a homogeneous sample of 91 co-natal pairs of stars with a well-defined selection function and identify at least seven new instances of planetary ingestion, corresponding to an occurrence rate of 8%. An independent Bayesian indicator is deployed, which can effectively disentangle the planet signatures from other factors, such as random abundance variation and atomic diffusion. Our study provides new evidence of planet signatures and facilitates a deeper understanding of the star-planet-chemistry connection by providing new observational constraints on the mechanisms of planet engulfment, formation and evolution., Comment: 29 pages, 11 figures. Author's submitted version before final edits. Published in Nature on March 21, 2024: https://www.nature.com/articles/s41586-024-07091-y

The Wide-field Spectroscopic Telescope (WST) is proposed as a new facility dedicated to the efficient delivery of spectroscopic surveys. This white paper summarises the initial concept as well as the corresponding science cases. WST will feature simultaneous operation of a large field-of-view (3 sq. degree), a high multiplex (20,000) multi-object spectrograph (MOS) and a giant 3x3 sq. arcmin integral field spectrograph (IFS). In scientific capability these requirements place WST far ahead of existing and planned facilities. Given the current investment in deep imaging surveys and noting the diagnostic power of spectroscopy, WST will fill a crucial gap in astronomical capability and work synergistically with future ground and space-based facilities. This white paper shows that WST can address outstanding scientific questions in the areas of cosmology; galaxy assembly, evolution, and enrichment, including our own Milky Way; origin of stars and planets; time domain and multi-messenger astrophysics. WST's uniquely rich dataset will deliver unforeseen discoveries in many of these areas. The WST Science Team (already including more than 500 scientists worldwide) is open to the all astronomical community. To register in the WST Science Team please visit https://www.wstelescope.com/for-scientists/participate, Comment: 194 pages, 66 figures. Comments are welcome (wstelescope@gmail.com)

We study the cosmological evolution of the two-point correlation functions of galactic gas-phase metal distributions using the 28 simulated galaxies from the Auriga Project. Using mock observations of the $z = 0$ snapshots to mimic our past work, we show that the correlation functions of the simulated mock observations are well matched to the correlation functions measured from local galaxy surveys. This comparison suggests that the simulations capture the processes important for determining metal correlation lengths, the key parameter in metallicity correlation functions. We investigate the evolution of metallicity correlations over cosmic time using the true simulation data, showing that individual galaxies undergo no significant systematic evolution in their metal correlation functions from $z\sim 3$ to today. In addition, the fluctuations in metal correlation length are correlated with but lag ahead fluctuations in star formation rate. This suggests that re-arrangement of metals within galaxies occurs at a higher cadence than star formation activity, and is more sensitive to the changes of environment, such as galaxy mergers, gas inflows / outflows, and fly-bys., Comment: 12 pages, 10 figures, 1 table, accepted for publication in MNRAS

Stellar abundances for a large number of stars are key information for the study of Galactic formation history. Large spectroscopic surveys such as DESI and LAMOST take median-to-low resolution ($R\lesssim5000$) spectra in the full optical wavelength range for millions of stars. However, line blending effect in these spectra causes great challenges for the elemental abundances determination. Here we employ the DD-PAYNE, a data-driven method regularised by differential spectra from stellar physical models, to the DESI EDR spectra for stellar abundance determination. Our implementation delivers 15 labels, including effective temperature $T_{\rm eff}$, surface gravity $\log g$, microturbulence velocity $v_{\rm mic}$, and abundances for 12 individual elements, namely C, N, O, Mg, Al, Si, Ca, Ti, Cr, Mn, Fe, Ni. Given a spectral signal-to-noise ratio of 100 per pixel, internal precision of the label estimates are about 20 K for $T_{\rm eff}$, 0.05 dex for $\log~g$, and 0.05 dex for most elemental abundances. These results are agree with theoretical limits from the Cr\'amer-Rao bound calculation within a factor of two. The Gaia-Enceladus-Sausage that contributes the majority of the accreted halo stars are discernible from the disk and in-situ halo populations in the resultant [Mg/Fe]-[Fe/H] and [Al/Fe]-[Fe/H] abundance spaces. We also provide distance and orbital parameters for the sample stars, which spread a distance out to $\sim$100 kpc. The DESI sample has a significant higher fraction of distant (or metal-poor) stars than other existed spectroscopic surveys, making it a powerful data set to study the Galactic outskirts. The catalog is publicly available., Comment: 18 pages, 12 figures. Submitted to ApJS

We present a novel, deep-learning based method -- dubbed Galactic-Seismology Substructures and Streams Hunter, or GS$^{3}$ Hunter for short, to search for substructures and streams in stellar kinematics data. GS$^{3}$ Hunter relies on a combined application of Siamese Neural Networks to transform the phase space information and the K-means algorithm for the clustering. As a validation test, we apply GS$^{3}$ Hunter to a subset of the Feedback in Realistic Environments (FIRE) cosmological simulations. The stellar streams and substructures thus identified are in good agreement with corresponding results reported earlier by the FIRE team. In the same vein, we apply our method to a subset of local halo stars from the Gaia Early Data Release 3 and GALAH DR3 datasets, and recover several, previously known dynamical groups, such as Thamnos 1+2, Hot Thick Disk, ED-1, L-RL3, Helmi 1+2, and Gaia-Sausage-Enceladus, Sequoia, VRM, Cronus, Nereus. Finally, we apply our method without fine-tuning to a subset of K-giant stars located in the inner halo region, obtained from the LAMOST Data Release 5 (DR5) dataset. We recover three, previously known structures (Sagittarius, Hercules-Aquila Cloud, and the Virgo Overdensity), but we also discover a number of new substructures. We anticipate that GS$^{3}$ Hunter will become a useful tool for the community dedicated to the search of stellar streams and structures in the Milky Way (MW) and the Local group, thus helping advance our understanding of the stellar inner and outer halos, and of the assembly and tidal stripping history in and around the MW., Comment: 41 pages, 34 figures, 3 tables. Accepted for Publication in the Astrophysical Journal

We explore the potential of enhancing LLM performance in astronomy-focused question-answering through targeted, continual pre-training. By employing a compact 7B-parameter LLaMA-2 model and focusing exclusively on a curated set of astronomy corpora -- comprising abstracts, introductions, and conclusions -- we achieve notable improvements in specialized topic comprehension. While general LLMs like GPT-4 excel in broader question-answering scenarios due to superior reasoning capabilities, our findings suggest that continual pre-training with limited resources can still enhance model performance on specialized topics. Additionally, we present an extension of AstroLLaMA: the fine-tuning of the 7B LLaMA model on a domain-specific conversational dataset, culminating in the release of the chat-enabled AstroLLaMA for community use. Comprehensive quantitative benchmarking is currently in progress and will be detailed in an upcoming full paper. The model, AstroLLaMA-Chat, is now available at https://huggingface.co/universeTBD, providing the first open-source conversational AI tool tailored for the astronomy community., Comment: 4 pages, 1 figure, model is available at https://huggingface.co/universeTBD, published in RNAAS

Recent measurements of the $4$-point correlation functions (4PCF) from spectroscopic surveys provide evidence for parity-violations in the large-scale structure of the Universe. If physical in origin, this could point to exotic physics during the epoch of inflation. However, searching for parity-violations in the 4PCF signal relies on a large suite of simulations to perform a rank test, or an accurate model of the 4PCF covariance to claim a detection, and this approach is incapable of extracting parity information from the higher-order $N$-point functions. In this work we present an unsupervised method which overcomes these issues, before demonstrating the approach is capable of detecting parity-violations in a few toy models using convolutional neural networks. This technique is complementary to the 4-point method and could be used to discover parity-violations in several upcoming surveys including DESI, Euclid and Roman., Comment: 8 pages. 5 figures. PRD Submitted

The High-Resolution Multi-Object Spectrograph (HRMOS) is a facility instrument that we plan to propose for the Very Large Telescope (VLT) of the European Southern Observatory (ESO), following the initial presentation at the VLT 2030 workshop held at ESO in June 2019. HRMOS provides a combination of capabilities that are essential to carry out breakthrough science across a broad range of active research areas from stellar astrophysics and exoplanet studies to Galactic and Local Group archaeology. HRMOS fills a gap in capabilities amongst the landscape of future instrumentation planned for the next decade. The key characteristics of HRMOS will be high spectral resolution (R = 60000 - 80000) combined with multi-object (20-100) capabilities and long term stability that will provide excellent radial velocity precision and accuracy (10m/s). Initial designs predict that a SNR~100 will be achievable in about one hour for a star with mag(AB) = 15, while with the same exposure time a SNR~ 30 will be reached for a star with mag(AB) = 17. The combination of high resolution and multiplexing with wavelength coverage extending to relatively blue wavelengths (down to 380\,nm), makes HRMOS a spectrograph that will push the boundaries of our knowledge and that is envisioned as a workhorse instrument in the future. The science cases presented in this White Paper include topics and ideas developed by the Core Science Team with the contributions from the astronomical community, also through the wide participation in the first HRMOS Workshop (https://indico.ict.inaf.it/event/1547/) that took place in Firenze (Italy) in October 2021., Comment: 88 pages, 39 figures. Comments and expressions of interest are welcome by contacting members of the Core Science Team

Dynamic astrophysical phenomena are predominantly described by differential equations, yet our understanding of these systems is constrained by our incomplete grasp of non-linear physics and scarcity of comprehensive datasets. As such, advancing techniques in solving non-linear inverse problems becomes pivotal to addressing numerous outstanding questions in the field. In particular, modeling hot galactic winds is difficult because of unknown structure for various physical terms, and the lack of \textit{any} kinematic observational data. Additionally, the flow equations contain singularities that lead to numerical instability, making parameter sweeps non-trivial. We leverage differentiable programming, which enables neural networks to be embedded as individual terms within the governing coupled ordinary differential equations (ODEs), and show that this method can adeptly learn hidden physics. We robustly discern the structure of a mass-loading function which captures the physical effects of cloud destruction and entrainment into the hot superwind. Within a supervised learning framework, we formulate our loss function anchored on the astrophysical entropy ($K \propto P/\rho^{5/3}$). Our results demonstrate the efficacy of this approach, even in the absence of kinematic data $v$. We then apply these models to real Chandra X-Ray observations of starburst galaxy M82, providing the first systematic description of mass-loading within the superwind. This work further highlights neural ODEs as a useful discovery tool with mechanistic interpretability in non-linear inverse problems. We make our code public at this GitHub repository (https://github.com/dustindnguyen/2023_NeurIPS_NeuralODEs_M82)., Comment: Accepted at the NeurIPS 2023 workshop on Machine Learning and the Physical Sciences

We present zephyr, a novel method that integrates cutting-edge normalizing flow techniques into a mixture density estimation framework, enabling the effective use of heterogeneous training data for photometric redshift inference. Compared to previous methods, zephyr demonstrates enhanced robustness for both point estimation and distribution reconstruction by leveraging normalizing flows for density estimation and incorporating careful uncertainty quantification. Moreover, zephyr offers unique interpretability by explicitly disentangling contributions from multi-source training data, which can facilitate future weak lensing analysis by providing an additional quality assessment. As probabilistic generative deep learning techniques gain increasing prominence in astronomy, zephyr should become an inspiration for handling heterogeneous training data while remaining interpretable and robustly accounting for observational uncertainties., Comment: 10 pages, 5 figures, accepted to NeurIPS 2023 workshop on Machine Learning and the Physical Sciences

Solar-like oscillations have been detected in thousands of stars thanks to modern space missions. These oscillations have been used to measure stellar masses and ages, which have been widely applied in Galactic archaeology. One of the pillars of such applications is the $\nu_{\max}$ scaling relation: the frequency of maximum power $\nu_{\max}$, assumed to be proportional to the acoustic cut-off frequency, $\nu_{\rm ac}$, scales with effective temperature and surface gravity. However, the theoretical basis of the $\nu_{\max}$ scaling relation is uncertain, and there is an ongoing debate about whether it can be applied to metal-poor stars. We investigate the metallicity dependence of the $\nu_{\max}$ scaling relation by carrying out 3D near-surface convection simulations for solar-type stars with [Fe/H] between -3 and 0.5 dex. Firstly, we found a negative correlation between $\nu_{\rm ac}$ and metallicity from the 3D models. This is in tension with the positive correlation identified by studies using 1D models. Secondly, we estimated theoretical $\nu_{\max}$ values using velocity amplitudes determined from first principles, by quantifying the mode excitation and damping rates with methods validated in our previous works. We found that at solar effective temperature and surface gravity, $\nu_{\max}$ does not show correlation with metallicity. This study opens an exciting prospect of testing the asteroseismic scaling relations against realistic 3D hydrodynamical stellar models., Comment: 14 pages, 7 figures, accepted for publication in ApJ

Stellar light curves contain valuable information about oscillations and granulation, offering insights into stars' internal structures and evolutionary states. Traditional asteroseismic techniques, primarily focused on power spectral analysis, often overlook the crucial phase information in these light curves. Addressing this gap, recent machine learning applications, particularly those using Convolutional Neural Networks (CNNs), have made strides in inferring stellar properties from light curves. However, CNNs are limited by their localized feature extraction capabilities. In response, we introduce $\textit{Astroconformer}$, a Transformer-based deep learning framework, specifically designed to capture long-range dependencies in stellar light curves. Our empirical analysis centers on estimating surface gravity ($\log g$), using a dataset derived from single-quarter Kepler light curves with $\log g$ values ranging from 0.2 to 4.4. $\textit{Astroconformer}$ demonstrates superior performance, achieving a root-mean-square-error (RMSE) of 0.017 dex at $\log g\approx3$ in data-rich regimes and up to 0.1 dex in sparser areas. This performance surpasses both K-nearest neighbor models and advanced CNNs. Ablation studies highlight the influence of receptive field size on model effectiveness, with larger fields correlating to improved results. $\textit{Astroconformer}$ also excels in extracting $\nu_{\max}$ with high precision. It achieves less than 2% relative median absolute error for 90-day red giant light curves. Notably, the error remains under 3% for 30-day light curves, whose oscillations are undetectable by a conventional pipeline in 30% cases. Furthermore, the attention mechanisms in $\textit{Astroconformer}$ align closely with the characteristics of stellar oscillations and granulation observed in light curves., Comment: 15 pages, 10 figures, Accepted by MNRAS

Large language models excel in many human-language tasks but often falter in highly specialized domains like scholarly astronomy. To bridge this gap, we introduce AstroLLaMA, a 7-billion-parameter model fine-tuned from LLaMA-2 using over 300,000 astronomy abstracts from arXiv. Optimized for traditional causal language modeling, AstroLLaMA achieves a 30% lower perplexity than Llama-2, showing marked domain adaptation. Our model generates more insightful and scientifically relevant text completions and embedding extraction than state-of-the-arts foundation models despite having significantly fewer parameters. AstroLLaMA serves as a robust, domain-specific model with broad fine-tuning potential. Its public release aims to spur astronomy-focused research, including automatic paper summarization and conversational agent development., Comment: 6 pages, 3 figures, submitted to IJCNLP-AACL 2023. Comments are welcome. The model can be found on Hugging Face - https://huggingface.co/universeTBD/astrollama

We conduct a line-by-line differential analysis of a sample of 125 co-moving pairs of stars (dwarfs and subgiants near solar metallicity). We obtain high precision stellar parameters with average uncertainties in effective temperature, surface gravity and metallicity of 16.5 K, 0.033 dex and 0.014 dex, respectively. We classify the co-moving pairs of stars into two groups, chemically homogeneous (conatal; |Delta[Fe/H]| $\le$ 0.04 dex) and inhomogeneous (non-conatal), and examine the fraction of chemically homogeneous pairs as a function of separation and effective temperature. The four main conclusions from this study are: (1) A spatial separation of \ds = 10$^6$ AU is an approximate boundary between homogeneous and inhomogeneous pairs of stars, and we restrict our conclusions to only consider the 91 pairs with \ds $\le$ 10$^6$ AU; (2) There is no trend between velocity separation and the fraction of chemically homogeneous pairs in the range \dv $\le$ 4 \kms; (3) We confirm that the fraction of chemically inhomogeneous pairs increases with increasing \teff\ and the trend matches a toy model of that expected from planet ingestion; (4) Atomic diffusion is not the main cause of the chemical inhomogeneity. A major outcome from this study is a sample of 56 bright co-moving pairs of stars with chemical abundance differences $\leq$ 0.02 dex (5\%) which is a level of chemical homogeneity comparable to that of the Hyades open cluster. These important objects can be used, in conjunction with star clusters and the \gaia\ ``benchmark'' stars, to calibrate stellar abundances from large-scale spectroscopic surveys., Comment: MNRAS in press (see source file for full versions of long tables)

The proto-Milky Way epoch forms the earliest stars in our Galaxy and sets the initial conditions for subsequent disk formation. Recent observations from APOGEE and H3 surveys showed that the [$\alpha$/Fe] ratio slowly declined between [Fe/H] $=-3$ and $-1.3$ until it reached the lowest value ($\sim 0.25$) among the selected in situ metal-poor stars that most likely formed during the proto-Galaxy epoch. [$\alpha$/Fe] rose to meet the traditional high value commonly associated with the thick disk population at [Fe/H] $=-1$. It was suggested that the rise in [$\alpha$/Fe] could be caused by an increase in the star formation efficiency (SFE), known as the "simmering" phase scenario. However, gas inflow also plays a vital role in shaping the star formation history and chemical evolution of galaxies. We investigate this unexpected [$\alpha$/Fe]-rise with a statistical experiment involving a galactic chemical evolution (GCE). Our model has five free parameters: the mass of the initial reservoir of the cold interstellar medium (ISM) at birth, the frequency of Type Ia supernovae (SNe Ia), the cooling timescale of the warm ISM, the SFE, and the inflow rate of fresh gas. The last two free parameters were allowed to change after [$\alpha$/Fe] reached its lowest value, dividing the proto-Galaxy epoch into two phases. We find that the rise in [$\alpha$/Fe] is caused by a large inflow of fresh gas and conclude that the [$\alpha$/Fe]-rise is a signature of the cold mode accretion whose materials formed the prototype Milky Way preceding disk formation. Although the SFE is essential in regulating the chemical evolution, it does not necessarily increase to facilitate the [$\alpha$/Fe]-rise., Comment: 15 pages, 10 figures

In this paper, we tackle the emerging challenge of unintended harmful content generation in Large Language Models (LLMs) with a novel dual-stage optimisation technique using adversarial fine-tuning. Our two-pronged approach employs an adversarial model, fine-tuned to generate potentially harmful prompts, and a judge model, iteratively optimised to discern these prompts. In this adversarial cycle, the two models seek to outperform each other in the prompting phase, generating a dataset of rich examples which are then used for fine-tuning. This iterative application of prompting and fine-tuning allows continuous refinement and improved performance. The performance of our approach is evaluated through classification accuracy on a dataset consisting of problematic prompts not detected by GPT-4, as well as a selection of contentious but unproblematic prompts. We show considerable increase in classification accuracy of the judge model on this challenging dataset as it undergoes the optimisation process. Furthermore, we show that a rudimentary model \texttt{ada} can achieve 13\% higher accuracy on the hold-out test set than GPT-4 after only a few rounds of this process, and that this fine-tuning improves performance in parallel tasks such as toxic comment identification.

We present the first detailed chemical-abundance analysis of stars from the dwarf-galaxy stellar stream Wukong/LMS-1 covering a wide metallicity range ($-3.5 < \rm[Fe/H] \lesssim -1.3$). We find abundance patterns that are effectively indistinguishable from the bulk of Indus and Jhelum, a pair of smaller stellar streams proposed to be dynamically associated with Wukong/LMS-1. We confirmed a carbon-enhanced metal-poor star ($\rm[C/Fe] > +0.7$ and $\rm[Fe/H] \sim -2.9$) in Wukong/LMS-1 with strong enhancements in Sr, Y, and Zr, which is peculiar given its solar-level [Ba/Fe]. Wukong/LMS-1 stars have high abundances of $\alpha$ elements up to $\rm[Fe/H] \gtrsim -2$, which is expected for relatively massive dwarfs. Towards the high-metallicity end, Wukong/LMS-1 becomes $\alpha$-poor, revealing that it probably experienced fairly standard chemical evolution. We identified a pair of N- and Na-rich stars in Wukong/LMS-1, reminiscent of multiple populations in globular clusters. This indicates that this dwarf galaxy contained at least one globular cluster that was completely disrupted in addition to two intact ones previously known to be associated with Wukong/LMS-1, which is possibly connected to similar evidence found in Indus. From these $\geq$3 globular clusters, we estimate the total mass of Wukong/LMS-1 to be ${\approx}10^{10} M_\odot$, representing ${\sim}1$% of the present-day Milky Way. Finally, the [Eu/Mg] ratio in Wukong/LMS-1 continuously increases with metallicity, making this the first example of a dwarf galaxy where the production of $r$-process elements is clearly dominated by delayed sources, presumably neutron-star mergers., Comment: Accepted to MNRAS. New version fixes abundance uncertainties, which significantly affects elements like Al and N. Use new version of abundance files for correct error bars

We infer the dynamical masses of stars across the Hertzsprung-Russell (H-R) diagram using wide binaries from the Gaia survey. Gaia's high-precision astrometry measures the wide binaries' orbital motion, which contains the mass information. Using wide binaries as the training sample, we measure the mass of stars across the two-dimensional H-R diagram using the combination of statistical inference and neural networks. Our results provide the dynamical mass measurements for main-sequence stars from 0.1 to 2 M$_\odot$, unresolved binaries and unresolved triples on the main sequence, and the mean masses of giants and white dwarfs. Two regions in the H-R diagram show interesting behaviors in mass, where one of them is pre-main-sequence stars, and the other one may be related to close compact object companions like M dwarf-white dwarf binaries. These mass measurements depend solely on Newtonian dynamics, providing independent constraints on stellar evolutionary models and the occurrence rate of compact objects., Comment: Fig. 5 and Fig. 12 are the key results. Submitted to MNRAS. Comments are welcome!

Large Language Models (LLMs) hold immense potential to generate synthetic data of high quality and utility, which has numerous applications from downstream model training to practical data utilisation. However, contemporary models, despite their impressive capacities, consistently struggle to produce both coherent and diverse data. To address the coherency issue, we introduce contrastive expert guidance, where the difference between the logit distributions of fine-tuned and base language models is emphasised to ensure domain adherence. In order to ensure diversity, we utilise existing real and synthetic examples as negative prompts to the model. We deem this dual-pronged approach to logit reshaping as STEER: Semantic Text Enhancement via Embedding Repositioning. STEER operates at inference-time and systematically guides the LLMs to strike a balance between adherence to the data distribution (ensuring semantic fidelity) and deviation from prior synthetic examples or existing real datasets (ensuring diversity and authenticity). This delicate balancing act is achieved by dynamically moving towards or away from chosen representations in the latent space. STEER demonstrates improved performance over previous synthetic data generation techniques, exhibiting better balance between data diversity and coherency across three distinct tasks: hypothesis generation, toxic and non-toxic comment generation, and commonsense reasoning task generation. We demonstrate how STEER allows for fine-tuned control over the diversity-coherency trade-off via its hyperparameters, highlighting its versatility.

Generative adversarial networks (GANs) are frequently utilized in astronomy to construct an emulator of numerical simulations. Nevertheless, training GANs can prove to be a precarious task, as they are prone to instability and often lead to mode collapse problems. Conversely, the diffusion model also has the ability to generate high-quality data without adversarial training. It has shown superiority over GANs with regard to several natural image datasets. In this study, we undertake a quantitative comparison between the denoising diffusion probabilistic model (DDPM) and StyleGAN2 (one of the most robust types of GANs) via a set of robust summary statistics from scattering transform. In particular, we utilize both models to generate the images of 21 cm brightness temperature mapping, as a case study, conditionally based on astrophysical parameters that govern the process of cosmic reionization. Using our new Fr\'echet Scattering Distance (FSD) as the evaluation metric to quantitatively compare the sample distribution between generative models and simulations, we demonstrate that DDPM outperforms StyleGAN2 on varied sizes of training sets. Through Fisher forecasts, we demonstrate that on our datasets, StyleGAN2 exhibits mode collapses in varied ways, while DDPM yields a more robust generation. We also explore the role of classifier-free guidance in DDPM and show the preference for a non-zero guidance scale only when the training data is limited. Our findings indicate that the diffusion model presents a promising alternative to GANs in the generation of accurate images. These images can subsequently provide reliable parameter constraints, particularly in the realm of astrophysics., Comment: 14 pages, 10 figures, 1 table. Accepted for publication in MNRAS. Comments welcome

The Magellanic Stream (MS) - an enormous ribbon of gas spanning $140^\circ$ of the southern sky trailing the Magellanic Clouds - has been exquisitely mapped in the five decades since its discovery. However, despite concerted efforts, no stellar counterpart to the MS has been conclusively identified. This stellar stream would reveal the distance and 6D kinematics of the MS, constraining its formation and the past orbital history of the Clouds. We have been conducting a spectroscopic survey of the most distant and luminous red giant stars in the Galactic outskirts. From this dataset, we have discovered a prominent population of 13 stars matching the extreme angular momentum of the Clouds, spanning up to $100^\circ$ along the MS at distances of $60-120$ kpc. Furthermore, these kinemetically-selected stars lie along a [$\alpha$/Fe]-deficient track in chemical space from $-2.5 < \mathrm{[Fe/H]} < -0.5$, consistent with their formation in the Clouds themselves. We identify these stars as high-confidence members of the Magellanic Stellar Stream. Half of these stars are metal-rich and closely follow the gaseous MS, whereas the other half are more scattered and metal-poor. We argue that the metal-rich stream is the recently-formed tidal counterpart to the MS, and speculate that the metal-poor population was thrown out of the SMC outskirts during an earlier interaction between the Clouds. The Magellanic Stellar Stream provides a strong set of constraints - distances, 6D kinematics, and birth locations - that will guide future simulations towards unveiling the detailed history of the Clouds., Comment: 21 pages, 12 figures. Submitted to ApJ

Astrophysical explorations are underpinned by large-scale stellar spectroscopy surveys, necessitating a paradigm shift in spectral fitting techniques. Our study proposes three enhancements to transcend the limitations of the current spectral emulation models. We implement an attention-based emulator, adept at unveiling long-range information between wavelength pixels. We leverage a domain-specific fine-tuning strategy where the model is pre-trained on spectra with fixed stellar parameters and variable elemental abundances, followed by fine-tuning on the entire domain. Moreover, by treating wavelength as an autonomous model parameter, akin to neural radiance fields, the model can generate spectra on any wavelength grid. In the case with a training set of O(1000), our approach exceeds current leading methods by a factor of 5-10 across all metrics., Comment: 7 pages, 3 figures, accepted to ICML 2023 Workshop on Machine Learning for Astrophysics

Stars' chemical signatures provide invaluable insights into stellar cluster formation. This study utilized the Weisfeiler-Lehman (WL) Graph Kernel to examine a 15-dimensional elemental abundance space. Through simulating chemical distributions using normalizing flows, the effectiveness of our algorithm was affirmed. The results highlight the capability of the WL algorithm, coupled with Gaussian Process Regression, to identify patterns within elemental abundance point clouds correlated with various cluster mass functions. Notably, the WL algorithm exhibits superior interpretability, efficacy and robustness compared to deep sets and graph convolutional neural networks and enables optimal training with significantly fewer simulations (O(10)), a reduction of at least two orders of magnitude relative to graph neural networks., Comment: 7 pages, 4 figures, accepted to the ICML 2023 Machine Learning for Astrophysics workshop

This study investigates the application of Large Language Models (LLMs), specifically GPT-4, within Astronomy. We employ in-context prompting, supplying the model with up to 1000 papers from the NASA Astrophysics Data System, to explore the extent to which performance can be improved by immersing the model in domain-specific literature. Our findings point towards a substantial boost in hypothesis generation when using in-context prompting, a benefit that is further accentuated by adversarial prompting. We illustrate how adversarial prompting empowers GPT-4 to extract essential details from a vast knowledge base to produce meaningful hypotheses, signaling an innovative step towards employing LLMs for scientific research in Astronomy., Comment: 8 pages, 3 figures, accepted to ICML ML4Astro Workshop. Comments and suggestions are welcome

We present initial results from our JWST NIRSpec program to study the $\alpha$-abundances in the M31 disk. The Milky Way has two chemically-defined disks, the low-$\alpha$ and high-$\alpha$ disks, which are closely related to the thin and thick disks, respectively. The origin of the two populations and the $\alpha$-bimodality between them is not entirely clear, although there are now several models that can reproduce the observed features. To help constrain the models and discern the origin, we have undertaken a study of the chemical abundances of the M31 disk using JWST NIRSpec, in order to determine whether stars in M31's disk also show an $\alpha$-abundance bimodality. Approximately 100 stars were observed in our single NIRSpec field at a projected distance of 18 kpc from the M31 center. The 1-D extracted spectra have an average signal-to-noise ratio of 85 leading to statistical metallicity precision of 0.016 dex, $\alpha$-abundance precision of 0.012 dex, and a radial velocity precision 8 km/s. The initial results indicate that, in contrast to the Milky Way, there is no $\alpha$-bimodality in the M31 disk, and no low-$\alpha$ sequence. The entire stellar population falls along a single chemical sequence very similar to the MW's high-alpha component which had a high star formation rate. While this is somewhat unexpected, the result is not that surprising based on other studies that found the M31 disk has a larger velocity dispersion than the MW and is dominated by a thick component. M31 has had a more active accretion and merger history than the MW which might explain the chemical differences., Comment: 8 pages, 4 figures, IAU Symposium 377, Early Disk-Galaxy Formation: From JWST to the Milky Way

Stellar ages are key for determining the formation history of the Milky Way, but are difficult to measure precisely. Furthermore, methods that use chemical abundances to infer ages may entangle the intrinsic evolution of stars with the chemodynamical evolution of the Galaxy. In this paper, we present a framework for making probabilistic predictions of stellar ages, and then quantify the contribution of both stellar evolution and Galactic chemical evolution to those predictions using SHAP values. We apply this interpretable prediction framework to both a simulated Milky Way sample containing stars in a variety of evolutionary stages and an APOGEE-mocked sample of red clump stars. We find that in the former case, stellar evolution is the dominant driver for age estimates, while in the latter case, the more restricted evolutionary information causes the model to proxy ages through the chemical evolution model. We show that as a result of the use of non-intrinsic Galactic chemical information, trends estimated with the predicted ages, such as the age-metallicity relation, can deviate from the truth., Comment: 18 pages, 17 figures, submitted to ApJ

Deriving stellar atmospheric parameters and chemical abundances from stellar spectra is crucial for understanding the evolution of the Milky Way. By performing a fitting with MARCS model atmospheric theoretical synthetic spectra combined with a domain-adaptation method, we estimate the fundamental stellar parameters (Teff, log g, [Fe/H], vmic, and vmac) and 11 chemical abundances for 1.38 million FGKM-type stars of the Medium-Resolution Spectroscopic Survey (MRS) from LAMOST-II DR8. The domain-adaptation method, Cycle-StarNet, is employed to reduce the gap between observed and synthetic spectra, and the L-BFGS algorithm is used to search for the best-fit synthetic spectra. By combining the 2MASS photometric survey data, Gaia EDR3 parallax, and MIST isochrones, the surface gravities of the stars are constrained after estimating their bolometric luminosities. The accuracy of Teff, log g, and [Fe/H] can reach 150 K, 0.11 dex, and 0.15 dex, evaluated by the PASTEL catalog, asteroseismic samples, and other spectroscopic surveys. The precision of these parameters and elemental abundances ([C/Fe], [Na/Fe], [Mg/Fe], [Si/Fe], [Ca/Fe], [Ti/Fe], [Cr/Fe], [Mn/Fe], [Co/Fe], [Ni/Fe], and [Cu/Fe]) is assessed by repeated observations and validated by cluster members. For spectra with signal-to-noise (S/N) ratios greater than 10, the precision of the three stellar parameters and elemental abundances can achieve 76 K, 0.014 dex, 0.096 dex, and 0.04-0.15 dex. For spectra with S/N ratios higher than 100, the precision stabilizes at 22 K, 0.006 dex, 0.043 dex, and 0.01-0.06 dex. The full LAMOST MRS stellar properties catalog is available online., Comment: Accepted for publication in ApJS

We demonstrate the potential of the state-of-the-art OpenAI GPT-4 large language model to engage in meaningful interactions with Astronomy papers using in-context prompting. To optimize for efficiency, we employ a distillation technique that effectively reduces the size of the original input paper by 50\%, while maintaining the paragraph structure and overall semantic integrity. We then explore the model's responses using a multi-document context (ten distilled documents). Our findings indicate that GPT-4 excels in the multi-document domain, providing detailed answers contextualized within the framework of related research findings. Our results showcase the potential of large language models for the astronomical community, offering a promising avenue for further exploration, particularly the possibility of utilizing the models for hypothesis generation., Comment: 3 pages, published in RNAAS