Xiao Kong

KIC 7211759 is a rapidly rotating δ Sct– γ Dor hybrid pulsator. Based on high-precision photometric data from the Kepler and TESS missions, we conduct a systematic asteroseismic study of this star. We detect a pulsation spectrum spanning a broad frequency range from low-frequency gravity modes to high-frequency pressure modes (0.6–90 day ⁻¹ ). Long-term amplitude and phase modulations have been observed in the dominant pulsation modes, exhibiting features distinct from known systematic effects. Our analysis confirms that these modulations arise from beating between closely spaced frequency components. By combining the Fourier transform, autocorrelation function, and histogram of frequency differences methods, we identify the p -mode large separation to two candidate values and constrain two candidate period spacings for g -modes. In addition, we infer a likely rotation frequency of ν _rot ≈ 2.2237 day ⁻¹ from the low-frequency signals. Further constraints from stellar evolution models, incorporating Δ ν , Δ P _g , and fundamental atmospheric parameters, yield a mass of 1.70 ± 0.08 M _⊙ , a radius of 2.28 ± 0.27 R _⊙ , and and an age of 1.19 ± 0.04 dex, placing the star at the evolutionary stage just past the second turning point of the main sequence.

Astronomical image interpretation requires not only visual perception, but also familiarity with domain-specific representations, observational conventions, and astrophysical reasoning. Although recent multimodal large language models (MLLMs) have shown strong general visual–language capabilities, their performance on specialized astronomical figures remains insufficiently evaluated. To address this gap, we introduce AstroMMBench, an astronomy-specific benchmark for evaluating MLLMs on figure-associated astronomical multiple-choice questions. AstroMMBench contains 592 expert-screened multiple-choice questions covering six major astrophysical subfields: Astrophysics of Galaxies, Cosmology and Extragalactic Astrophysics, Earth and Planetary Astrophysics, High Energy Astrophysical Phenomena, Instrumentation and Methods for Astrophysics, and Solar and Stellar Astrophysics. The questions were generated from astronomical figures and associated textual context through an automated pipeline, followed by multi-stage filtering and expert review to assess scientific correctness, image–question alignment, and answer uniqueness. Using AstroMMBench, we evaluated 25 selected MLLMs, including 22 open-source and 3 closed-source models. The results show substantial performance differences across models and astrophysical subfields. Ovis2-34B achieved the highest point-estimate overall accuracy of 71.3%, with performance comparable to strong closed-source models such as ChatGPT-4o and Doubao-1.5-vision-pro. Subfield-level analysis shows the lowest mean point-estimate performance in cosmology, whereas high-energy astrophysics exhibits substantial between-model variability rather than uniformly low performance. Question-only results on the 592 benchmark questions yielded 27.53% for Qwen2.5-VL-7B and 24.83% for InternVL3-38B, compared with image-present accuracies of 58.11% and 68.24%, respectively. These descriptive comparisons suggest a substantial contribution from visual input in the two models. AstroMMBench provides a structured and extensible resource for comparing model behavior on figure-associated astronomical questions.

Large photometric surveys require classification methods that retain sensitivity to sparsely represented periodic variable-star classes. We present a two-stage decoupled multimodal network (TDMN) for imbalanced periodic-variable-star classification using phase-folded light curves, three-channel morphology images, and global features. Stage 1 organizes the three modalities using proxy contrastive objectives and real-observation class prototypes. Stage 2 retains low-level morphology extractors while adapting higher-level representations and combining class-balanced, logit-adjusted, and cosine-focal experts. Across three training runs on fixed survey-specific splits, TDMN achieves mean macro F1-scores of 0.85 on CRTS, 0.83 on the OGLE-IV 15-subclass task, and 0.87 on the OGLE-IV 6-superclass task. A retrained UPSILoN-feature random forest evaluated on the same test objects yields corresponding mean scores of 0.66, 0.78, and 0.88. Thus, TDMN has higher mean macro F1 on CRTS and the OGLE-IV subclass task, but not on the superclass task. Rotational variables remain difficult, and increased Type II Cepheid recall with TDMN is accompanied by lower precision. The class-wise recovery and purity estimates can support the construction of known periodic-variable samples and the prioritization of candidates for follow-up observations. The study remains a catalogue-based, closed-set classification analysis and assumes an externally supplied period for TDMN; period discovery, unknown-class recognition, and nonperiodic transients lie outside its scope.

Modern time-domain surveys discover transients faster than dedicated spectroscopic facilities can classify them. We test whether archival large-scale multiobject spectroscopy can partly relieve this bottleneck using 108 LAMOST low-resolution spectra of 102 cataloged transients. The sample includes 53 primary supernova (SN) candidates for classifier comparison and 19 externally spectroscopically confirmed SNe for reference. We apply a host-continuum suppression workflow in which direct host-spectrum subtraction is adopted when a clean independent host spectrum provides a stable residual, and STARLIGHT stellar-continuum modeling is adopted when such a spectrum is unavailable or ambiguous. We benchmark SNID, GELATO, and DASH on spectra affected by host light in the 3 . ″ 3 LAMOST fiber. For the 53 SN candidates, GELATO returns candidate classifications for all objects, compared with 24 (45%) for DASH and 12 (23%) for SNID v5.0/templates-2.0; these fractions measure returned yield, not accuracy. In the 12 SNID-pass cases, SNID and GELATO agree at the broad-family level for seven objects (58%); the lower SNID yield is consistent with low-contrast broad SN features in host-contaminated residual spectra, although tool-specific templates, preprocessing, and thresholds also contribute. In this sample, DASH assigns stripped-envelope labels to several spectra for which GELATO returns Type II-family candidate labels. Sixteen candidates (30%) precede maximum light and 18 (34%) are at or before +5 days, reflecting temporal selection and retrospective mining of the decade-scale LAMOST archive rather than dedicated transient monitoring. These results define a practical procedure for serendipitous transient spectroscopy with multifiber survey data.

We present a clean, homogeneous, and AI-ready stellar dataset assembled from LAMOST DR11 low-resolution spectra, tailored for data-driven models. This dataset contains 111,417 high-quality spectra for 91,542 AFGK-type stars, with stellar labels such as T _eff , log g , [Fe/H], [ α /M], stellar mass, and absolute G -band magnitude. It reaches a metallicity of [Fe/H] ∼ −3.8 dex, significantly below the LAMOST pipeline’s lower limit of −2.5 dex, and covers T _eff ≈ 3800–13,000 K, log g ≈ 0.1–5.0 dex, [ α /M] from −0.4 to 0.6 dex, mass from 0.4 to 6.0 M _⊙ , and M _G from −4.0 to 8.0 mag. Self-consistency validation using neural networks confirms the reliability of stellar labels and shows only minor differences between the adopted and network-predicted parameters. By combining this dataset with the 5105 M-type stars from our previous work, we trained a neural network model and applied it to the complete set of LAMOST DR12 AFGKM-type spectra, obtaining stellar parameters for 9 million spectra. The model yields precise estimates that agree well with APOGEE and Gaia estimates, with uncertainties of about 55 K in T _eff , 0.10 dex in log g , 0.07 dex in [Fe/H], 0.02 dex in [ α /M], a relative error of about 4.2% in stellar mass, and 0.13 mag in M _G . Distances inferred from M _G show a relative error of 7.4%. This work delivers a public, AI-ready dataset and a value-added catalog for LAMOST DR12, extends investigations of very metal-poor stars toward lower metallicities, and supports direct, efficient estimation of stellar parameters, mass, and distance from large spectroscopic surveys.

Photometric redshift ( z _photo or photo- z ) estimation for quasars is crucial for large-scale surveys. However, estimating redshifts from photometric images alone often suffers from limited accuracy and a large dispersion of estimation errors, as quasars generally appear as point sources with few informative pixels. To address these limitations, we propose QuasarCLIP-z, a CLIP-based (Contrastive Language–Image Pre-training) cross-modal method for quasar photometric redshift estimation. At the training stage, QuasarCLIP-z aligns paired SDSS multi-band images and optical spectra within a shared latent space via contrastive learning. This alignment transfers knowledge from spectra, which provide more redshift-sensitive information, to image features. At the application stage, QuasarCLIP-z requires only SDSS multi-band photometric images as input, without requiring spectroscopic data. Evaluated on an SDSS quasar sample, QuasarCLIP-z simultaneously reduces the normalized median absolute deviation ( σ _NMAD ) from 0.0659 to 0.0592 and increases the fraction of objects satisfying ∣( z _photo − z _spec )/(1 + z _spec )∣ < 0.1 (spectroscopic redshift, denoted as z _spec or spec- z ) from 77.46% to 81.90% compared to the image-only ResNet-18 (Residual Network-18), demonstrating that our method effectively improves estimation accuracy and reduces the dispersion of estimation errors. In addition, QuasarCLIP-z exhibits strong few-shot performance: fine-tuning with only a 10% subset of the training data yields an accuracy close to that achieved with the full set. Overall, our method provides a practical way for quasar photometric redshift estimation in spectroscopically limited surveys.

To improve the accuracy and efficiency of high-dimensional stellar parameter inference in large spectroscopic datasets, we propose a projection-assisted parameter-inference framework—Projected-Space Inference of Stellar Parameters (PISP). PISP constructs an orthonormal basis and optimizes in the projected space, reducing the impact of parameter correlations on inference. The basis is constructed using either principal component analysis (PCA) or the active-subspace (AS) method and is combined with two inference strategies—Non-L1, which optimizes the projection coefficients for a user-specified projected dimensionality, and L1, which introduces L1 regularization in the full projected space to adaptively select projection directions—yielding four strategies: PCA-Non-L1, AS-Non-L1, PCA-L1, and AS-L1. For different computational scenarios, we implement two versions: PISP-CurveFit for fast single-spectrum inference and PISP-Adam for large-scale GPU-parallel inference. Using a fully connected neural network and a residual network as spectral emulators, we evaluate PISP on Kurucz synthetic spectra and on 722,896 APOGEE DR17 observed spectra. Compared to the baseline strategy, PISP improves inference accuracy for multiple parameters across all emulator–optimizer combinations. In synthetic data, PCA-L1 performs best, reducing the standard deviation of differences ( σ (Δ)) by at least 0.01 dex for 12 of 20 elemental abundances, with [N/H], [O/H], [Na/H], [Co/H], [P/H], [V/H], [Cu/H] showing 0.05–0.72 dex reductions. In observed data, PCA-Non-L1 reduces σ (Δ) by >30 K for effective temperature and by at least 0.01 dex for 9 of 17 elemental abundances, with [O/H], [Na/H], [V/H] showing 0.05–0.20 dex reductions, while achieving a ∼4× efficiency gain, slightly outperforming PCA-L1.

In recent years, large language models (LLMs) have transformed natural language understanding through vast data sets 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 data sets, our goal is to learn robust and informative embeddings that support diverse downstream applications. As a proof of concept, SpecCLIP involves pretraining on two spectral types—LAMOST low-resolution and Gaia XP—followed by contrastive alignment using the Contrastive Language–Image Pretraining (CLIP) 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, the former being achieved by maximizing mutual information between embeddings and input spectra. The result is a cross-spectrum framework that enables intrinsic calibration and flexible applications across instruments. We demonstrate that fine-tuning these models on moderate-sized labeled data sets 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. In addition, 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 on GitHub ✎.

To enhance the efficiency, scalability, and cross-survey applicability of stellar parameter inference in large spectroscopic datasets, we present a modular, parallelized Python framework with automated error estimation, built on the LAMOST Atmospheric Parameter Pipeline (LASP) originally implemented in IDL. Rather than a direct code translation, this framework refactors LASP with two complementary modules: LASP-CurveFit, a new implementation of the LASP fitting procedure that runs on a CPU, preserving legacy logic while improving data I/O and multithreaded execution efficiency; and LASP-Adam-GPU, a GPU-accelerated method that introduces grouped optimization by constructing a joint residual function over multiple observed and model spectra, enabling high-throughput parameter inference across tens of millions of spectra. Applied to 10 million LAMOST spectra, the framework reduces runtime from 84 to 48 hr on the same CPU platform and to 7 hr on an NVIDIA A100 GPU, while producing results consistent with those from the original pipeline. The inferred errors agree well with the parameter variations from repeat observations of the same target (excluding radial velocities), while the official empirical errors used in LASP are more conservative. When applied to DESI DR1, our effective temperatures and surface gravities agree better with APOGEE than those from the DESI pipeline, particularly for cool giants, while the latter performs slightly better in radial velocity and metallicity. These results suggest that the framework delivers reliable accuracy, efficiency, and transferability, offering a practical approach to parameter inference in large spectroscopic surveys. The code and DESI-based catalog are available via DOI: 10.12149/101679 and DOI: 10.12149/101675 , respectively.

Sky background subtraction is a critical step in the Multi-objective Fiber spectra process. However, current subtraction relies mainly on sky fiber spectra to build a super sky. These average spectra are lacking in the modeling of the environment surrounding the objects. To address this issue, a sky background estimation model: Sky background building based on Mutual Information (SMI) is proposed. SMI is based on mutual information and an incremental training approach. It utilizes spectra from all fibers in the plate to estimate the sky background. SMI contains two main networks, the first network applies a wavelength calibration module to extract sky features from spectra, and can effectively solve the feature shift problem according to the corresponding emission position. The second network employs an incremental training approach to maximize mutual information between representations of different spectra to capture the common component. Then, it minimizes the mutual information between adjoining spectral representations to obtain individual components. This network yields an individual sky background at each location of the object. To verify the effectiveness of the method in this paper, we conducted experiments on the spectra of LAMOST. Results show that SMI can obtain a better object sky background during the observation, especially in the blue end.