arXiv subjects

Yang Sun

Publications and source records attributed to Yang Sun.

At least 55 records · Page 3Linked to original sources

arXiv2026

JADES Dark Horse: demonstrating high-multiplex observations with JWST/NIRSpec dense-shutter spectroscopy in the JADES Origins Field

We present JWST/NIRSpec dense-shutter spectroscopy (DSS). This novel observing strategy with the NIRSpec/MSA deliberately permits a high number of controlled spectral overlaps to reach extreme multiplex while retaining the low background of slit spectroscopy. In a single configuration over the JADES Origins Field, we opened shutters on all faint (mF444W$<$30 mag) z$_\mathrm{phot}>$3 candidates, prioritising emission-line science and rejecting only bright continuum sources. Using 33.6 and 35.8 ks on-source in G235M and G395M, we observed a single mask with $\sim$850 sources, obtaining spectroscopic redshifts for $\sim$540 galaxies over 2.5$<$z$<$8.9. The per-configuration target density in DSS mode is 4-5x higher than standard no- and low-overlap MSA strategies ($<$200 sources), with no loss in redshift precision or accuracy. Line-flux sensitivities are 30 percent lower at fixed exposure time, matching the expected increase in background noise, but the gain in survey speed is 5x in our setup, more than justifying the penalty. The measured line sensitivity exceeds NIRCam/WFSS by at least $\sim$5x ($\sim$25x in exposure time) at $λ\sim4\,μ$m, demonstrating that DSS is a compelling method to gain deep, wide-band spectra for large samples. Crucially, NIRSpec/MSA could deliver even higher target allocation densities than those used here. We derive H$α$-based SFRs, gas-phase metallicities (including a large sample suitable for strong-line calibrations), and identify rare mini-quenched galaxies and broadline AGN. DSS is immediately applicable wherever deep imaging enables robust pre-selection and astrometry, providing an efficient method to obtain large samples of faint emission-line galaxies, a compelling middle ground between the completeness of slitless surveys and the sensitivity and bandwidth of NIRSpec/MSA.

astro-ph.GA↗

arXiv2026

Disentangling electronic and phononic contributions to high-temperature superconductivity in X2MH6 hydrides

Understanding the factors that control superconductivity is essential for discovering new superconducting materials using high-throughput elemental substitution. Focusing on the recently predicted ambient-pressure superconducting X2MH6 family, we disentangle the phononic and electronic contributions to Tc to determine how isoelectronic substitution alters superconductivity. While substitution affects both phononic and electronic properties, the electronic contribution plays the dominant role in determining Tc in the X2MH6 family. We show that the electronic contribution is affected by three key factors: the X-H bond distance, the electron localization function networking value of hydrogen, and the hydrogen-projected density of states at the Fermi level. A combined figure of merit derived from these parameters exhibits a robust correlation with Tc across the family. We further show that pressure produces competing effects on superconductivity: it enhances the electronic contribution by shortening X-H bonds, but simultaneously weaken the phononic contribution by increasing phonon frequencies. The net pressure dependence of Tc therefore results from the balance between these opposing tendencies. By disentangling and analyzing the electronic and phononic mechanisms, this work provides comprehensive insight into superconductivity in X2MH6 hydrides and offers practical guidance for designing new high-Tc hydride superconductors.

cond-mat.supr-con↗

arXiv2026

Calibrating Photometric Mid-Infrared Star Formation Rates for JWST

The mid-infrared (IR) spectrum of galaxies has a long history as a valuable proxy for the dust-obscured star formation rate (SFR) in massive galaxies. Now, with JWST, we can explore the mid-IR's full potential as a SFR tracer over four orders of magnitude in total infrared luminosity (9<~log LIR/Lo<~13). First, combining the SMILES and FRESCO surveys, we evaluate MIRI photometry against the Pa-alpha emission line - a gold standard SFR indicator - in Main Sequence (MS) galaxies at cosmic noon. We find the rest-frame 6-8um luminosity has a steeply superlinear relation with SFR(Pa-alpha) below ~8 Mo/yr, in contrast with the unity slope seen in coeval massive galaxies. We derive broken power-law SFR indicators from single-band MIRI photometry plus a representative dust template, with a scatter typical of IR SFRs (~0.2-0.3 dex). Despite the break in the mid-IR behavior and our simplifying assumption of a single dust SED, we next successfully formulate a UV+IR composite relation (scatter ~0.15 dex) under the usual assumption of energy balance. This implies that the rest-frame 6-8um primarily tracks the global dust-obscuration fraction - which decreases rapidly at log M*/Mo<~10 - rather than reflecting a deficit in PAH abundances at low mass. Our results thus support MIRI photometry as a robust SFR proxy at log M*/Mo>~9 up to z~3. Finally, extending to local and z>~1 ultraluminous infrared galaxies not represented in SMILES, we examine when Pa-alpha and the IR reliably track SFR in the bright regime.

astro-ph.GA↗

arXiv2026

Many-body correlations as the origin of Gamow-Teller quenching in nuclear $β$-decay

The longstanding quenching problem of Gamow-Teller (GT) strength in nuclear $β$-decay is attributed to missing contributions in the transition operator and/or incomplete nuclear correlations in the many-body wavefunction. Recent studies have predominantly emphasized operator renormalization, including chiral two-body currents, while the effects of many-body correlations--especially in heavy open-shell nuclei--remain underappreciated. We present a large-configuration shell-model calculation that incorporates chiral two-body weak current and treats both mechanisms on equal footing. Taking the neutrinoless double $β$-decay candidate $^{76}$Ge as an example, we demonstrate that strong nuclear correlations drive a substantial portion of GT strength to high excitation energies, leading to a pronounced suppression of low-energy strength responsible for the apparent quenching. We identify that the quenching originates mainly from deformation, cross-shell correlations, and mixing among densely-spaced highly excited states. In contrast, the chiral two-body current contributes only a modest $5-15\%$ reduction, depending on the coupling constants employed. Our results thus suggest many-body correlations as the primary origin of GT quenching and provide a unified microscopic explanation for this phenomenon in nuclear $β$-decay.

nucl-th↗

arXiv2026

Beyond Real Data: Synthetic Data through the Lens of Regularization

Synthetic data can improve generalization when real data is scarce, but excessive reliance may introduce distributional mismatches that degrade performance. In this paper, we present a learning-theoretic framework to quantify the trade-off between synthetic and real data. Our approach leverages algorithmic stability to derive generalization error bounds, characterizing the optimal synthetic-to-real data ratio that minimizes expected test error as a function of the Wasserstein distance between the real and synthetic distributions. We motivate our framework in the setting of kernel ridge regression with mixed data, offering a detailed analysis that may be of independent interest. Our theory predicts the existence of an optimal ratio, leading to a U-shaped behavior of test error with respect to the proportion of synthetic data. Empirically, we validate this prediction on CIFAR-10 and a clinical brain MRI dataset. Our theory extends to the important scenario of domain adaptation, showing that carefully blending synthetic target data with limited source data can mitigate domain shift and enhance generalization. We conclude with practical guidance for applying our results to both in-domain and out-of-domain scenarios.

stat.ML↗

arXiv2026

Natural Adversaries: Fuzzing Autonomous Vehicles with Realistic Roadside Object Placements

The emergence of Autonomous Vehicles (AVs) has spurred research into testing the resilience of their perception systems, i.e., ensuring that they are not susceptible to critical misjudgements. It is important that these systems are tested not only with respect to other vehicles on the road, but also with respect to objects placed on the roadside. Trash bins, billboards, and greenery are examples of such objects, typically positioned according to guidelines developed for the human visual system, which may not align perfectly with the needs of AVs. Existing tests, however, usually focus on adversarial objects with conspicuous shapes or patches, which are ultimately unrealistic due to their unnatural appearance and reliance on white-box knowledge. In this work, we introduce a black-box attack on AV perception systems that creates realistic adversarial scenarios (i.e., satisfying road design guidelines) by manipulating the positions of common roadside objects and without resorting to "unnatural" adversarial patches. In particular, we propose TrashFuzz, a fuzzing algorithm that finds scenarios in which the placement of these objects leads to substantial AV misperceptions -- such as mistaking a traffic light's colour -- with the overall goal of causing traffic-law violations. To ensure realism, these scenarios must satisfy several rules encoding regulatory guidelines governing the placement of objects on public streets. We implemented and evaluated these attacks on the Apollo autonomous driving system, finding that TrashFuzz induced violations of 15 out of 24 traffic laws.

cs.CV↗

arXiv2026

A Potential Link between Nuclear Winds and Cold Gas Outflows on Kiloparsec Scales in Reionization-Era Quasars

Feedback from accreting supermassive black holes may regulate galaxy evolution, but statistical evidence linking nuclear winds to kiloparsec-scale cold gas outflows remains limited in the early universe. Here we report statistical evidence for such a connection in a sample of luminous quasars at $z \sim 5.5$. We compare stacked [C II] 158 $μ$m emission profiles from ALMA observations, which trace galactic-scale neutral gas, for quasars with and without broad absorption lines (BALs) that indicate powerful nuclear winds on sub-kiloparsec scales. A total of 5 BAL and 11 non-BAL quasar spectra are included in the stacking analysis. The BAL quasar stack exhibits a potential blueshifted broad component in the [C II] line profile, with a velocity offset of $Δv_{\rm b} = -2.1 \times 10^2\,\rm km\,s^{-1}$ and a full width at half maximum of $1.18 \times 10^3\,\rm km\,s^{-1}$, whereas the non-BAL stack shows no obvious broad component. Using a conservative ``clean-stack'' selection that excludes quasars with partial [C II] spectral coverage, the BAL broad residual is reduced to a hint-level feature. We estimate that up to a few percent to one-quarter of the nuclear wind energy may be transferred to cold neutral gas on kiloparsec scales. Although the sample size is limited, these results suggest a potential link between BAL winds and cold gas feedback in quasar host galaxies. These results provide empirical motivation for future tests of how multiphase outflows relate to the diversity of quasar host properties, including $M_{\rm BH}/M_*$.

astro-ph.GA↗

arXiv2026

Novel phases in the Fe-Si-O system at terapascal pressures

The Fe-Si-O ternary system, central to modeling the interiors of terrestrial planets, remains poorly constrained at Terapascal (TPa) pressures characteristic of super-Earth mantles. Using a combination of crystal-structure prediction and ab initio calculations, we identify three ternary compounds stable near 1 TPa: P3 FeSiO4, P3 Fe4Si5O18, and P-3 FeSi2O6. The first two phases are thermodynamically stable at low temperatures, whereas P-3 FeSi2O6 becomes favored above approximately 2000 K. All three are metallic, paramagnetic, and adopt pseudo-binary arrangements derived from the FeO2 and SiO2 end-member structures. Their crystal structures emerge through substitutions of Fe for Si in Fe2P-type SiO2 or of Si for Fe in Pnma-type FeO2, the stable elemental oxides at ~1 TPa. This structural continuity suggests that Fe preferentially substitutes for Si in the canonical Mg-silicates expected at TPa pressures. Notably, these new pseudo-binaries accommodate Fe in six- and nine-fold coordination, in contrast to the eight-fold cubic coordination found in FeO at similar pressures. The thermodynamic conditions under which these phases form from FeO2 and SiO2 mixtures are clarified through quasi-harmonic free-energy calculations. These phases imply a fundamentally different pattern of Fe incorporation into Mg-silicates at TPa pressures compared with that inferred for Earth's mantle, i.e., mainly [Fe]Mg. Their stability may trigger silicate dissociation into oxides ((Mg,Fe)2(Si,Fe)O4 -> 2(Mg,Fe)O + (Si,Fe)O2) at pressures below ~3 TPa, as expected in the Mg-Si-O system, with the extent of dissociation governed by iron content.

cond-mat.mtrl-sci↗

arXiv2026

Protonic thermoelectric effect of Superionic H2O and magnetic field generation in Uranus and Neptune

Uranus and Neptune are characterized by anomalously tilted and multi-dipole magnetic fields, which poses substantial challenges for elucidating the internal mechanisms generating magnetic fields. Recent investigations confirmed that superionic H2O is thermodynamically stable and constitutes the dominant H2O phase within their icy mantles. In this study, we demonstrate that the superionic H2O ice exhibits a pronounced protonic thermoelectric effect, in which the maximum Seebeck coefficient within the interior of Uranus can reach approximately 620 uV/K, whereas that of Neptune is lower, within the range of 570-585 uV/K. Consequently, temperature gradients in the icy mantles can induce proton convection, which in turn drives magnetic field generation. Based on this novel mechanism, the disparities in magnetic field strength between Uranus and Neptune can be accounted for exclusively by their differing internal temperature gradients, and the predicted values are in agreement with observations.

astro-ph.EP↗

arXiv2026

AoE: Always-on Egocentric Human Video Collection for Embodied AI

Embodied foundation models require large-scale, high-quality real-world interaction data for pre-training and scaling. However, existing data collection methods suffer from high infrastructure costs, complex hardware dependencies, and limited interaction scope, making scalable expansion challenging. In fact, humans themselves are ideal physically embodied agents. Therefore, obtaining egocentric real-world interaction data from globally distributed "human agents" offers advantages of low cost and sustainability. To this end, we propose the Always-on Egocentric (AoE) data collection system, which aims to simplify hardware dependencies by leveraging humans themselves and their smartphones, enabling low-cost, highly efficient, and scene-agnostic real-world interaction data collection to address the challenge of data scarcity. Specifically, we first employ an ergonomic neck-mounted smartphone holder to enable low-barrier, large-scale egocentric data collection through a cloud-edge collaborative architecture. Second, we develop a cross-platform mobile APP that leverages on-device compute for real-time processing, while the cloud hosts automated labeling and filtering pipelines that transform raw videos into high-quality training data. Finally, the AoE system supports distributed Ego video data collection by anyone, anytime, and anywhere. We evaluate AoE on data preprocessing quality and downstream tasks, demonstrating that high-quality egocentric data significantly boosts real-world generalization.

cs.CV↗

arXiv2026

Symmetry-imposed correlation in nuclear level statistics: The spin distribution

Despite long-term research, the origin of spin cutoff in the angular-momentum (spin) distribution of nuclear level densities remains incompletely elucidated. We demonstrate that this problem can be traced back to Bethe's assumption that nucleons in finite Fermi systems are independent random variables. By constructing a statistical ensemble that enforces rotational invariance through angular-momentum coupling, we obtain an analytical expression for the spin cutoff parameter, which includes a previously unidentified finite-population correction. Our results show that, even in the absence of interactions, nuclear many-body states exhibit non-negligible correlations arising from fermionic antisymmetry and angular-momentum coupling. From this perspective, spin cutoff may be interpreted as a quantitative measure of correlation imposed by symmetry in nuclear level statistics.

nucl-th↗

arXiv2026

JADES: Low Surface Brightness Galaxies at 0.4 < z < 0.8 in GOODS-S

Low surface brightness galaxies (LSBs) are an important class of galaxies that allow us to broaden our understanding of galaxy formation and test various cosmological models. We present a survey of low surface brightness galaxies at $0.4 < z_{\rm phot} < 0.8$ in the GOODS-S field using JADES data. We model LSB surface brightness profiles, identifying those with $\barμ_{\rm eff} > 24$ mag arcsec$^{-2}$ in the F200W JWST/NIRCam filter. We study the spatial distribution, number density, Sérsic profile parameters, and rest-frame colours of these LSBs. We compare the photometrically-derived star formation histories, mass-weighted ages, and dust attenuations of these galaxies with a high surface brightness (HSB) sample at similar redshift and a lower redshift ($z_{\rm phot} < 0.4$) LSB sample, all of which have stellar masses $\lesssim 10^8 M_{\odot}$. We find that all samples have low star formation (SFR$_{100} \lesssim 0.01$ $M_{\odot}$ yr$^{-1}$). The higher redshift LSBs and HSBs have similar star formation histories which show that the LSBs and HSBs possibly come from the same progenitors at $z \gtrsim 2$, though the histories are not well constrained for the LSB samples. The LSBs appear to have minimal dust, with most of our LSB samples showing $A_V < 1$ mag. JWST has pushed our understanding of LSBs beyond the local Universe.

astro-ph.GA↗

arXiv2026

Deformation-Recovery Diffusion Model (DRDM): Instance Deformation for Image Manipulation and Synthesis

In medical imaging, the diffusion models have shown great potential for synthetic image generation tasks. However, these approaches often lack the interpretable connections between the generated and real images and can create anatomically implausible structures or illusions. To address these limitations, we propose the Deformation-Recovery Diffusion Model (DRDM), a novel diffusion-based generative model that emphasises morphological transformation through deformation fields rather than direct image synthesis. DRDM introduces a topology-preserving deformation field generation strategy, which randomly samples and integrates multi-scale Deformation Velocity Fields (DVFs). DRDM is trained to learn to recover unrealistic deformation components, thus restoring randomly deformed images to a realistic distribution. This formulation enables the generation of diverse yet anatomically plausible deformations that preserve structural integrity, thereby improving data augmentation and synthesis for downstream tasks such as few-shot learning and image registration. Experiments on cardiac Magnetic Resonance Imaging and pulmonary Computed Tomography show that DRDM is capable of creating diverse, large-scale deformations, while maintaining anatomical plausibility of deformation fields. Additional evaluations on 2D image segmentation and 3D image registration tasks indicate notable performance gains, underscoring DRDM's potential to enhance both image manipulation and generative modelling in medical imaging applications. Project page: https://jianqingzheng.github.io/def_diff_rec/

eess.IV↗

arXiv2026

Discern Truth from Falsehood: Reducing Over-Refusal via Contrastive Refinement

Large language models (LLMs) aligned for safety often suffer from over-refusal, the tendency to reject seemingly toxic or benign prompts by misclassifying them as toxic. This behavior undermines models' helpfulness and restricts usability in sensitive or nuanced contexts. While prior work has proposed mitigation strategies such as data augmentation and activation steering, these approaches often face a trade-off: reducing over-refusal typically degrades the model's ability to reject genuinely harmful content. We argue that this issue arises from the ambiguous influence of toxic and seemingly toxic prompts on the model's learning dynamics. To address it, we introduce a preceding alignment stage, DCR: Discernment via Contrastive Refinement. Both theoretically and empirically, we demonstrate that contrastive refinement improves an LLM's capacity to distinguish truly toxic prompts from superficially toxic ones. Evaluation across diverse benchmarks shows that our method effectively reduces over-refusal while preserving the safety benefits of alignment. Importantly, it achieves this with minimal degradation of general capabilities, offering a more principled and robust direction for safety alignment.

cs.CL↗

arXiv2026

Incorporating Gibbs free energy into interatomic potential fitting

We develop a method to fit high-temperature Gibbs free energy data for the development of interatomic potentials for atomic systems. The approach is based on Hamiltonian thermodynamic integration, enabling the identification of suitable potential parameters such that the system's free energy matches a specified target. The method can be readily combined with conventional fitting techniques for properties such as elastic tensors and liquid pair distribution functions. We validate the effectiveness of the approach using the Uhlenbeck-Ford model and embedded-atom method potentials for pure Ni phases and binary Fe1-xOx liquids under high-pressure and high-temperature conditions. Our framework provides an efficient strategy for incorporating free energy into interatomic potential fitting.

cond-mat.mtrl-sci↗

arXiv2026

JWST Advanced Deep Extragalactic Survey (JADES) Data Release 5: NIRCam Imaging in GOODS-S and GOODS-N

We present the Near Infrared Camera (NIRCam) imaging products of the fifth data release (DR5) of the James Webb Space Telescope (JWST) Advanced Deep Extragalactic Survey (JADES). The JADES survey is one of the most ambitious programs yet conducted on JWST, producing deep infrared imaging and multiobject spectroscopy on the GOODS-S and GOODS-N extragalactic deep fields in order to explore galaxies to the earliest epoch. Here we describe the NIRCam data reduction procedures that result in deep and well-characterized mosaics in up to 18 filters covering 469 arcmin$^2$, with 250 arcmin$^2$ having at least 8 filters of coverage. This release contains the full NIRCam imaging of JADES, over 800 JWST mission hours, as well as co-reductions of 19 other programs in these two premier deep fields. We perform detailed tests on the final data products, thereby characterizing the photometric properties, point-spread function, and astrometric alignment. We release mosaics for individual programs (or epochs, depending on scheduling) and the mosaics combining data from all programs in order to facilitate photometric variability studies and the deepest possible photometry.

astro-ph.GA↗

arXiv2026

JWST Advanced Deep Extragalactic Survey (JADES) Data Release 5: MIRI Coordinated Parallels in GOODS-S and GOODS-N

Medium to ultra-deep mid-infrared imaging surveys with the James Webb Space Telescope (JWST)'s Mid-Infrared Instrument (MIRI) are reframing our view of the early Universe, from the emergence of ultra-red dusty and quiescent galaxies to the epoch of reionization to the first galaxies. Here we present the MIRI coordinated parallels component of the JADES program, which obtained ultra-deep (155 ks) imaging at $7.7 μ$m over $\sim10$ arcmin$^2$ as well as medium depth ($\sim5-15$ ks) imaging at $7.7, 12.8$, and $15 μ$m over $\sim36$, 25, and 22 arcmin$^2$, respectively, in the GOODS-S and GOODS-N fields. This paper describes the data reduction, which combines the official JWST Calibration Pipeline with custom steps to optimize flagging of warm/hot pixels and optimize background subtraction. We further introduce a new step to address artifacts caused by persistence from saturating sources. The final, fully reduced JADES/MIRI mosaics are being released as part of JADES Data Release 5, along with prior-based forced photometry using NIRCam detection images, providing critical rest-frame near-infrared and optical constraints on early galaxy populations.

astro-ph.GA↗

arXiv2026

JWST Advanced Deep Extragalactic Survey (JADES) Data Release 5: Photometric Catalog

JADES Data Release 5 (DR5) photometric catalogs and describes the methodologies used for source detection, deblending, photometry, uncertainty estimation, and catalog curation. The catalogs are constructed from 35 space-based imaging mosaics obtained with JWST/NIRCam, JWST/MIRI, HST/ACS, and HST/WFC3, combining approximately 1250 hours of JADES imaging with extensive additional public JWST and HST observations in the GOODS fields. Sources are identified using custom signal-to-noise-based detection and deblending algorithms optimized for the depth, resolution, and complex point-spread-function structure of JWST imaging. Source centroids, shapes, and photometric apertures are determined using a new fast two-dimensional Gaussian regression method applied to detection-image profiles. We provide forced circular-aperture photometry, ellipsoidal Kron photometry, and curve-of-growth measurements for every source in every band. We introduce a new pixel-level regression framework to model photometric uncertainties as a function of aperture size and local mosaic properties, accounting for correlated noise in heterogeneous JWST mosaics. Photometric redshifts are computed using template-based fitting applied to both small-aperture photometry on unconvolved images and Kron photometry on common-PSF mosaics. The JADES DR5 catalogs supersede previous JADES photometric releases, and are publicly released through the Mikulski Archive for Space Telescopes and an interactive web interface.

astro-ph.GA↗