arXiv subjects

Yuan-Sen Ting

Publications and source records attributed to Yuan-Sen Ting.

At least 55 records · Page 3Linked to original sources

arXiv2025

SpecCLIP: Aligning and Translating Spectroscopic Measurements for Stars

In recent years, large language models (LLMs) have transformed natural language understanding through vast datasets and large-scale parameterization. Inspired by this success, we present SpecCLIP, a foundation model framework that extends LLM-inspired methodologies to stellar spectral analysis. Stellar spectra, akin to structured language, encode rich physical and chemical information about stars. By training foundation models on large-scale spectral datasets, our goal is to learn robust and informative embeddings that support diverse downstream applications. As a proof of concept, SpecCLIP involves pre-training on two spectral types--LAMOST low-resolution and Gaia XP--followed by contrastive alignment using the CLIP (Contrastive Language-Image Pre-training) framework, adapted to associate spectra from different instruments. This alignment is complemented by auxiliary decoders that preserve spectrum-specific information and enable translation (prediction) between spectral types, with the former achieved by maximizing mutual information between embeddings and input spectra. The result is a cross-spectrum framework enabling intrinsic calibration and flexible applications across instruments. We demonstrate that fine-tuning these models on moderate-sized labeled datasets improves adaptability to tasks such as stellar-parameter estimation and chemical-abundance determination. SpecCLIP also enhances the accuracy and precision of parameter estimates benchmarked against external survey data. Additionally, its similarity search and cross-spectrum prediction capabilities offer potential for anomaly detection. Our results suggest that contrastively trained foundation models enriched with spectrum-aware decoders can advance precision stellar spectroscopy. Our code SpecCLIP is publicly available at https://github.com/Xiaosheng-Zhao/SpecCLIP

astro-ph.IM↗

arXiv2025

Metallicity fluctuation statistics in the interstellar medium and young stars -- II. Elemental cross-correlations and the structure of chemical abundance space

Observations of Milky Way stars by multiplexed spectroscopic instruments and of gas in nearby galaxies using integral field units have made it possible to measure the abundances of multiple elements in both the interstellar medium and the stars that form out of it. These observations have revealed complex correlations between elemental abundances, but thus far there has been no analytic theoretical framework to interpret these data. In this paper we extend the simple stochastically-forced diffusion model of Krumholz & Ting (2018), which has proven successful at explaining the spatial abundance patterns of single elements, to multiple elements, clarifying why elements are correlated and what controls their degree of correlation, and making quantitative predictions for the degree of correlation in both gas and young stars. We show that our results are qualitatively consistent with observed patterns, and point out how application of this theory to measured correlations should enable determination of currently unknown parameters describing r-process nucleosynthesis.

astro-ph.GA↗

arXiv2025

AstroMLab 5: Structured Summaries and Concept Extraction for 400,000 Astrophysics Papers

We present a dataset of 408,590 astrophysics papers from arXiv (astro-ph), spanning 1992 through July 2025. Each paper has been processed through a multi-stage pipeline to produce: (1) structured summaries organized into six semantic sections (Background, Motivation, Methodology, Results, Interpretation, Implication), and (2) concept extraction yielding 9,999 unique concepts with detailed descriptions. The dataset contains 3.8 million paper-concept associations and includes semantic embeddings for all concepts. Comparison with traditional ADS keywords reveals that the concepts provide denser coverage and more uniform distribution, while analysis of embedding space structure demonstrates that concepts are semantically dispersed within papers-enabling discovery through multiple diverse entry points. Concept vocabulary and embeddings are publicly released at https://github.com/tingyuansen/astro-ph_knowledge_graph.

astro-ph.IM↗

arXiv2025

SPICE -- modelling synthetic spectra of stars with non-homogeneous surfaces

In the era of large time-domain spectro-photometric surveys, surface variations such as starspots, chemical inhomogeneities, pulsations, rotational distortions, and binary interactions can now be directly detected and modelled. Accurately interpreting these phenomena requires stellar spectral synthesis frameworks that go beyond the assumption of homogeneous surface properties. Yet most existing tools remain limited by this simplification, hindering their applicability to stars with complex surface structures. To address this need, we present SPICE (SPectral Integration Compiled Engine), an open-source Python package for generating high-resolution spectra and photometry from non-homogeneous stellar surface models. SPICE integrates angle-dependent specific intensities from each surface element, enabling forward modelling of both photometric and spectroscopic variability. Case studies demonstrate applications to spotted stars, Cepheid pulsations, and eclipsing binaries, making SPICE well-suited for analysing current and upcoming survey data. In addition, SPICE can directly import meshes from PHOEBE, enabling the modelling of complex binary configurations beyond these case studies.

astro-ph.SR↗

arXiv2025

Mephisto: Self-Improving Large Language Model-Based Agents for Automated Interpretation of Multi-band Galaxy Observations

Astronomical research has long relied on human expertise to interpret complex data and formulate scientific hypotheses. In this study, we introduce Mephisto -- a multi-agent collaboration framework powered by large language models (LLMs) that emulates human-like reasoning for analyzing multi-band galaxy observations. Mephisto interfaces with the CIGALE codebase (a library of spectral energy distribution, SED, models) to iteratively refine physical models against observational data. It conducts deliberate reasoning via tree search, accumulates knowledge through self-play, and dynamically updates its knowledge base. Validated across diverse galaxy populations -- including the James Webb Space Telescope's recently discovered "Little Red Dot" galaxies -- we show that Mephisto demonstrates proficiency in inferring the physical properties of galaxies from multi-band photometry, positioning it as a promising research copilot for astronomers. Unlike prior black-box machine learning approaches in astronomy, Mephisto offers a transparent, human-aligned reasoning process that integrates seamlessly with existing research practices. This work underscores the possibility of LLM-driven agent-based research for astronomy, establishes a foundation for fully automated, end-to-end artificial intelligence (AI)-powered scientific workflows, and unlocks new avenues for AI-augmented discoveries in astronomy.

astro-ph.IM↗

arXiv2025

Large Language Models Achieve Gold Medal Performance at the International Olympiad on Astronomy & Astrophysics (IOAA)

While task-specific demonstrations show early success in applying large language models (LLMs) to automate some astronomical research tasks, they only provide incomplete views of all necessary capabilities in solving astronomy problems, calling for more thorough understanding of LLMs' strengths and limitations. So far, existing benchmarks and evaluations focus on simple question-answering that primarily tests astronomical knowledge and fails to evaluate the complex reasoning required for real-world research in the discipline. Here, we address this gap by systematically benchmarking five state-of-the-art LLMs on the International Olympiad on Astronomy and Astrophysics (IOAA) exams, which are designed to examine deep conceptual understanding, multi-step derivations, and multimodal analysis. With average scores of 85.6% and 84.2%, Gemini 2.5 Pro and GPT-5 (the two top-performing models) not only achieve gold medal level performance but also rank in the top two among ~200-300 participants in all four IOAA theory exams evaluated (2022-2025). In comparison, results on the data analysis exams show more divergence. GPT-5 still excels in the exams with an 88.5% average score, ranking top 10 among the participants in the four most recent IOAAs, while other models' performances drop to 48-76%. Furthermore, our in-depth error analysis underscores conceptual reasoning, geometric reasoning, and spatial visualization (52-79% accuracy) as consistent weaknesses among all LLMs. Hence, although LLMs approach peak human performance in theory exams, critical gaps must be addressed before they can serve as autonomous research agents in astronomy.

astro-ph.IM↗

arXiv2025

Interpreting deep learning-based stellar mass estimation via causal analysis and mutual information decomposition

End-to-end deep learning models fed with multi-band galaxy images are powerful data-driven tools used to estimate galaxy physical properties in the absence of spectroscopy. However, due to a lack of interpretability and the associational nature of such models, it is difficult to understand how the information that is included in addition to integrated photometry (e.g., morphology) contributes to the estimation task. Improving our understanding in this field would enable further advances into unraveling the physical connections among galaxy properties and optimizing data exploitation. Therefore, our work is aimed at interpreting the deep learning-based estimation of stellar mass via two interpretability techniques: causal analysis and mutual information decomposition. The former reveals the causal paths between multiple variables beyond nondirectional statistical associations, while the latter quantifies the multicomponent contributions (i.e., redundant, unique, and synergistic) of different input data to the stellar mass estimation. Using data from the Sloan Digital Sky Survey (SDSS) and the Wide-field Infrared Survey Explorer (WISE), we obtained meaningful results that provide physical interpretations for image-based models. Our work demonstrates the gains from combining deep learning with interpretability techniques, and holds promise in promoting more data-driven astrophysical research (e.g., astrophysical parameter estimations and investigations on complex multivariate physical processes).

astro-ph.IM↗

arXiv2025

C3PO V: Comoving stellar pairs indicate rotational spin-down drives the main-sequence Li-Dip

The lithium-dip observed in mid-F dwarfs remains a long-standing challenge for stellar evolution models. We present high-precision stellar parameters and A(Li) for 22 new comoving pairs, primarily located on the hotter side of the Li-Dip. Combined with pairs from the C3PO catalog, our sample includes 124 stars with Teff between 6000 and 7300 K, encompassing and extending slightly beyond the Li-Dip. Among them, 49 comoving pairs (98 stars) have both components within the temperature range of interest. Using this expanded set of comoving pairs observed with high-resolution spectroscopy, we show that rotational spin-down is the dominant process responsible for Li depletion in the Li-Dip. First, within comoving pairs, the star with v sin i > 12 km/s shows higher A(Li) than its more slowly rotating companion within the Li-Dip, indicating that rotation-dependent mixing drives lithium depletion. Second, we observe a correlation between A(Li) and v sin i: fast rotators retain higher A(Li) with less scatter, while slow rotators show lower A(Li) and greater dispersion. Third, among slow rotators, A(Li) varies widely, suggesting that differences in initial rotation rates and spin-down histories influence how much Li is depleted. Some stars may have formed as fast rotators and spun down rapidly, leading to more Li depletion, while others may have started as slow rotators and retained more of their initial Li. These results demonstrate that rotational induced mixing plays an important role in shaping the Li-Dip beyond the effects of stellar age and mass.

astro-ph.SR↗

arXiv2025

Teaching LLMs to Speak Spectroscopy

Pre-trained Large Language Models (LLMs) have revolutionized text processing, yet adapting Transformer-based neural networks to non-textual scientific modalities typically requires specialized architectures and extensive computational resources. We demonstrate that LLaMA-3.1-8B can be efficiently repurposed to predict galaxy redshifts from spectroscopic data through Low-Rank Adaptation (LoRA), achieving competitive performance while preserving its linguistic capabilities. Using only 16 GPU-hours and adapting 0.04% of model parameters, our approach achieves a mean absolute error of 0.04 in redshift prediction while retaining over 85% of performance on AstroBench and 89% on general QA tasks from eval-harness. This minimal-effort adaptation--requiring only simple standard fine-tuning APIs--lowers barriers to entry for domain scientists and enables integrated agentic workflows where a single model handles both spectroscopic data for quantitative analysis and natural language for reasoning.

astro-ph.IM↗

arXiv2025

Planets Around Solar Twins/Analogs (PASTA) II: chemical abundances, systematic offsets, and clues to planet formation

Context. Previous studies have suggested that the Sun is relatively depleted in refractory elements compared to other solar twins or analogs, potentially as a result of planet formation. However, such conclusions are often limited by inhomogeneous samples and a lack of direct comparison with stars known to host planets. Aims. We aim to perform a homogeneous and precise abundance analysis of solar twins and analogs that host planets, to investigate possible chemical signatures associated with planet formation. Methods. We obtain high-resolution, high signal-to-noise ratio Magellan/MIKE spectra for 25 solar-like stars, including 22 confirmed or candidate planet hosts and three comparison stars. Stellar parameters and elemental abundances for 23 elements (from C to Eu) are derived through a strict line-by-line differential analysis relative to the Sun. Results. Our sample spans [Fe/H] = -0.23 to +0.18 dex and includes 20 solar analogs, six of which are solar twins. Typical abundance uncertainties range from 0.01 to 0.05 dex for lighter elements (e.g., Fe, Si, C, O, Na) and up to 0.1 dex for neutron-capture elements. The Sun is consistently depleted in refractory elements relative to all solar analogs and twins, regardless of planet type. Stars hosting small planets tentatively show slightly stronger refractory element depletion than those hosting giant planets, though the difference is not yet statistically significant. Conclusions. We emphasize the need for strictly differential, line-by-line analyses relative to the Sun, as well as careful consideration of systematic differences between instruments, to ensure consistency and the homogeneity required to achieve our goals.

astro-ph.EP↗

arXiv2025

Effective Training Data Synthesis for Improving MLLM Chart Understanding

Being able to effectively read scientific plots, or chart understanding, is a central part toward building effective agents for science. However, existing multimodal large language models (MLLMs), especially open-source ones, are still falling behind with a typical success rate of 30%-50% on challenging benchmarks. Previous studies on fine-tuning MLLMs with synthetic charts are often restricted by their inadequate similarity to the real charts, which could compromise model training and performance on complex real-world charts. In this study, we show that modularizing chart generation and diversifying visual details improves chart understanding capabilities. In particular, we design a five-step data synthesis pipeline, where we separate data and function creation for single plot generation, condition the generation of later subplots on earlier ones for multi-subplot figures, visually diversify the generated figures, filter out low quality data, and finally generate the question-answer (QA) pairs with GPT-4o. This approach allows us to streamline the generation of fine-tuning datasets and introduce the effective chart dataset (ECD), which contains 10k+ chart images and 300k+ QA pairs, covering 25 topics and featuring 250+ chart type combinations with high visual complexity. We show that ECD consistently improves the performance of various MLLMs on a range of real-world and synthetic test sets. Code, data and models are available at: https://github.com/yuweiyang-anu/ECD.

cs.CV↗

arXiv2025

Interpreting Multi-band Galaxy Observations with Large Language Model-Based Agents

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.

astro-ph.IM↗

arXiv2025

Finetuning Stellar Spectra Foundation Models with LoRA

Foundation models are beginning to impact stellar spectroscopy, where spectra encode rich physical information in a structured, language-like form. A key challenge is adapting these models across heterogeneous surveys with differing resolution and coverage. We apply Low-Rank Adaptation (LoRA) to fine-tune SpecCLIP--a contrastively pre-trained model on LAMOST and Gaia XP spectra--for downstream tasks on DESI Early Data Release (EDR) spectra. We show that LoRA enables few-shot learning on DESI, with performance varying by fine-tuned module and benefiting from Gaia XP knowledge embedded in the pre-trained model. Our results demonstrate that LoRA provides a lightweight and effective strategy for extending spectral foundation models to new instruments and survey domains.

astro-ph.IM↗

arXiv2025

Why Machine Learning Models Systematically Underestimate Extreme Values

A persistent challenge in astronomical machine learning is a systematic bias where predictions compress the dynamic range of true values-high values are consistently predicted too low while low values are predicted too high. Understanding this bias has important consequences for astronomical measurements and our understanding of physical processes in astronomical inference. Through analytical examination of linear regression, we show that this bias arises naturally from measurement uncertainties in input features and persists regardless of training sample size, label accuracy, or parameter distribution. In the univariate case, we demonstrate that attenuation becomes important when the ratio of intrinsic signal range to measurement uncertainty ($σ_{\text{range}}/σ_x$) is below $O(10)$-a regime common in astronomy. We further extend the theoretical framework to multivariate linear regression and demonstrate its implications using stellar spectroscopy as a case study. Even under optimal conditions-high-resolution APOGEE-like spectra ($R=24,000$) with high signal-to-noise ratios (SNR=100) and multiple correlated features-we find percent-level bias. The effect becomes even more severe for modern-day low-resolution surveys like LAMOST and DESI due to the lower SNR and resolution. These findings have broad implications, providing a theoretical framework for understanding and addressing this limitation in astronomical data analysis with machine learning.

astro-ph.IM↗

arXiv2025

Millions of Main-Sequence Binary Stars from Gaia BP/RP Spectra

We present the main-sequence binary (MSMS) Catalog derived from Gaia Data Release 3 BP/RP (XP) spectra. Leveraging the vast sample of low-resolution Gaia XP spectra, we develop a forward modeling approach that maps stellar mass and photometric metallicity to XP spectra using a neural network. Our methodology identifies binary systems through statistical comparison of single- and binary-star model fits, enabling detection of binaries with mass ratios between 0.4 and 1.0 and flux ratios larger than 0.1. From an initial sample of 35 million stars within 1 kpc, we identify 14 million binary candidates and define a high-confidence "golden sample" of 1 million binary systems. This large, homogeneous sample enables detailed statistical analysis of binary properties across diverse Galactic environments, providing new insights into binary star formation and evolution. In addition, the $χ^2$ comparison allows us to distinguish stars with luminous companions from single stars or binaries with dark companions, such as white dwarfs, neutron stars and black hole candidates, improving our understanding of compact object populations.

astro-ph.SR↗

arXiv2025

Differentiable Stellar Atmospheres with Physics-Informed Neural Networks

We present Kurucz-a1, a physics-informed neural network (PINN) that emulates 1D stellar atmosphere models under Local Thermodynamic Equilibrium (LTE), addressing a critical bottleneck in differentiable stellar spectroscopy. By incorporating hydrostatic equilibrium as a physical constraint during training, Kurucz-a1 creates a differentiable atmospheric structure solver that maintains physical consistency while achieving computational efficiency. Kurucz-a1 can achieve superior hydrostatic equilibrium and more consistent with the solar observed spectra compared to ATLAS-12 itself, demonstrating the advantages of modern optimization techniques. Combined with modern differentiable radiative transfer codes, this approach enables data-driven optimization of universal physical parameters across diverse stellar populations-a capability essential for next-generation stellar astrophysics.

astro-ph.SR↗

arXiv2025

Anchoring Stellar Age Indicators: A Cross-Calibration of [C/N] and Gyrochronology Ages via the Age-Velocity-Dispersion Relation

Determining stellar ages is challenging, as it depends on other stellar parameters in a non-linear way and often relies on stellar evolution models to infer the underlying relation between these parameters and age. This complexity increases when comparing different age-dating methods, as they rely on distinct indicators and are often applicable to non-overlapping regions of the color-magnitude diagram. Moreover, many empirical calibration methods rely on pre-determined ages, often from open clusters or asteroseismology, which only cover a limited parameter space. Fortunately, the age-velocity-dispersion relation (AVR), in which the velocity dispersion increases with age, is a universal feature among stars of all evolutionary stages. In this paper, we 1) explore the parameter space in which [C/N] and gyrochronology are applicable, extending beyond the domains probed by asteroseismology and open clusters, and 2) assess whether the traditionally assumed [C/N] and gyrochronology relations yield ages on a consistent physical scale, after calibrating both using the same AVR. We find gyrochronology can be applied to all partially convective stars after they have converged onto the slow rotating sequence and before they experience weakened magnetic braking; [C/N] can be used to infer ages for all giants with metallicity > -0.8 dex and [C/N] < -0.05 dex, and can be used as an age-indicator down to [Fe/H] of -1 dex if only selecting the low-$α$ disk. Lastly, ages obtained from [C/N] and gyrochronology agree within uncertainty after accounting for systematic offsets.

astro-ph.SR↗

arXiv2025

Causal Discovery of Latent Variables in Galactic Archaeology

Galactic archaeology--the study of stellar migration histories--provides insights into galaxy formation and evolution. However, establishing causal relationships between observable stellar properties and their birth conditions remains challenging, as key properties like birth radius are not directly observable. We employ Rank-based Latent Causal Discovery (RLCD) to uncover the causal structure governing the chemodynamics of a simulated Milky Way galaxy. Using only five observable properties (metallicity, age, and orbital parameters), we recover in a purely data-driven manner a causal graph containing two latent nodes that correspond to real physical properties: the birth radius and guiding radius of stars. Our study demonstrates the potential of causal discovery models in astrophysics.

astro-ph.GA↗