arXiv subjects

Xin Wang

Publications and source records attributed to Xin Wang.

At least 91 records · Page 5Linked to original sources

arXiv2026

Quiescent Galaxies at $z > 2$ from CAPERS spectroscopy and their Number densities

Recent JWST observations have revealed an overabundance of quiescent galaxies at high redshift. Photometric methods, such as photometry-derived sSFRs, $UVJ$ colors, and synthetic $ugi_s$ colors, are commonly used to identify quiescent galaxies, but their completeness and purity remain poorly constrained. Here we present a spectroscopically selected sample of 19 quiescent galaxies at $2 \leq z \leq 4$ from the CANDELS-Area Prism Epoch of Reionization Survey (CAPERS). They are selected with $M_\ast > 10^{9.5} M_\odot$ and a specific star formation rate below 20% of the inverse cosmic age. These galaxies formed 50% of their stellar mass at $z \sim 3.0$--4.6 and quenched rapidly on timescales of $\sim0.1$--0.2 Gyr. These quenching timescales are comparable to those of other high-redshift QGs but shorter than those at lower redshift. Using the spectroscopic sample as a benchmark, we assess commonly adopted photometric QG selection methods. We find completeness of 63--89% and purities of only 40--63%, with strong emission-line galaxies constituting the dominant source of contamination. Applying completeness and purity corrections calibrated from the spectroscopic sample, we measure the number density of quiescent galaxies at $2 \leq z \leq 5$. The resulting number densities exceed the predictions of most current galaxy-formation simulations at $2<z<4$, with the largest discrepancy at $M_\ast \geq 10^{10.5} M_\odot$. This discrepancy suggests that current cosmological models may not assemble and quench massive galaxies sufficiently early or efficiently.

astro-ph.GA↗

arXiv2026

ForeTac-VLA: A Forecasting-Based Tactile-Vision-Language-Action Model for Contact-Rich Robotic Manipulation

Vision-language-action (VLA) models have demonstrated strong capabilities in robotic manipulation, yet their reliance on visual perception limits robustness in contact-rich environments, where critical physical interaction states may not be visually observable. Existing tactile-enhanced VLA methods improve physical grounding using observed tactile feedback, but most remain largely reactive rather than explicitly modeling how contact may evolve. Therefore, we propose ForeTac-VLA, a forecasting-based tactile-vision-language fusion model that predicts future tactile states to guide action generation. Specifically, ForeTac-VLA encodes recent tactile observations into temporal representations and integrates them with vision-language features through bidirectional cross-attention. Further, a transformer-based forecasting module predicts multi-step future tactile states, enabling the model to reason jointly over observed and anticipated contact. Finally, the fused multimodal representations and predicted future tactile states are fed into the VLA backbone to condition action generation. To stabilize training, a ground-truth-to-prediction curriculum is employed when early forecasts are unreliable. Across four real-world contact-rich manipulation tasks, ForeTac-VLA achieves an average success rate of 95%, outperforming the fine-tuned VLA model by 36.25 percentage points and state-of-the-art tactile-enhanced VLA baselines by over 22 percentage points. ForeTac-VLA also maintains strong performance under low-illumination and visually cluttered conditions. Video demonstrations can be found on https://foretac-vla.github.io/

cs.RO↗

arXiv2026

Almost One Bit Violation of Minimum-Output Rényi Entropy Additivity Simultaneously at All Orders

We prove that minimum-output Rényi-entropy additivity can fail by almost one bit simultaneously at every nonnegative order. For every $\varepsilon\in(0,\log2)$, there exists a finite-dimensional quantum channel with a real Stinespring isometry such that the same maximally entangled input witnesses a tensor-square entropy gap of at least $\log2-\varepsilon$ for all $p\in[0,\infty]$. The output dimension can be chosen to be $O(\varepsilon^{-3})$ as $\varepsilon\downarrow0$. The construction uses direct products of free groups: tensorized Haagerup estimates control the one-copy outputs, while commutation between distinct factors forces exact Bell-branch collisions at two copies. Strong convergence gives both an existential realization through finite-dimensional representations of right-angled Artin groups followed by realification, and a Haar-orthogonal model whose success probability tends to one as the matrix dimension grows. We also determine the exact Bell quotient, prove asymptotically sharp regular-radius bounds, and show that the cubic output-dimension scale is optimal within the present purity--rank certificate.

quant-ph↗

arXiv2026

MSA-3D: A Diversity of Dust Attenuation Profiles Across the Epoch of Thin Disk Emergence

We present spatially resolved measurements of dust attenuation and star formation in 18 main-sequence star-forming galaxies at z$\sim$1 from the MSA-3D survey, obtained by mapping the Balmer emission lines at $\sim$1 kpc resolution with JWST/NIRSpec's MSA in a slit-stepping strategy. We investigate the diversity of radial attenuation profiles, and how the spatial variation affects attenuation and star formation rates (SFR) derived from single-aperture measurements. We find a notable diversity among radial attenuation profiles: some galaxies exhibit centrally peaked attenuation, but the majority exhibit flat or even positive radial profiles, with large variation at a fixed stellar mass. This diversity may reflect different evolutionary pathways shaped by various mechanisms such as disk settling, merging, and internal processes. We examine possible biases arising from single-aperture and integrated measurements and find that, while they can under- or over-estimate attenuation and SFRs for individual galaxies, the sample-averaged trends remain roughly unchanged, with the derived SFRs consistent with the star-forming main sequence, and a small scatter. From our sample, we find a median stellar-to-nebular reddening ratio f = E(B-V)$_{\rm star}$/E(B-V)$_{\rm gas}$ of 0.88 with an interquartile range of 0.51-0.96, suggesting relatively uniform dust distributions even in intermediate-mass galaxies (stellar masses $\sim 10^9$-$10^{10.5}~M_{\odot}$). Our results highlight the importance of spatially resolved attenuation measurements for accurately tracing star formation and understanding the evolving dust geometry in galaxies during a critical epoch of morphological transformation.

astro-ph.GA↗

arXiv2026

An End-to-End Automated Pipeline for Controllable Crack Data Synthesis

Vision-based crack inspection depends on segmentation networks whose reliability depends on the quantity, diversity and label quality of their training data. Pixel-level annotations are costly, and crack images of specific structures are scarce. Generative augmentation can supply additional data, but existing methods address isolated steps. They reuse annotated masks, offer limited control over crack geometry, and adopt the conditioning mask as the label without checking it. This paper presents an end-to-end pipeline that produces labelled crack data without manual annotation and assesses the reliability of these data and of the detectors trained on them. Procedurally sampled Bézier skeletons with guaranteed geometric properties are converted into crack masks by a generative adversarial network (GAN). A dual-ControlNet Stable Diffusion model renders the masks as crack images, either on text-described surfaces or on user-provided backgrounds. An ensemble of segmentation networks trained on real images combines its agreement with the inherited label and its internal disagreement into a pixel-wise label confidence. This confidence weights the training loss instead of removing samples with a threshold. The trained detectors are evaluated with image-space probability of detection (POD) and calibration analyses. On CRACK500 and CrackTree200, the pipeline improves five segmentation networks over conventional, diffusion-based and flow-matching-based augmentation, and on CRACK500 confidence weighting yields a higher accuracy than threshold filtering at every tested threshold. On CRACK500, the crack width that U-Net detects with 90\% probability at 95\% confidence decreases from 8.0 to 4.3 pixels, and the expected calibration error decreases from 14.2\% to 9.6\%.

cs.CV↗

arXiv2026

Quantum Compiler Design for Fault-Tolerant Quantum Computing

Scalable quantum computation is expected to rely on fault-tolerant quantum computation (FTQC), in which quantum error correction (QEC) suppresses physical errors sufficiently to support reliable logical operations. This requires quantum compilation to move beyond general-purpose circuit optimization toward encoding-aware and protocol-structured compilation across the full stack of fault-tolerant quantum computers. Beyond circuit synthesis and hardware mapping, an FTQC compiler must lower algorithm-level operations into the logical gate set supported by the chosen code, coordinate encoded data and ancilla resources, realize logical operations together with repeated syndrome extraction under hardware constraints, and provide the resulting measurement stream to real-time decoding. This survey presents a full-stack view of compiler design for QEC-protected quantum computation. We organize existing work into three interacting layers: logical-level QEC compilation, physical-level QEC realization, and decoder runtime integration. At the logical level, we review surface-code lattice-surgery compilers, beyond-surface-code code-surgery frameworks including emerging qLDPC approaches, and compilation support for non-Clifford operations such as magic-state distillation and code switching. At the physical level, we survey hardware-aware QEC realization on superconducting, trapped-ion, and neutral-atom platforms. We further examine decoder models, real-time decoding systems, and frame-management mechanisms that close the feedback loop during fault-tolerant execution. Finally, we identify open challenges in cross-layer optimization, qLDPC compilation, compiler-decoder co-design, runtime adaptivity, and the development of integrated and benchmarkable FTQC compilation stacks. An actively maintained paper list is available at: github.com/chenghongz/QEC-compiler-design.

quant-ph↗

arXiv2026

Policy and World Modeling Co-Training for Language Agents

Reinforcement learning (RL) improves large language model (LLM) agents by teaching them which actions lead to high rewards, but provides little supervision on what those actions do to the environment. World modeling (WM) can fill this gap, yet existing approaches often require separate simulators, extra training stages, or additional inference-time computation. We observe that on-policy RL rollouts already contain the needed signal: each transition pairs an action with its resulting next observation. Based on this observation, we propose PaW, a Policy and World modeling co-training framework that adds auxiliary WM supervision to the same policy during RL, without changing the inference paradigm. To make auxiliary WM supervision informative and stable, PaW introduces three components: action-entropy-based WM data selection, noise-tolerant WM loss, and reward-adaptive loss balancing. Experiments on three agentic task benchmarks show consistent improvements over strong RL baselines across models and RL algorithms. These results suggest that standard RL rollouts are a practical source of WM supervision for language-agent training.

cs.LG↗

arXiv2026

Per-Matrix Optimality Is Not Enough: Three-Level Optimization for Low-Rank LLM Compression

Per-matrix singular value decomposition (SVD) truncation is Eckart-Young optimal in the whitened Frobenius norm, but errors from independently compressed matrices compound through the block's nonlinear forward pass. Inspired in part by hierarchical variational optimization in quantum many-body methods, we introduce a three-level chain that widens optimization scope from individual matrices to Transformer blocks to the full model: whitened SVD~(L1), block-level joint optimization~(L2), and end-to-end language-modeling loss refinement~(L3), all from 256 calibration sequences, with no instruction or recovery data. On LLaMA-7B at 60% compression, the chain reduces WikiText-2 perplexity from 42.1 to 19.1 to 11.4. The block-level stage acts as a regularizer: skipping it worsens Penn Treebank (PTB) perplexity by 24 points, a gap that additional end-to-end training did not close in our experiments. Perplexity gains hold across 20-80% compression, five architectures up to 13B parameters, and both in-distribution and out-of-distribution benchmarks, though the cross-architecture rows use architecture-specific configurations and the ratio sweep was not run under one common protocol. With more calibration data, skipping the block-level stage becomes competitive, revealing an offline compute--data trade-off. We therefore claim improvements only in perplexity and compression fidelity; downstream accuracy remains well below the dense model.

cs.LG↗

arXiv2026

Phase-Stable Hologram Updates for Large-Scale Neutral-Atom Array Reconfiguration

Dynamic holographic optical tweezers provide a programmable route to array assembly and reconfiguration essential for scalable neutral-atom quantum computation. However, phase mismatch between successive holograms can cause destructive interference during finite spatial light modulator (SLM) refresh. In this work, we analyze finite SLM refresh to establish a phase-stability criterion for prescribed trap amplitudes and develop a weighted-projective Gerchberg-Saxton (WPGS) method that efficiently approximates the corresponding phase-only complex-field optimization. Enforcing this phase constraint also reduces the number of iterations required for each update, achieving convergence within five iterations and enabling hologram generation within a few milliseconds. Numerical simulations of 2D and 3D reconfiguration involving more than $10^3$ traps, together with nonuniform-intensity interlayer transport, show that WPGS preserves endpoint intensity quality while suppressing inter-frame phase mismatch, transient intensity degradation, trap splitting, and motional heating. Similar suppression of phase mismatch and transient-intensity degradation is maintained under Gaussian SLM-plane illumination. These results establish the phase-stability criterion as a practical design principle for dynamic holographic control and scalable neutral-atom array reconfiguration.

quant-ph↗

arXiv2026

Beyond Distribution Matching: Semantics-Consistent Tabular Diffusion with Weak Semantic Priors

Synthetic tabular data can match real data distributions while still violating the semantic constraints that govern valid tabular rows. This reveals a key limitation of existing tabular generators: they mainly optimize distributional fidelity, but do not explicitly model weak semantic priors encoded in tabular schema and textual descriptions. In this paper, we propose \ours, a semantics-consistent tabular diffusion framework for high-fidelity synthetic data generation under weakly specified semantic priors. \ours\ first constructs two types of priors, namely intra-column semantics and inter-column symbolic rules, with LLM-assisted extraction from metadata and validation on the real training split. These priors are then used as generation conditions rather than post-hoc filters. Specifically, \ours\ maps heterogeneous column values, column identities, and semantic priors into a unified semantic space, and performs column-wise forward corruption and prior-conditioned reverse denoising to preserve both marginal distributions and rule-consistent cross-column dependencies. Extensive experiments on six real-world tabular benchmarks show that \ours\ consistently improves distributional fidelity, semantic consistency, and downstream task utility over representative VAE-, GAN-, LLM-, and diffusion-based baselines. Additional analyses further demonstrate the robustness of \ours\ when semantic priors are partially unavailable.

cs.LG↗

arXiv2026

Convex Order and Moment Persuasion

If a probability measure is a mean-preserving contraction of another, the two measures are said to be in convex order. This paper provides a complete characterization of mean-preserving contractions and their extreme points in the multidimensional setting, extending the results of Kleiner, Moldovanu, Strack, and Whitmeyer (2024) to incorporate fully revealing regions and lower-dimensional pooling regions. A central feature of our characterization is a "concentration " structure: each state can be mapped only to posterior means lying within its own irreducible component. We apply these results to the moment persuasion problem.

econ.TH↗

arXiv2026

The quasi-star model for Little Red Dots: potential and challenges

(Abridged) Little Red Dots (LRDs) are a class of sources discovered by JWST observationally defined by a "V-shaped" rest-frame UV-Optical SED, a compact or unresolved morphology, and for having, frequently, broad hydrogen emission lines. Among various models, those involving a quasi-star interpret LRDs as an intermediate stage in the evolution of a super-massive black hole (SMBH) seed into a classic AGN. In this paper, we employ the radiative-transfer code \texttt{Cloudy} to study whether this model is able to reproduce the spectral features commonly observed in LRDs. The model consists of an accreting SMBH ($M_{\rm BH}\sim10^{5-6} \ M_\odot$) surrounded by a convective layer where a black-body (BB) spectrum with $T\sim5000 \ {\rm K}$ and $L\sim10^{44.4} \ {\rm erg \ s}^{-1}$ is produced. This BB is then reprocessed by a concentric thick ($ΔR\sim1000 \ {\rm AU}$) shell of dense ($n_{\rm H}\sim10^{11} \ {\rm cm}^{-3}$) gas partially ionised by thermal collisions. The emerging radiation is further reprocessed by a diffuse clumpy medium surrounding the quasi-star. We fit this model to JWST/NIRSpec spectra of LRDs from the literature, deriving the main physical parameters and the SMBH masses. Once coupled with the UV emission from a host galaxy, this model is able to reproduce the shape of the UV-to-NIR continuum, including the presence of a Balmer break, as well as the luminosity of the hydrogen emission lines. However, this quasi-star model does not natively account for the presence of broad helium lines and for the possible presence of hot dust, needing additional components to match these observables. Our main result is to show how some LRDs can be modeled as quasi-stars, highlighting that a significant degeneracy exists among different LRD models. This has important consequences for our understanding of the mechanisms driving black hole growth in the early Universe.

astro-ph.GA↗

arXiv2026

SAEExplainer: Interpreting SAE Features with Activation-Guided Preference Optimization

Although Sparse Autoencoders (SAEs) have mitigated the opacity of large language models (LLMs) by decomposing dense representations into sparse features, explaining these features still remains a central challenge. Current explanation methods, however, typically operate within an open-loop paradigm, failing to leverage mechanistic feedback for further refinement. In this paper, we propose SAEExplainer, a training framework that utilizes activation scores as an objective reward signal to train the model for self-correction and iterative bootstrapping. By iteratively verifying and correcting foundational explanations through a two-round optimization process, SAEExplainer achieves continuous improvement in its explanatory capabilities. This mechanism significantly reduces explanation hallucinations and reinforces causal triggering patterns. Extensive experiments demonstrate our approach improves upon established baselines across most metrics, especially in causal triggering and discriminative activation. The code is available at https://github.com/he-jingyi/SAEExplainer.

cs.CL↗

arXiv2026

In Defense of Using Worst-case Privacy Disclosure as Privacy Evaluation Metric of Voice Anonymization

The voice anonymization community mainly uses Equal Error Rate (EER) to evaluate the performance of voice identity protection. While alternative metrics such as privacy-ZEBRA and a rank-based metric have been proposed, their underlying assumptions and differences may not be well known, especially to newcomers. This paper is motivated to fill the gap. Based on the concept of Shannon's perfect secrecy (or privacy), this paper positions itself as a defense of the privacy-ZEBRA framework. While no new metric is proposed, this paper explains how an `ideal' system in terms of EER may fail to gauge the information leakage on individual speakers in the log-likelihood ratio (LLR) space. The paper also shows how the rank-based metric can be cast into a metric that follows the same principle of perfect secrecy and how their best solutions are equivalent. Furthermore, the paper explains how the method of estimating LLRs may affect the evaluation results. These discussions are, to the best of the authors' knowledge, not explored or explained in detail in existing papers. Last but not least, the findings are demonstrated on simulated and VoicePrivacy Challenge data.

eess.AS↗

arXiv2026

Mr.LHDR: A Benchmark for Multimodal Real-World Long-Horizon Deep Research Agents

Deep research agents are increasingly capable of web search, tool use, multimodal evidence analysis, and information synthesis. However, existing benchmarks mainly evaluate medium-horizon exploration and rarely test whether agents can sustain long, dependency-heavy research processes. We introduce Mr. LHDR (Multimodal real-world Long-Horizon Deep Research), a benchmark for evaluating real-world deep research over long, irreducible chains of interdependent evidence across eight categories. Each question is constructed from a hidden Node-Relation graph and requires an average of 12.1 necessary intermediate conclusions with a mean dependency depth of 10.4 before reaching a short, unique, and verifiable answer. Questions incorporate multimodal evidence, including images, maps, PDFs, logos, charts, tables, and video frames, with at least one non-text element that changes the reasoning state. Mr. LHDR evaluates both final answers and the correctness of intermediate conclusions under annotated dependencies. We evaluate general models, deep research systems, and agent frameworks using Overall Accuracy (OA), Strict Accuracy (SA), Checklist Score (CS), and Dependency-Aware Checklist Score (DACS). Results show that even the strongest system achieves only 43.1% OA and 34.3% SA, indicating that final-answer accuracy substantially overestimates complete research success. Removing images reduces DACS by 12.6 points, demonstrating the importance of multimodal evidence, while SA consistently declines as reasoning chains become longer. These findings reveal sustained, dependency-consistent evidence integration, rather than isolated fact retrieval, as a key bottleneck for current deep research agents.

cs.AI↗

arXiv2026

Pre- and Post-Treatment Brain Metastases Segmentation Using nnU-Net with Post-Processing for BraTS 2026

Brain metastases exhibit high inter-lesion variability in size, enhancement pattern, and post-treatment appearance, making volumetric segmentation of both pre- and post-treatment cases the central challenge of the BraTS 2026 Task 1 (Brain Metastases). We build a pragmatic pipeline on a 5-fold nnU-Net ResEnc-L ensemble, in which each fold is trained independently for 1,000 epochs with the standard Dice + cross-entropy loss on 1,296 four-modality training cases. This ensemble is followed by a rule-based post-processing cascade tuned for the lesion-wise Dice similarity coefficient (LW-DSC), a detection-oriented metric that behaves very differently from the traditional global Dice. The final pipeline reaches an LW-DSC of 0.733 / 0.751 / 0.713 / 0.549 on the enhancing tumour (ET), tumour core (TC), whole tumour (WT), and resection cavity (RC) sub-regions on the official validation leaderboard. Rather than trusting these leaderboard gains, we audit every post-processing stage with a five-fold out-of-fold (OOF) analysis with no model-training leakage over all 1,296 training cases, scored with the official BraTS evaluation code (BraTS_evaluation): it confirms two stages as robust, per-fold-consistent improvements while the third improves only the leaderboard and does not reproduce out-of-fold. We further provide a mechanistic analysis of the LW-DSC metric that explains why recall-recovering post-processing carries low risk whereas component deletion does not, and we report thirteen negative results spanning loss engineering, alternative backbones, and inference-time settings, several of which run counter to widely held intuitions. Source code is released under Apache-2.0 at https://github.com/hornbeamliu/brats2026-met.

cs.CV↗

arXiv2026

An exchange-assisted entangling gate between 87Rb and 171Yb Rydberg atoms

Neutral-atom tweezer arrays support scalable quantum information processing. Dual-species $^{87}\mathrm{Rb}$--$^{171}\mathrm{Yb}$ arrays combine long-lived ytterbium nuclear-spin data qubits with fast, species-selective rubidium ancilla control and readout. However, realizing interspecies gates without inducing destructive Stark mixing in divalent atoms remains an outstanding problem. Here, we identify an optically accessible $S{+}S\leftrightarrow P{+}P$ Förster resonance at zero electric field, providing strong dipole-dipole exchange at array pitch. Using a shaped optical pulse under finite control response, we demonstrate a $0.36\,μ\mathrm{s}$ exchange-assisted controlled-$Z$ gate with an intrinsic fidelity of $99.91\%$, remaining above $99.85\%$ under bounded perturbations. We also identify an auxiliary repulsive van der Waals channel, providing a comprehensive toolbox for hybrid quantum processors.

quant-ph↗

arXiv2026

Learning Quantum Matter through Attention in Complex Space

Magnetic many-electron wavefunctions require amplitude and phase to be optimized together. Whether a complex internal representation improves this variational search is a practical question for neural wavefunction design. We introduce Complex Psiformer for interacting electrons in a magnetic moiré continuum, combining complex hidden features and Hermitian-magnitude attention with magnetic boundary conditions and fermionic antisymmetry. After the same number of optimization steps, Complex Psiformer reaches lower energies than Real Psiformer in two finite supercells. Both Psiformers also improve on their respective neural Hartree-Fock references. Across five training seeds in the 25-cell system, the mean Complex advantage is 1.458 meV per electron, with a smaller observed spread. A separately trained two-electron Complex state has a smaller energy gap to a finite configuration interaction reference than its Real counterpart. In the Complex states, flux scans show nonmonotonic density correlations and weaker honeycomb mean-density modulation at higher flux, while connected fluctuations persist. Gauge invariant current maps provide a qualitative comparison of local circulation in the optimized states. These benchmarks support the combined architecture as a variational ansatz for studying energies and charge arrangements in finite magnetic systems.

cond-mat.str-el↗