arXiv subjects

Yixiao Zhou

Publications and source records attributed to Yixiao Zhou.

At least 19 recordsLinked to original sources

arXiv2026

Robust Bimanual Vision-Language-Action Models via Embarrassingly Simple Modality Masking

Query-based Vision-Language-Action (VLA) models offer low-latency inference that is attractive for bimanual robotic manipulation, but we observe that they can still exhibit discontinuous actions and execution failures in complex dual-arm tasks. We hypothesize that unstable multi-view and language fusion is one contributing factor in these failures, often coinciding with attention spreading to distracting regions. To improve robustness, we introduce the Modality Masking Mechanism (M3), an embarrassingly simple, training-only strategy that requires no architectural changes or large-scale robot pretraining. M3 stochastically masks subsets of modality channels during training, exposing the policy to controlled partial observations and encouraging it to rely less on distracting cues and more on evidence that remains reliable. We evaluate M3 on ten bimanual tasks from RoboTwin 2.0 and on three long-horizon real-world tasks. Compared with the Adapter baseline, M3 improves average success by 21.7% in the Clean setting and 11.4% in Clean2Rand, where policies are trained on clean demonstrations and evaluated on randomized scenes, while also improving averaged real-world full-task success by over 30%. These results suggest that structured training-time masking is a practical way to improve the robustness of query-based VLA policies for bimanual manipulation.

cs.RO↗

arXiv2026

GASTAG evolutionary tracks and isochrones from coupled 1D and 3D models: Systematic temperature offsets in red giants

Models of stellar structure and evolution describe the global and internal properties of stars throughout their lifetimes and are indispensable for studies of stellar clusters and Galactic evolution. However, most 1D evolutionary calculations rely on simplified treatment of convection, resulting in inaccurate near-surface structures and non-negligible uncertainties in the predicted fundamental parameters of low-mass stars. In a series of previous works, a novel approach was developed to couple 1D stellar interior models with 3D model atmospheres throughout the evolutionary calculation. This 1D-3D coupling method makes predicted stellar properties effectively independent of the mixing-length parameter. To expand this framework to ensemble studies of stars and age determinations of clusters, we present the GASTAG stellar evolutionary tracks and isochrones constructed using the 1D-3D coupling approach. Comparing effective temperatures from the APOGEE-Kepler catalog with GASTAG predictions, we find the theoretical temperatures are cooler by about 70 K near solar metallicity. Our isochrones are compared with observed color-magnitude diagrams of star clusters spanning from $\rm [Fe/H] = 0.3$ to $-1.9$. In all cases, the synthesized and observed diagrams agree excellently in the main-sequence, turn-off, and subgiant regions, while isochrones predict systematically cooler red giant branches. Taking these independent findings together reveals that the temperature mismatch is most likely due to deficiencies in stellar models. Because GASTAG is constructed using a method that substantially reduces uncertainties associated with surface boundary conditions and the mixing-length parameter, the difference between modeling and observation can be more confidently attributed to other ingredients in the models, such as $α$-element abundances or uncertainties in low-temperature opacities.

astro-ph.SR↗

arXiv2026

GSDrive: Reinforcing Driving Policies by Multi-mode Future Trajectory Probing with 3D Gaussian Splatting Environment

End-to-end (E2E) autonomous driving aims to directly map sensory observations to driving actions, but its real-world deployment is hindered by evolving data distributions and the high cost of continual annotation. While combining imitation learning (IL) and reinforcement learning (RL) is a common strategy for policy improvement, conventional RL training relies on delayed, event-based rewards, where policies learn only from catastrophic outcomes such as collisions, leading to premature convergence to suboptimal behaviors. To address these limitations, we propose GSDrive, a framework that uses a differentiable 3D Gaussian Splatting (3DGS) environment for future-aware trajectory probing and reward shaping in E2E driving. GSDrive first learns a multi-mode trajectory probe via IL and then uses RL to evaluate multiple candidate futures in the 3DGS environment, converting their simulated returns into dense shaping rewards for policy optimization. This yields a cyclic hybrid IL-RL training loop, where IL supplies structured future priors and RL provides interactive feedback for iterative refinement. Evaluated on the reconstructed nuScenes dataset, our method outperforms other simulation-based RL approaches in closed-loop experiments. Code is available at https://github.com/ZionGo6/GSDrive.

cs.RO↗

arXiv2026

One Refiner to Unlock Them All: Inference-Time Reasoning Elicitation via Reinforcement Query Refinement

Large Language Models (LLMs) often fail to utilize their latent reasoning capabilities due to a distributional mismatch between ambiguous human inquiries and the structured logic required for machine activation. Existing alignment methods either incur prohibitive $O(N)$ costs by fine-tuning each model individually or rely on static prompts that fail to resolve query-level structural complexity. In this paper, we propose ReQueR (\textbf{Re}inforcement \textbf{Que}ry \textbf{R}efinement), a modular framework that treats reasoning elicitation as an inference-time alignment task. We train a specialized Refiner policy via Reinforcement Learning to rewrite raw queries into explicit logical decompositions, treating frozen LLMs as the environment. Rooted in the classical Zone of Proximal Development from educational psychology, we introduce the Adaptive Solver Hierarchy, a curriculum mechanism that stabilizes training by dynamically aligning environmental difficulty with the Refiner's evolving competence. ReQueR yields consistent absolute gains of 1.7\%--7.2\% across diverse architectures and benchmarks, outperforming strong baselines by 2.1\% on average. Crucially, it provides a promising paradigm for one-to-many inference-time reasoning elicitation, enabling a single Refiner trained on a small set of models to effectively unlock reasoning in diverse unseen models. Code is available at https://github.com/newera-xiao/ReQueR.

cs.CL↗

arXiv2026

A Test of Substellar Evolutionary Models with High-Precision Ages from Asteroseismology and Gyrochronology for the Benchmark System HR 7672AB

We present high-precision measurements for HR~7672AB, composed of a Sun-like (G0V) star and an L~dwarf companion. Three nights of precise (70 cm/s) radial velocity (RV) asteroseismology with the Keck Planet Finder clearly detect 5-minute oscillations from the primary HR~7672A, and modeling of the frequency spectrum yields an asteroseismic age of $1.87\pm0.65$~Gyr. We also determine a gyrochronological age of $2.58\pm0.47$~Gyr, and we combine these two results for a final age of $2.26\pm0.40$~Gyr. In addition, we obtained new RVs for HR~7672A and new astrometry for the companion HR~7672B. From a joint orbit fit, we measured a dynamical mass of $1.111\pm0.017$~$\text{M}_\odot$ for HR~7672A and $75.39\pm0.67$~$\text{M}_{\text{Jup}}$ for HR~7672B. This places the companion near the stellar/substellar boundary and thus particularly sensitive to differences in model predictions. The joint precision in host star age (18\% uncertainty) and companion mass (0.9\% uncertainty) makes HR~7672AB an exceptional substellar benchmark. Combined with the companion's luminosity, we use these measurements to test predictions from six brown dwarf cooling models. The best agreement occurs with the Chabrier et al. (2023) models, which incorporate a new equation of state, resulting in predictions that agree within $<$0.3$σ$ with all the observations. The other 5 sets of models agree at the 1--3$σ$ level depending on the particular test, and some models struggle to predict a sufficient low luminosity for HR~7672B at any age given its dynamical mass. We also detected a weak seismic signal in near-simultaneous TESS photometry of HR~7672A, with the resulting RV-to-photometry oscillation amplitude ratio consistent with solar values.

astro-ph.SR↗

arXiv2026

Look Inward to Explore Outward: Learning Temperature Policy from LLM Internal States via Hierarchical RL

Reinforcement Learning from Verifiable Rewards (RLVR) trains large language models (LLMs) from sampled trajectories, making decoding strategy a core component of learning rather than a purely inference-time choice. Sampling temperature directly controls the exploration--exploitation trade-off by modulating policy entropy, yet existing methods rely on static values or heuristic adaptations that are decoupled from task-level rewards. We propose Introspective LLM, a hierarchical reinforcement learning framework that learns to control sampling temperature during generation. At each decoding step, the model selects a temperature based on its hidden state and samples the next token from the resulting distribution. Temperature and token policies are jointly optimized from downstream rewards using a coordinate ascent scheme. Experiments on mathematical reasoning benchmarks show that learned temperature policies outperform fixed and heuristic baselines, while exhibiting interpretable exploration behaviors aligned with reasoning uncertainty.

cs.LG↗

arXiv2026

Granulation signatures as seen by Kepler short-cadence data. I. A decoupling between granulation and oscillation timescales for dwarfs

Granulation is the observable signature of convection in envelopes of low-mass stars, forming the background in stellar power spectra. While well-studied in evolved giants, granulation on the MS has received less attention. We here study and characterise granulation signatures of MS and SGB stars, extending previous studies of giants to provide a continuous physical picture across evolutionary stages. We analyse 753 Kepler short-cadence stars using a Bayesian nested-sampling framework to evaluate three background descriptions and compare model preferences. This yields full posterior distributions for all parameters, enabling robust comparisons across a diverse stellar sample. No universal preference between background models is found. Assuming a Gaussian oscillation envelope, $ν_\mathrm{max}$ estimates are sensitive to model misspecification, with the resulting systematics exceeding the formal uncertainties. The envelope width scales with $ν_\mathrm{max}$ across models and shows a dependence on effective temperature. Total granulation amplitudes in dwarfs broadly follow giant-based scalings, however a decoupling appears between the timescale of the primary granulation and the oscillations for MS stars cooler than the Sun. The prolonged granulation timescale is reproduced by 3D simulations of a K-dwarf, driven by reduced convective velocities due to more efficient convective energy transport in denser envelopes. The prolonged granulation timescale increases the frequency separation to the oscillation excess, potentially aiding seismic detectability, while the reduced convective velocities may influence the excitation of stellar oscillations and relate to the low amplitudes observed in cool dwarfs. Finally, we contribute a dataset linking granulation, oscillations, and stellar parameters, providing a foundation for future investigations into their interdependence across the HR diagram.

astro-ph.SR↗

arXiv2026

Listen, Look, Drive: Coupling Audio Instructions for User-aware VLA-based Autonomous Driving

Vision Language Action (VLA) models promise an open-vocabulary interface that can translate perceptual ambiguity into semantically grounded driving decisions, yet they still treat language as a static prior fixed at inference time. As a result, the model must infer continuously shifting objectives from pixels alone, yielding delayed or overly conservative maneuvers. We argue that effective VLAs for autonomous driving need an online channel in which users can influence driving with specific intentions. To this end, we present EchoVLA, a user-aware VLA that couples camera streams with in situ audio instructions. We augment the nuScenes dataset with temporally aligned, intent-specific speech commands generated by converting ego-motion descriptions into synthetic audios. Further, we compose emotional speech-trajectory pairs into a multimodal Chain-of-Thought (CoT) for fine-tuning a Multimodal Large Model (MLM) based on Qwen2.5-Omni. Specifically, we synthesize the audio-augmented dataset with different emotion types paired with corresponding driving behaviors, leveraging the emotional cues embedded in tone, pitch, and speech tempo to reflect varying user states, such as urgent or hesitant intentions, thus enabling our EchoVLA to interpret not only the semantic content but also the emotional context of audio commands for more nuanced and emotionally adaptive driving behavior. In open-loop benchmarks, our approach reduces the average L2 error by $59.4\%$ and the collision rate by $74.4\%$ compared to the baseline of vision-only perception. More experiments on nuScenes dataset validate that EchoVLA not only steers the trajectory through audio instructions, but also modulates driving behavior in response to the emotions detected in the user's speech.

eess.AS↗

arXiv2025

Validating the 1D-3D coupling stellar models via Asteroseismology of 18 Kepler main-sequence stars

Standard 1D stellar evolution model has poor descriptions of the near-surface layers of stars, and this can be improved by using the atmosphere model computed from 3D hydrodynamical simulations. In this work, we validated the model inferences of the 1D-3D coupling models using 18 well-studied stars from the Kepler LEGACY Sample. We compared our estimates of the fundamental parameters determined with other six pipelines and obtained good consistency. The results indicate that the 1D-3D coupling models can be applied to characterizing solar-like stars with confidence. Our analysis showed similar pattern for the surface term in stars with effective temperature range from ~5000 K to ~6400 K, suggesting that the surface term of the 1D-3D coupling models is simpler and easier to deal with than that of models using classical atmosphere.

astro-ph.SR↗

arXiv2025

Dropping Experts, Recombining Neurons: Retraining-Free Pruning for Sparse Mixture-of-Experts LLMs

Sparse Mixture-of-Experts (SMoE) architectures are widely used in large language models (LLMs) due to their computational efficiency. However, though only a few experts are activated for each token, SMoE still requires loading all expert parameters, leading to high memory usage and challenges in deployment. Previous work has tried to reduce the overhead by pruning and merging experts, but primarily focused on expert-level operations, leaving neuron-level structure underexplored. We propose DERN (Dropping Experts, Recombining Neurons), a task-agnostic and retraining-free framework for expert pruning and reconstruction. We observe that experts are often misaligned and contain semantic conflicts at the neuron level, which poses challenges for direct merging. To solve this, DERN works in three steps: it first prunes redundant experts using router statistics; then it decomposes them into neuron-level expert segments, assigning each segment to its most compatible retained expert; and finally, it merges segments within each retained expert to build a compact representation. Experiments on Mixtral, Qwen, and DeepSeek SMoE models show that DERN improves performance by more than 5% on commonsense reasoning and MMLU benchmarks under 50% expert sparsity, without extra training. It also greatly reduces the number of experts and memory usage, making SMoE LLMs easier to deploy in practice.

cs.CL↗

arXiv2025

Asteroseismology of the G8 subgiant beta Aquilae with SONG-Tenerife, SONG-Australia and TESS

We present time-series radial velocities of the G8 subgiant star beta Aql obtained in 2022 and 2023 using SONG-Tenerife and, for the first time, SONG-Australia. We also analyse a sector of TESS photometry that overlapped with the 2022 SONG data. The resulting power spectrum clearly shows solar-like oscillations centred at 430 muHz. The TESS light curve shows the oscillations at lower signal-to-noise, reflecting the fact that photometric measurements are much more affected by the granulation background than are radial velocities. The simultaneous observations in velocity and photometry represent the best such measurements for any star apart from the Sun. They allowed us to measure the ratio between the bolometric photometric amplitude and the velocity amplitude to be 26.6 +/- 3.1 ppm/(m/s). We measured this ratio for the Sun from published SOHO data to be 19.5 +/- 0.7 ppm/(m/s) and, after accounting for the difference in effective temperatures of and the Sun, these values align with expectations. In both the Sun and beta Aql, the photometry-to-velocity ratio appears to be a function of frequency. We also measured the phase shift of the oscillations in beta Aql between SONG and TESS to be -113 +/- 7 deg, which agrees with the value for the Sun and also with a 3-D simulation of a star with similar properties to beta Aql. Importantly for exoplanet searches, we argue that simultaneous photometry can be used to predict the contribution of oscillations to radial velocities. We measured frequencies for 22 oscillation modes in beta Aql and carried out asteroseismic modelling, yielding an excellent fit to the frequencies. We derived accurate values for the mass and age, and were able to place quite strong constraints on the mixing-length parameter. Finally, we show that the oscillation properties of beta Aql are very similar to stars in the open cluster M67.

astro-ph.SR↗

arXiv2025

Granulation signatures in 3D hydrodynamical simulations: evaluating background model performance using a Bayesian nested sampling framework

Understanding the granulation background signal is of vital importance when interpreting the asteroseismic diagnostics of solar-like oscillators. Various descriptions exist in the literature for modelling the surface manifestation of convection, the choice of which affects our interpretations. We aim to evaluate the performance of and preference for various granulation background models for a suite of 3D hydrodynamical simulations of convection across the HR diagram, thereby expanding the number of simulations and coverage of parameter space for which such studies have been made. We take a statistical approach by considering the granulation in power density spectra of 3D simulations, where no biases or systematics of observational origin are present. To properly contrast the performance of the models, we develop a Bayesian nested sampling framework for model inference and comparison. This framework was extended to real stellar data using KIC 8006161 (Doris) and the Sun. We find that multi-component models are consistently preferred over a single-component model, with each tested multi-component model demonstrating merit in specific cases. This occurs for simulations with no magnetic activity, thus ruling out stellar faculae as the sole source of the second granulation component. Like a previous study, we find that a hybrid model with a single overall amplitude and two characteristic frequencies performs well for numerous simulations. Additionally, a tentative third granulation component beyond the value of $ν_\mathrm{max}$ is seen for some simulations, but its potential presence in observations requires further efforts. Studying the granulation signatures in these simulations paves the way to studying stars with accurate granulation models. This deeper understanding of the granulation signal may lead to complementary methods to existing algorithms for determining stellar parameters.

astro-ph.SR↗

arXiv2025

Each Rank Could be an Expert: Single-Ranked Mixture of Experts LoRA for Multi-Task Learning

Low-Rank Adaptation (LoRA) is widely used for adapting large language models (LLMs) to specific domains due to its efficiency and modularity. Meanwhile, vanilla LoRA struggles with task conflicts in multi-task scenarios. Recent works adopt Mixture of Experts (MoE) by treating each LoRA module as an expert, thereby mitigating task interference through multiple specialized LoRA modules. While effective, these methods often isolate knowledge within individual tasks, failing to fully exploit the shared knowledge across related tasks. In this paper, we establish a connection between single LoRA and multi-LoRA MoE, integrating them into a unified framework. We demonstrate that the dynamic routing of multiple LoRAs is functionally equivalent to rank partitioning and block-level activation within a single LoRA. We further empirically demonstrate that finer-grained LoRA partitioning, within the same total and activated parameter constraints, leads to better performance gains across heterogeneous tasks. Building on these findings, we propose Single-ranked Mixture of Experts LoRA (\textbf{SMoRA}), which embeds MoE into LoRA by \textit{treating each rank as an independent expert}. With a \textit{dynamic rank-wise activation} mechanism, SMoRA promotes finer-grained knowledge sharing while mitigating task conflicts. Experiments demonstrate that SMoRA activates fewer parameters yet achieves better performance in multi-task scenarios.

cs.LG↗

arXiv2025

Coupling 1D stellar evolution with 3D-hydrodynamical simulations on-the-fly III: stellar evolution at different metallicities

A major weakness in one-dimensional (1D) stellar structure and evolution modeling is the simplified treatment of convection, which leads to erroneous near-surface stratification and considerable uncertainties in predicted effective temperatures and luminosities of low-mass stars. In a series of preceding works, a novel method for coupling 1D stellar structural models with a grid of 3D surface convection simulations during stellar evolution was developed, at solar metallicity. This 1D-3D coupling method slightly shifts evolutionary tracks relative to standard calculations, meanwhile providing oscillation frequencies that agree more closely with asteroseismic observations. Here we extend this method to model metal-poor and metal-rich FGK-type stars, by implementing interpolations on-the-fly across metallicity ($\rm -3 < [Fe/H] < 0.5$) for mean 3D models during stellar evolution. We demonstrate quantitatively that the fundamental stellar parameters modeled within our framework are insensitive to the mixing-length parameter. A 20% change in the mixing-length parameter results in evolutionary tracks with a temperature shift of less than 30 K, compared to a difference of over 200 K in standard evolution calculations. Our extension is validated against eclipsing binary systems with extremely precise observational constraints as well as stars in binaries with asteroseismic data. Using a fixed mixing-length parameter that merely governs convective heat transport in the near-adiabatic layers, the 1D-3D coupling method successfully reproduces most observational constraints for all target stars. Coupling 1D stellar evolution models with 3D simulations greatly reduces uncertainties associated with the choice of atmosphere boundary conditions and mixing-length parameters, hence offering a powerful tool for characterizing stars with seismic measurements and determining ages for globular clusters.

astro-ph.SR↗

arXiv2024

Realistic Uncertainties for Fundamental Properties of Asteroseismic Red Giants and the Interplay Between Mixing Length, Metallicity and $ν_{\rm max}$

Asteroseismic modelling is a powerful way to derive stellar properties. However, the derived quantities are limited by built-in assumptions used in stellar models. This work presents a detailed characterisation of stellar model uncertainties in asteroseismic red giants, focusing on the mixing-length parameter $α_{\rm MLT}$, the initial helium fraction $Y_{\rm init}$, the solar abundance scale, and the overshoot parameters. First, we estimate error floors due to model uncertainties to be $\approx$0.4\% in mass, $\approx$0.2\% in radius, and $\approx$17\% in age, primarily due to the uncertain state of $α_{\rm MLT}$ and $Y_{\rm init}$. The systematic uncertainties in age exceed typical statistical uncertainties, suggesting the importance of their evaluation in asteroseismic applications. Second, we demonstrate that the uncertainties from $α_{\rm MLT}$ can be entirely mitigated by direct radius measurements or partially through $ν_{\rm max}$. Utilizing radii from Kepler eclipsing binaries, we determined the $α_{\rm MLT}$ values and calibrated the $α_{\rm MLT}$--[M/H] relation. The correlation observed between the two variables is positive, consistent with previous studies using 1-D stellar models, but in contrast with outcomes from 3-D simulations. Third, we explore the implications of using asteroseismic modelling to test the $ν_{\rm max}$ scaling relation. We found that a perceived dependency of $ν_{\rm max}$ on [M/H] from individual frequency modelling can be largely removed by incorporating the calibrated $α_{\rm MLT}$--[M/H] relation. Variations in $Y_{\rm init}$ can also affect $ν_{\rm max}$ predictions. These findings suggest that $ν_{\rm max}$ conveys information not fully captured by individual frequencies, and that it should be carefully considered as an important observable for asteroseismic modelling.

astro-ph.SR↗

arXiv2024

An extended and refined grid of 3D STAGGER model atmospheres. Processed snapshots for stellar spectroscopy

Context: Traditional one-dimensional (1D) hydrostatic model atmospheres introduce systematic modelling errors into spectroscopic analyses of FGK-type stars. Aims: We present an updated version of the STAGGER-grid of 3D model atmospheres, and explore the accuracy of post-processing methods in preparation for spectral synthesis. Methods: New and old models were (re)computed following an updated workflow, including an updated opacity binning technique. Spectroscopic tests were performed in 3D LTE for a grid of 216 fictitious Fe I lines, spanning a wide range in oscillator strength, excitation potential and central wavelength, and eight model atmospheres that cover the stellar atmospheric parameter range (Teff, log g, [Fe/H]) of FGK-type stars. Using this grid, the impact of vertical and horizontal resolution, and temporal sampling of model atmospheres on spectroscopic diagnostics was tested. Results: We find that downsampling the horizontal mesh from its original size of 240 x 240 grid cells to 80 x 80 cells, i.e. sampling every third grid cell, introduces minimal errors on the equivalent width and normalized line flux across the line and stellar parameter space. Regarding temporal sampling, we find that sampling ten statistically independent snapshots is sufficient to accurately model the shape of spectral line profiles. For equivalent widths, a subsample consisting of only two snapshots is sufficient, introducing an abundance error of less than 0.015 dex. Conclusions: We have computed 32 new model atmospheres and recomputed 116 old model atmospheres present in the original grid. The public release of the STAGGER-grid contains 243 models, excluding models with [Fe/H] = -4.00, and the processed snapshots can be used to improve the accuracy of spectroscopic analyses.

astro-ph.SR↗

arXiv2023

Does the $ν_{\max}$ scaling relation depend on metallicity? Insights from 3D convection simulations

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 $ν_{\max}$ scaling relation: the frequency of maximum power $ν_{\max}$, assumed to be proportional to the acoustic cut-off frequency, $ν_{\rm ac}$, scales with effective temperature and surface gravity. However, the theoretical basis of the $ν_{\max}$ 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 $ν_{\max}$ 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 $ν_{\rm ac}$ and metallicity from the 3D models. This is in tension with the positive correlation identified by studies using 1D models. Secondly, we estimated theoretical $ν_{\max}$ 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, $ν_{\max}$ does not show correlation with metallicity. This study opens an exciting prospect of testing the asteroseismic scaling relations against realistic 3D hydrodynamical stellar models.

astro-ph.SR↗

arXiv2023

3D Stagger model atmospheres with FreeEOS I. Exploring the impact of microphysics on the Sun

Three-dimensional radiation-hydrodynamics (3D RHD) simulations of stellar surface convection provide valuable insights into many problems in solar and stellar physics. However, almost all 3D near-surface convection simulations to date are based on solar-scaled chemical compositions, which limit their application on stars with peculiar abundance patterns. To overcome this difficulty, we implement the robust and widely-used FreeEOS equation of state and our Blue opacity package into the Stagger 3D radiation-magnetohydrodynamics code. We present a new 3D RHD model of the solar atmosphere, and demonstrate that the mean stratification as well as the distributions of key physical quantities are in good agreement with those of the latest Stagger solar model atmosphere. The new model is further validated by comparing against solar observations. The new model atmospheres reproduce the observed flux spectrum, continuum centre-to-limb variation, and hydrogen line profiles at a satisfactory level, thereby confirming the realism of the model and the underlying input physics. These implementations open the prospect for studying other stars with different $α$-element abundance, carbon-enhanced metal-poor stars and population II stars with peculiar chemical compositions using 3D Stagger model atmospheres.

astro-ph.SR↗