arXiv subjects

Xin Zhang

Publications and source records attributed to Xin Zhang.

At least 145 records · Page 8Linked to original sources

arXiv2026

Nuclear shell evolution near N = 6, 14, 20 and 28: insights from nuclear charge radii of short-lived nuclei derived from binding energies

A deep understanding of the evolution of nuclear shell structure correlating with the nucleon number is crucial for unraveling the fundamental properties of the nuclear structure and for exploring new nuclear physics phenomena far from the $β$-stability line. Although significant progress has been made in probing nuclear shell evolution via the measurements of nuclear root-mean-square charge radii, $R_{\text{ch}}$, the scarcity of new data for short-lived and exotic nuclei due to the increasing difficulty of measurements presents a formidable challenge in obtaining deeper and more universal insights into the nature of shell evolution. To mitigate this issue, we develop an improved method, accounting for the exchange term, charge-symmetry breaking effect, and odd-even staggering effect in the Coulomb energy formulation compared with that proposed by Liu et al. [Phys. Lett. B 872, 140046 (2026)], to determine unmeasured $R_{\text{ch}}$ values. Using the improved method, the $R_{\text{ch}}$ values of 59 nuclei are determined from their measured binding energies ($B$) and the respective $B$ and $R_{\text{ch}}$ of their mirror partners. We then systematically study the shell evolution near $N=6$, 14, 20 and 28 (sub)shells by placing the newly obtained $R_{\text{ch}}$ values into the corresponding isotopic chains. More comprehensive insights into the properties of nuclear shell evolution, particularly for the neutron-deficient sectors of the studied shell regions, e.g., $p$, $sd$ and $pf$ shells, are acquired, advancing our understanding of nuclear shell evolution in the light and intermediate mass region.

nucl-th↗

arXiv2026

GCN-DevLSTM: Path Development for Skeleton-Based Action Recognition

Skeleton-based action recognition (SAR) in videos is an important but challenging task in computer vision. The recent state-of-the-art (SOTA) models for SAR are primarily based on graph convolutional neural networks (GCNs), which are powerful in extracting the spatial information from skeleton data. However, their ability to capture temporal dynamics remains limited. To address this, we propose the G-Dev layer, which leverages path development-a principled and parsimonious representation for sequential data based on Lie group structures-to enhance temporal modeling. By integrating the G-Dev layer, the proposed DevLSTM module summarizes local temporal dynamics, reducing the time dimension while retaining high-frequency information. It can be conveniently applied to any temporal graph data, complementing existing advanced GCN-based models. Our empirical studies on the NTU-60, NTU-120 and Chalearn2013 datasets demonstrate that our proposed GCN-DevLSTM network consistently improves the strong GCN baseline models and achieves competitive performance. The code repository is publicly available at https://github.com/DeepIntoStreams/GCN-DevLSTM.

cs.CV↗

arXiv2026

HIcosmo: a differentiable JAX-based framework for cosmology inference

The Stage IV cosmological surveys, such as Euclid, LSST, DESI, and SKA, will deliver observational data of unprecedented volume, calling for efficient and reliable inference tools. This paper presents HIcosmo (High-performance Inference for Cosmology), an open-source JAX-based framework for cosmology inference. In HIcosmo, the forward model, distance integrals, likelihood evaluations, posterior sampling, and Fisher forecasts are all built from JAX primitives, so that gradients and Hessians of the log-likelihood are obtained directly by automatic differentiation, without any finite-difference approximation. The framework implements the $Λ$CDM, $w$CDM, $w_0 w_a$CDM, and interacting dark-energy models, and provides likelihoods for Type Ia supernovae (Pantheon+, DES-SN5YR, Union3), baryon acoustic oscillations (DESI DR1/DR2, SDSS), Planck 2018 distance priors, local $H_0$ measurements, and strong-lensing time delays. Its scope is restricted to background cosmology, with Boltzmann solvers and full perturbation-level likelihoods left to external tools. We validate HIcosmo against the reference implementation of each likelihood and against Cobaya. $χ^2$ values agree to absolute differences of $10^{-6}$-$10^{-2}$, and the marginalized constraints from the two codes differ by less than $0.2σ$ in every analysis tested. Leveraging just-in-time compilation and automatic differentiation, HIcosmo achieves about $8.7\times$ the end-to-end sampling throughput of Cobaya on CPU. As the dataset grows to survey scale, GPU acceleration over CPU reaches up to $20\times$. As applications, we present multi-probe $Λ$CDM joint constraints, dark-energy equation-of-state constraints, and Fisher forecasts for six 21 cm intensity-mapping surveys, including SKA1, MeerKAT, BINGO, Tianlai, and CHIME.

astro-ph.CO↗

arXiv2026

Exploring the Cosmic Dawn through the 21 cm Forest and High-redshift Radio Sources with the SKA

The 21~cm forest, seen as absorption features in the spectra of distant radio sources, is produced by intervening neutral hydrogen and offers a direct probe of the neutral intergalactic medium during the epoch of reionization (EoR). Because it is sensitive to small-scale structure and gas temperature, it can constrain the thermal history of the early Universe and physics that affects structure formation. Detecting individual absorption lines is challenging, mainly because of their weakness and the scarcity of high-redshift radio-bright sources. Recent progress, however, has made 21~cm forest studies increasingly feasible: new statistical observables can improve sensitivity within realistic observing times, updated radio-source counts have revised expectations for suitable background quasars, and deep-learning methods can extract physical information more efficiently. In addition, new approaches have been developed to separate astrophysical effects from early galaxies from fundamental-physics effects on small-scale structure. With the Square Kilometre Array (SKA), the 21~cm forest will therefore provide a promising route to study early heating, possible exotic energy injection, dark matter properties, neutrino mass, the running spectral index, and baryon--dark-matter relative velocity. This chapter reviews recent developments in 21~cm forest research and discusses observational strategies and prospects for constraining the first galaxies and fundamental physics with SKA-Low.

astro-ph.CO↗

arXiv2026

Cosmological preference for a positive neutrino mass at 2.7$σ$: A joint analysis of DESI DR2, DESY5, and DESY1 data

Neutrinos and dark energy (DE) have entered a new era of investigation, as the latest DESI baryon acoustic oscillation measurements tighten the constraints on the neutrino mass and suggest that DE may be dynamical rather than a cosmological constant. In this work, we obtain a high-confidence measurement of the neutrino mass within a dynamical DE framework. A joint analysis of DESI DR2, cosmic microwave background, DESY5 supernova, and DESY1 weak lensing data yields a total neutrino mass of $\sum m_ν= 0.098^{+0.016}_{-0.037}\,\mathrm{eV}$, indicating a measurement for a non-zero, positive neutrino mass at the $2.7σ$ level within the $w_0w_a$CDM framework. This high-confidence measurement is driven mainly by these factors: (i) the DESI's preference for a dynamical DE with its equation of state evolving from $w< -1$ at early times to $w> -1$ at late times, thus leading to a larger neutrino mass; (ii) treating $N_{\mathrm{eff}}$ as a free parameter together with the inclusion of weak lensing data, which likewise allows for an increased neutrino mass. In the future, even higher-confidence measurements of neutrino mass are expected with stronger preferences for dynamical DE in light of more complete DESI data releases.

astro-ph.CO↗

arXiv2026

Alleviating the $H_0$ tension through the interacting dark energy model from quantum gravitational field theory in light of DESI DR2

Recent DESI DR2 data has shown a significant preference for dynamical dark energy, yet this has further exacerbated the $H_0$ tension. In this work, we explore the potential of interacting dark energy models ($\widetildeΛ$CDM and $e\widetildeΛ$CDM) within the asymptotic-safety framework of quantum gravitational field theory to alleviate the $H_0$ tension. We perform observational constraints using the latest baryon acoustic oscillation data from DESI DR2, cosmic microwave background (CMB) data from Planck and ACT, and type Ia supernova data from DESY5 and PantheonPlus, as well as the SH0ES data. From our analysis, we observe the dynamical scale parameter of the cosmological constant, $δ_Λ = -0.270\pm 0.100$, in the $e\widetildeΛ$CDM model using the CMB+DESI+SH0ES data, which deviates from $Λ$CDM at the $2.7σ$ level. Simultaneously, we find $H_0 = 70.84\pm 0.74~\mathrm{km\,s^{-1}\,Mpc^{-1}}$, reducing the $H_0$ tension to $1.7σ$. This increase in the inferred $H_0$ is due to the anti-correlation between $δ_Λ$ and $H_0$, whereby a negative $δ_Λ$ leads to a higher $H_0$ value. Furthermore, for the CMB+DESI+SH0ES combination, we obtain $Δχ^2_{\min}=-14.14$ and $Δ\mathrm{DIC}=-9.18$, favoring the $e\widetildeΛ$CDM model over $Λ$CDM. Overall, the $e\widetildeΛ$CDM model can improve the fit and ease the $H_0$ tension, especially for the data combinations that provide the strongest statistical support.

astro-ph.CO↗

arXiv2026

Perceiving exposure segregation with open urban imagery

Socioeconomic exposure segregation -- the lack of daily interaction between income groups -- erodes social capital and entrenches inequality, yet the specific physical features that drive these behavioral restrictions remain poorly understood. Prior research has quantified where segregation occurs using mobility data, but has not identified how the built environment facilitates or inhibits these interactions. Here we introduce VISAGE, a large multi-modal model-enabled framework that perceives exposure segregation directly from open satellite and street-level imagery across 10,030 communities in 31 U.S. cities. Moving beyond black-box correlations, we operationalize cross-disciplinary sociological theory into an interpretable visual codebook to detect physical regulators of social mixing. We find that the built environment encodes a legible grammar of segregation: "defensible" architectural forms (e.g., fences, gated enclosures) and monofunctional zoning systematically predict higher social isolation, whereas mixed-use infrastructure fosters interaction, explaining substantial variance in mobility-derived segregation patterns (Pearson $r=0.770$). Crucially, we show that inclusionary housing policies manifest in distinct visual signatures associated with higher mixing, suggesting that policy interventions successfully alter the physical landscape to encourage diversity. Our findings offer a scalable pathway to decipher the social production of space, providing a mechanism-based lens to understand how the built environment shapes social behavior.

physics.soc-ph↗

arXiv2026

Gravitational radiations from periodic orbits around a black hole in the effective field theory extension of general relativity

The study of periodic orbits in extreme-mass-ratio inspirals is essential for understanding the dynamics of small bodies orbiting supermassive black holes. In this paper, we study the periodic orbits and their corresponding gravitational wave emissions within the framework of an effective field theory-based extension of general relativity (EFTGR), which incorporates higher-order curvature terms into the Einstein-Hilbert action. We start with a brief analysis of the modified black hole spacetime in EFTGR and examine how its parameters influence the dynamics of a massive neutral particle using the Lagrangian formalism. Focusing on the impact of the higher-order curvature terms in EFTGR, we examine the properties of periodic orbits, which are characterized by three topological integers $(z, w, v)$ that uniquely classify their trajectories. By analyzing these orbits within EFTGR, we aim to provide new insights into how strong-field deviations from general relativity may manifest in observable phenomena. We then calculate the gravitational waveforms generated by these periodic orbits, identifying potential observational signatures. Our analysis reveals a direct connection between the zoom-whirl orbital behavior of the small compact object and the gravitational waveforms it emits: higher zoom numbers lead to increasingly intricate waveform substructures. The results contribute to a clearer understanding of the dynamical features of EFTGR and open new avenues for probing black hole properties via gravitational wave detection.

gr-qc↗

arXiv2026

New Crystal Structures Hide in Plain Sight: A Stress Test for AI-Guided Materials Discovery

New types of crystal structures are discovered only rarely, and the artificial intelligence (AI) models now reshaping materials discovery have so far produced new chemical compositions within known structural families rather than genuinely new structures. We report GdNiSn4 and LuNiSn4, intermetallics that adopt a previously unreported structure type, found not by computation but by exploratory synthesis. Single-crystal diffraction shows that the structure is an intergrowth of two known structural units. We then use this system as a benchmark for two leading generative models, MatterGen and DiffCSP++. For DiffCSP++, the benchmark is performed in its crystallographically constrained setting, using the required space-group and Wyckoff-position inputs. Under our sampling budget, neither model recovers the experimentally reported monoclinic structure within the structural-matching tolerance. The generated structures are evaluated without further structural relaxation using the nonmagnetic analog LuNiSn4, where we rule out 4f magnetism as the cause. Because the new structure is built from familiar building blocks, it should be derivable. We argue that encoding chemical reasoning, such as the stacking of known motifs, is a concrete path toward AI that can discover structurally novel materials.

cond-mat.mtrl-sci↗

arXiv2026

Theoretical Grounding of Out-Of-Distribution Detection With Reinforcement Learning Optimizer

Out-of-distribution (OOD) detection in dynamic open-world environments requires a model to continually adapt to evolving data distributions while generalizing to covariate-shifted inputs and rejecting semantic-shifted OOD examples. Most existing OOD detection methods optimize only the current-step objective and do not explicitly account for how post-deployment environment changes affect future OOD behavior. In this paper, we establish a theoretical grounding for dynamic OOD detection using a reinforcement learning (RL)-guided optimizer that explicitly favors updates that reduce the semantic OOD false positive rate over time. We develop a novel augmented optimizer that uses an RL-guided correction term on top of standard gradient descent (GD) and show its improvement over both future-domain generalization and semantic-OOD rejection. We analyze temporal error decomposition in terms of model-change and environment-change generalization errors and develop a new theoretical framework for comparing the generalization errors under both GD and RL-guided optimizers.

cs.CV↗

arXiv2026

Synergy between CSST and future gravitational-wave detectors: Probing primordial black holes by cross-correlating dark sirens with galaxies

Gravitational-wave (GW) events and galaxies both trace the cosmic matter distribution, but the mergers of astrophysical black holes and primordial black holes (PBHs) are expected to populate different environments and therefore to cluster with different biases. The GW clustering bias is thus a statistical observable that can separate the two populations. We assess how well this can be done by cross-correlating the photometric galaxy survey of the Chinese Space-station Survey Telescope (CSST) with mock GW catalogs from two future detector networks: the third-generation ET2CE network (the Einstein Telescope and two Cosmic Explorer detectors) and the multi-band BDET2CE network, which adds the space-based baseline Decihertz Interferometer Gravitational-Wave Observatory. We find that CSST combined with 10 years of ET2CE observations can reveal a PBH contribution once its fraction in the total merger rate exceeds about $40\%$, while the much sharper sky localization of BDET2CE lowers this threshold to about $20\%$. The improvement comes from recovering the small-scale clustering information that localization errors would otherwise erase. These results show that combining future GW detector networks with CSST galaxy clustering offers a promising and largely independent route to identifying PBHs statistically.

astro-ph.CO↗

arXiv2026

Constraints on the Sum of Neutrino Masses from ACT DR6 and DESI DR2 Considering Isocurvature Initial Conditions

We present a robust assessment of cosmological constraints on the sum of neutrino masses ($\sum m_ν$) when relaxing the standard assumption of purely adiabatic primordial initial conditions. Allowing for a neutrino density isocurvature (NDI) component alongside the adiabatic mode, we analyse the latest CMB-SPA combination (Planck 2018, ACT DR6, and SPT-3G), DESI DR2 baryon acoustic oscillation data, and the DES Year 5 supernova sample. Within the $Λ$CDM model, the 95\% upper limit weakens only marginally from $\sum m_ν< 0.052$ eV (purely adiabatic) to $< 0.057$ eV (including NDI), with the NDI amplitude consistent with zero. In the CPL dynamical dark energy model, the adiabatic limit is $< 0.111$ eV, shifting to $< 0.115$ eV with NDI, yet the isocurvature mode remains undetected. While these limits are robust against the inclusion of isocurvature perturbations, they are highly sensitive to both the assumed dark energy equation of state and the prior lower bound on $\sum m_ν$. Notably, the adiabatic $Λ$CDM limit of $0.052$ eV lies below the minimum sum required by the normal neutrino mass hierarchy ($0.05878$ eV), indicating that this bound is an artifact of the statistical prior extending to zero. Imposing a physically motivated hierarchy-informed prior raises the limit to $< 0.092$ eV. Our results demonstrate that current data show no evidence for NDI modes and that the inferred neutrino mass upper limit is robust against this extension, but a definitive, model-independent bound requires addressing prior dependencies and dark energy uncertainties. This work provides the first joint constraint on $\sum m_ν$ and NDI using the full CMB-SPA+DESI DR2+DES dataset.

astro-ph.CO↗

arXiv2026

How to Train a Real-World Silicon Concierge? Internalizing Complex Business Workflow to Only OneModel

Traditional industrial agents rely on modular pipelines, including Router, Retriever, Planner, Executor, Responder, Reviewer, and other components. These systems often fracture into a labyrinth of ad-hoc patches, leading to cascading errors and high latency. We propose OneModel, an applicable paradigm shift from external workflows to internalized knowledge representation. Unlike modular systems that slice fluid user intents into static steps, OneModel consolidates complex business logic and SOPs directly into the model parameters. Through Continual Pre-training (CPT) and logic-compilation SFT, we transform fragmented business rules into intuitive model reasoning within a unified attention space. Deployed in our global financial service system, OneModel effectively breaks the trade-off between latency, accuracy, and complexity. Online A/B testing demonstrates an end-to-end latency reduction of more than 50 percent, from 18.7 seconds to 8.0 seconds, while the Intelligent Resolution Rate (IRR) increases from 64.3 percent to 83.3 percent. The results show that OneModel can replace brittle engineering logic with internalized cognitive intuition, offering a scalable blueprint for transitioning industrial agents from complex, error-prone workflows to unified model architectures.

cs.CL↗

arXiv2026

Universal geometric framework for black hole phase transitions: from multivaluedness to classification

Recent studies have revealed synchronized multivalued behavior in thermodynamic, dynamical, and geometric quantities during the black hole first-order phase transition, which enables a diagnosis from different perspectives, yet its fundamental origin has remained poorly understood. By constructing a unified geometric framework integrating real analysis and covering space theory, we reveal the universal mathematical mechanism behind this phenomenon. We prove that this multivaluedness originates from two non-degenerate critical points in the temperature function $T(r_+)$, where $r_+$ is the horizon radius, which fold the parameter space into a three-sheeted covering structure. As a direct application, we propose that a black hole undergoes a first-order phase transition if and only if its $T(r_+)$ curve has two extrema. Accordingly, we establish a classification scheme, denoted $A1$, $A2$, and $B$ for black holes. This scheme offers a complementary perspective to classifications based on global topological invariants. Our work provides a theoretical foundation for diagnosing phase transitions via multivaluedness and establishes a unified geometric perspective on black hole thermodynamics, chaotic dynamics, and spacetime structure during first-order phase transitions.

gr-qc↗

arXiv2026

The Curse and Blessing of Mean Bias in FP4-Quantized LLM Training

FP4 training promises substantial memory and compute savings for large language models, but remains fragile because blockwise quantization is dictated by extreme activation magnitudes, which inflate dynamic range and compress long-tail signals. We identify a counterintuitive source of this failure: dominant activation outliers are not merely arbitrary sparse events, but are largely induced by a coherent rank-one mean bias, whose direction aligns with the leading anisotropic spectral component. This mean component strengthens during training, is amplified and reshaped by attention and FFN operators, and increasingly dominates top activation magnitudes. Crucially, this discovery reveals that a seemingly complex outlier-suppression problem admits a truly simple solution: isolate the coherent mean before quantization. We therefore propose Averis, a mean-residual splitting quantization method that separates the mean component using only reductions and elementwise subtractions before FP4 quantization. Across Qwen3 0.6B Dense trained on 100B tokens and Qwen3 7B A1.5B MoE trained on 50B tokens, Averis enables robust W4A4G4 FP4 training, reducing BF16 loss gaps to 1.19%/0.81% versus 2.05%/1.10% for NVIDIA's recently released Hadamard-based outlier-smoothing method, while limiting downstream gaps to 0.89/0.71 points. With only 2.20% end-to-end overhead over vanilla NVFP4, about 30% of NVIDIA's Hadamard-based design, Averis provides a hardware-efficient path to stable low-bit LLM training. Complementary to Hadamard, Averis further reduces the Qwen3-0.6B loss and downstream gaps to 0.94% and 0.73 points when combined. Code is available at: https://anonymous.4open.science/r/averis-504D.

cs.LG↗

arXiv2026

Tests of general relativity at the fourth post-Newtonian order with GW230627 and GW250114

Gravitational wave (GW) observations provide an unprecedented laboratory for testing general relativity (GR) in the strong-field, highly dynamic, and relativistic regimes. Within the parameterized post-Newtonian (PN) formalisms, waveform generation tests have conventionally been limited to constraining inspiral coefficients up to the 3.5PN order. Leveraging the recent theoretical breakthrough that extended the analytical compact binary phasing to the 4.5PN order, we present the first observational constraints on these higher-order effects. Our analysis utilizes two exceptional events detected by the LIGO-Virgo-KAGRA (LVK) network: GW250114\_082203, which boasts the highest signal-to-noise ratio (SNR) recorded to date, and GW230627\_015337, which features a uniquely prolonged inspiral phase and the highest inspiral phase SNR to date. By performing Bayesian inference on the dimensionless deviation parameters ($δϕ_i$) associated with the 4PN and 4.5PN coefficients, we find that our results are fully consistent with the predictions of GR. While the current 90\% credible intervals for the four deviation parameters are of order $\mathcal{O}(1) \text{-} \mathcal{O}(10)$, the general relativistic null values ($δ\hatϕ_a= 0$) are entirely encapsulated within the bounds. This investigation establishes the first empirical baseline for 4PN and 4.5PN inspiral tests of GR, paving the way for high-precision null tests of GR with current and next-generation GW detectors.

gr-qc↗

arXiv2026

GEAR-VLA: Learning Geometry-Aware Action Representations for Generalizable Robotic Manipulation

Vision-Language-Action (VLA) models achieve strong benchmark performance but still struggle in real-world deployment with unseen objects, background shifts, and different robot embodiments. We argue that this stems from the lack of a unified geometry-aware manipulation representation, leaving existing VLAs vulnerable to low-level trajectory supervision, misaligned 3D features, and embodiment differences. To address this, we propose GEAR-VLA, a VLA framework for learning unified geometry-aware action representations for generalizable robotic manipulation. GEAR-VLA adopts coarse-to-fine action learning, where multi-source embodied pretraining equips the VLM with embodied reasoning and discrete action understanding before latent action tokens connect action semantics to a gradient-decoupled DiT continuous action expert. It further performs semantic-aligned 3D integration by aligning a trainable 3D spatial backbone with the VLA representation while freezing the original VLM-aligned visual pathway. To share this representation across robots, GEAR-VLA uses embodiment canonicalization, where embodiment-aware states and embodiment-invariant actions confine robot differences to the low-level interface. Extensive simulation and real-world experiments demonstrate strong generalization: GEAR-VLA achieves state-of-the-art performance on LIBERO, zero-shot LIBERO-Plus, and RoboTwin 2.0, reaches 85.9% success on AgileX and 81.0% on the pretraining-unseen LDT-01 embodiment, and obtains 90.1% success on a 6,360-trial universal grasping benchmark with 212 unseen objects. Code and models will be released at https://github.com/babynabeauty/GEAR-VLA.

cs.RO↗

arXiv2026

Nonlinearization of bilinear equations of the sine-Gordon type, nonlinear Schrödinger type and Benjamin-Ono type

This is a continuation of the paper [Commun. Theor. Phys., 77 (2025) 115006] on the nonlinearization of bilinear equations. The sine-Gordon type and nonlinear Schrödinger type bilinear equations are introduced by Jarmo Hietarinta during his search for integrable bilinear equations. In this paper, we provide a formulation to convert these two types of bilinear equations into nonlinear forms. In addition, the nonlinearization related to the equations involving the Hilbert transformations is also considered. Bell polynomials are employed in the nonlinearization and illustrative examples are provided.

nlin.SI↗