Searcharxiv⌕ Search

arXiv subjects

Yu Shi

Publications and source records attributed to Yu Shi.

At least 19 recordsLinked to original sources
arXiv2026

Tracing Gluon Saturation through Hadronization at EIC

Gluon saturation provides a window into the nonlinear nature of the strong interaction in nuclear matter. One direct consequence of saturation is the $\boldsymbol{k}_T$ broadening in the final state.We investigate how hadronization reshapes conventional signatures of gluon saturation at EIC within a complete event-generator framework. To this end, we implement in eHIJING an initial-state-radiation algorithm based on nonlinear small-$x$ evolution and complete the events with beam remnants, final-state radiation, and hadronization. In our simulations, the signal of parton-level nuclear $\boldsymbol{k}_T$ broadening is strongly diluted in both the nucleon energy correlator and leading-dihadron azimuthal decorrelation after hadronization. Global hadronic recoil, by contrast, remains sensitive to the underlying $\boldsymbol{k}_T$ broadening. We further demonstrate that Bayesian unfolding of the global hadronic recoil provides access to the underlying hard-scattering $\boldsymbol{k}_T$ distribution. These results establish the global hadronic recoil as a promising saturation observable at the EIC.

hep-ph↗
arXiv2026

Holographic subregion complexity in insulator/superconductor transition

We study holographic subregion complexity (HSC) across a fully backreacted insulator/superconductor transition in an AdS-soliton background and compare it with holographic entanglement entropy (HEE) and holographic complexity based on the complexity=volume (CV) proposal. Both HSC and HEE signal the second-order transition. For a strip subsystem, competing connected and disconnected Ryu-Takayanagi surfaces give rise to a confinement/deconfinement transition. At fixed chemical potential in the superconducting phase, HSC exhibits a finite jump at the critical width, whereas HEE remains continuous. Beyond this width, HSC grows linearly with the strip width, while HEE is constant. At fixed strip width, HSC first decreases and then increases with chemical potential for $\ell<\ell_c$, opposite to HEE, but increases monotonically for $\ell>\ell_c$. After consistent normalization and subtraction of the respective insulating references, the half-space HSC and CV complexity densities are analytically identical. These results show that HSC can diagnose the insulator/superconductor transition, but its qualitative response remains sensitive to the subsystem scale and entanglement-wedge topology.

hep-th↗
arXiv2026

Exploring $Z/\gamma$-mediated heavy FCNCs at the FCC-ee

The flavor structure of the Standard Model (SM) remains one of the most compelling questions in particle physics, with the third generation being particularly intriguing due to its significantly larger masses and comparatively less precisely measured properties. These features make third-generation flavor transitions particularly interesting in context of search for physics beyond the SM. In this work, we investigate flavor-violating transitions between the third and the first two generations, mediated by the neutral gauge bosons ($Z/\gamma$), within the framework of the SM Effective Field Theory (SMEFT), using dipole and Higgs-current operators. We determine the optimal sensitivities using the optimal observable technique (OOT) at different center-of-mass energies of the upcoming Future Circular Collider in the $e^+e^-$ mode (FCC-ee). We further derive complementary constraints on the relevant SMEFT operators from low-energy flavor-violating observables and heavy fermion decay channels. Our analysis also reveals characteristic interference patterns among the dipole contributions, which depend on the underlying flavor transition and exhibit distinct behavior between the $Z$ pole and higher-energy FCC-ee stages. The FCC-ee provides a complementary and direct probe of flavor-violating interactions at the electroweak scale, with the projections showing improved sensitivity for several interactions and comparable sensitivity to existing flavor constraints for several others. This highlights the importance of a systematic assessment across the different FCC-ee energy stages, which provides a comprehensive picture of its potential to explore flavor-violating phenomena and its complementarity with low-energy flavor experiments.

hep-ph↗
arXiv2026

EVOMAL: Self-Poisoning in Self-Evolving Coding Agents

Self-evolving LLM coding agents write their own tools by imitating retrieved skills from shared skill libraries. We identify a vulnerability in this loop: during authoring, a retrieved malicious skill can become the template for a new skill that preserves the payload. We call this self-poisoning: the agent authors, stores, and runs the resulting malicious skill. We exploit it through EvoMal, an attack that amplifies self-poisoning by wrapping an interchangeable payload in a banner, a set of benign-looking structural elements that induces an imitating agent to reproduce the enclosed code. The attacker plants malicious skills in the library without invoking them. The agent then authors and executes new skills carrying the harmful code. Each authored copy can re-enter the library and be imitated again, forming a self-propagating worm that persists after the planted skills are removed. We define the agent self-poisoning rate (ASPR) as the fraction of tasks that add a newly authored malicious skill to the library. Across six models on 153 tool-relevant SWE-bench Verified tasks, ASPR ranges from 20.3% to 41.8%, and the poisoned libraries hold 4.9 to 9.0 times as many malicious skills as were planted. The vulnerability also appears without a banner: DeepSeek-V4-Pro reaches 11.1% ASPR with the payload alone. Tailoring the planted skill descriptions to one task family raises ASPR to 86.7%. After the planted skills are removed, Qwen3 retains a round-5 ASPR of 68% because agent-authored copies remain. These copies evade existing defenses, which focus on attacker-submitted names, code, and signatures. We propose counter-prompt, a defense that discourages banner-style copying and reduces EvoMal's ASPR to at most 6.7% with no significant task-completion loss.

cs.CR↗
arXiv2026

SkillShield: Prompt-Space Security Skills for LLM Coding Agents

A coding agent edits files and executes shell commands with its developer's privileges, allowing malicious requests to translate directly into harmful actions or functional malware. Existing defenses have complementary limitations: weight-level alignment is unavailable to API-only deployers, whereas input filters and execution-boundary monitors require auxiliary classification or checking components along the agent's trajectory. We therefore introduce SkillShield, a system-prompt defense that synthesizes security skills offline from known attacks or recorded agent failures. These skills are injected into the system prompt at session start and remain active throughout the tool-use loop. Unlike a reference monitor, they protect the system by defining the security policies the model should follow during execution. Due to the limited system-prompt space, we examine three fixed-budget provisioning scopes: all-classes, with one skill covering all threat classes, per-bundle, with one skill targeting a related subset, and per-class, with one skill dedicated to a single known class and used as the upper-bound reference. None requires runtime request classification or routing. Across six large language models on RedCode, the default all-classes skill reduces malware-generation severity from 3.37 to 0.58 and achieves a 43.6% execution attack success rate, comparable to Llama Guard 3's 42.7% without its separate 8B classifier. The per-bundle and class-fixed per-class settings further reduce this rate to 36.2% and 14.5%, respectively. Under two non-adaptive jailbreak families, SkillShield continues to outperform all baselines on malware generation. Across 731 benign task descriptions, SkillShield yields a mean safety-refusal rate of 0.14%. These results demonstrate the potential of prompt-space security skills to prevent harmful actions and malware generation for LLM coding agents.

cs.CR↗
arXiv2026

Beyond Success and Failure: Length-Aware Contrastive Learning for GUI Agents

Graphical User Interface (GUI) agents powered by Multimodal Large Language Models (MLLMs) have shown strong potential for automating tasks across diverse digital environments, where reinforcement learning (RL) has become a dominant training paradigm. However, widely used methods such as Group Relative Policy Optimization (GRPO) suffer from reward-gradient misalignment, leading to inefficient and unstable optimization. Recent work addresses this issue by reformulating RL with verifiable rewards (RLVR) as contrastive or classification-based objectives, which improve stability by eliminating problematic gradient behaviors. Despite this progress, existing contrastive RLVR methods rely primarily on outcome-level supervision and fail to capture fine-grained differences in trajectory quality within the same outcome category. In this paper, we propose Length-Aware Contrastive Learning for GUI Agents (LACL-GUI), a contrastive RLVR framework that incorporates trajectory-level quality signals into policy optimization. LACL-GUI introduces structured preferences within both successful and failed trajectories, encouraging concise successful executions and differentiating failure quality based on divergence from successful trajectories, while preserving optimization stability. Experiments on GUI agent benchmarks show that LACL-GUI provides more effective learning signals and consistently improves agent performance over prior methods, highlighting the value of trajectory-level supervision in contrastive RLVR.

cs.AI↗
arXiv2026

Holographic subregion complexity in unbalanced St\"{u}ckelberg holographic superconductors

Within the subregion complexity-volume conjecture, we numerically compare holographic subregion complexity (HSC) and holographic entanglement entropy (HEE) for a strip in unbalanced St\"{u}ckelberg holographic superconductors. Varying the St\"{u}ckelberg parameter $\gamma$ yields both second- and first-order transitions. Both observables signal these transitions, but with markedly different robustness. The qualitative HEE signatures persist across strip widths, and the finite part of HEE remains smaller in the superconducting phase than in the normal phase. The HSC is instead strongly width dependent: its temperature trend is opposite to that of HEE at small $\ell$ and agrees with it at large $\ell$. Consequently, the superconducting and normal HSC branches reverse their relative ordering, creating a crossover region where they nearly coincide. There, HSC alone cannot reliably determine the occurrence or order of the transition, and the physical branch must be selected from the grand potential. Thus, HEE provides a more robust diagnostic, whereas HSC is a scale-dependent probe whose interpretation depends explicitly on the subsystem size.

hep-th↗
arXiv2026

Refusing Intent, Not Form: Wrapper-Based Intent-Group Supervision for LLM Safety

Safety tuning can improve harmful refusal, but models may learn surface-form shortcuts: wrapped harmful prompts bypass safety, while similarly wrapped benign prompts are over-refused. We propose Wrapper-Based Intent-Form Augmentation (WIFA), an automatic intent-group augmentation method that pairs wrapped harmful examples with structurally matched wrapped benign counterexamples, requiring no external teacher or manual per-wrapper intent labels. We use WIFA as a common data layer for two complementary fine-tuning routes: WIFA-Boost, a two-stage high-safety recipe, and Anchored Group-Consistent Refusal Training (A-GCRT), which regularizes refusal/compliance decision scores across same-intent wrappers and anchors harmful and benign groups on opposite sides of a margin. In the Qwen setting, WIFA-Boost reaches the strongest transformed-harmful refusal, while A-GCRT reduces OR-Bench over-refusal from 25.7\% for the base model to 17.4\%; reproduced baselines do not match these operating points. Llama results and ablations over data structure, two-stage order, and A-GCRT components support this intent-group interpretation without claiming universal below-base over-refusal.

cs.CL↗
arXiv2026

Gluonic nucleon energy correlators and fracture functions for Color Glass Condensate

Nucleon energy correlators (NECs) and fracture functions provide novel tools for probing nucleon and nuclear structure at small $x$ through measurements in the target fragmentation region (TFR) of deep inelastic scattering (DIS). We investigate gluonic NECs and fracture functions using the Color Glass Condensate effective theory at eikonal accuracy. We find that only the unpolarized and linearly polarized gluon components in an unpolarized target are nonvanishing at this order, and that both are determined by the adjoint dipole $S$-matrix. Furthermore, we show that the linearly polarized gluonic NEC $h_{1}^{t,g}$ generates a characteristic $\cos 2\phi$ azimuthal asymmetry in the DIS energy pattern in the TFR. Unlike analogous observables in the current fragmentation region, this asymmetry is governed by the ratio of the linearly polarized gluon NEC $h_{1}^{t,g}$ to the unpolarized quark NEC $f_{1}^{q}$, making it particularly sensitive to the saturation scale. Our numerical analysis shows that this asymmetry exhibits substantial nuclear suppression, providing a novel window into the onset of gluon saturation at the future Electron-Ion Collider.

hep-ph↗
arXiv2026

When Latents Forget Pixels: Restoring Fidelity in Diffusion Transformer Super-Resolution

Image super-resolution (SR) with large generative models has recently achieved remarkable perceptual quality, yet maintaining fidelity to the LR observation remains challenging. In particular, we observe that diffusion transformers (DiTs) built on latent representations suffer from a critical limitation: the compression bottleneck of the VAE weakens fine-grained spatial information, leading to hallucinated details that are weakly grounded in the input image. In this work, we revisit generative SR from a representation perspective and propose a pixel-grounded super-resolution (PGSR) framework that preserves LR-observed pixel evidence before VAE compression and reuses it throughout restoration. Instead of relying solely on the compressed latent condition, PGSR extracts pre-VAE pixel evidence from the upsampled LR image and reuses it at two stages. First, Condition-Side Trajectory Guidance fuses LR-derived pixel evidence with the latent LR condition to guide the latent restoration trajectory. Second, Decoder-Side Pixel Grounding injects multi-scale pixel features into the frozen VAE decoder to ground the final rendering with LR-observed cues. To efficiently adapt large pretrained DiT models, we keep the latent autoencoder and main flow-matching backbone frozen, and train only lightweight restoration modules. We further study an efficient local-window attention variant for improved high-resolution efficiency and scalability. Extensive experiments demonstrate that PGSR improves the realism--fidelity trade-off and produces more faithful, visually convincing results than existing latent generative SR approaches.

cs.CV↗
arXiv2026

WHALE: A Scalable Unified Model for Recommendation with Wukong-HSTU Architecture

As scalability becomes increasingly important in recommendation modeling, recent architectures have advanced the modeling of two broad sources of ranking signals along separate paths: non-sequence features, including user, item, context, and cross features; and sequence features from user behavior histories. Wukong and HSTU have emerged as representative scalable backbones for these paths: Wukong scales high-order non-sequence feature-interaction modeling, while HSTU scales long user-behavior sequence modeling. Despite their complementary strengths, practical architectures that combine these two types of feature modeling remain underexplored. We present WHALE, a scalable unified recommendation architecture that jointly models non-sequence and sequence features on top of Wukong and HSTU. Each WHALE layer contains a Wukong module, an HSTU module, and an attention-based fusion module in which Wukong-derived interaction representations query HSTU-derived behavior representations. This design keeps both backbones active throughout the network and enables progressive Wukong-HSTU exchange, allowing high-order feature crosses to repeatedly retrieve fine-grained evidence from long user histories. To make WHALE practical for industrial deployment, we introduce customized Triton kernels and other model-systems co-design techniques to improve training and inference efficiency. On large-scale industrial recommendation data, WHALE achieves consistent gains in offline experiments. Additionally, it delivers positive online gains with a modest serving-throughput trade-off. The method has been deployed in production systems. Overall, WHALE provides a practical example of how these two sources of information can be scalably unified in an industrial recommendation model.

cs.IR↗
arXiv2026

Regular Black Hole Formation and Gamma-Ray Burst from Matter Conversion

During the gravitational collapse of a massive star into a regular black hole, a new form of matter must be produced in order to prevent the formation of a central singularity. Since such matter is not present in the initial stellar configuration, it must emerge dynamically during the collapse. This formation process is expected to be accompanied by a strong release of energy in the form of electromagnetic radiation, which may be observable. Here we investigate the gravitational collapse of baryonic matter into Dymnikova-Hayward-Bardeen regular black holes. We estimate the radiation density and the corresponding bolometric luminosity generated by the formation of the matter sector responsible for singularity avoidance. We show that such processes provide a possible mechanism for gamma-ray bursts. Moreover, compatibility with gamma-ray burst requires small regularization effects. As a result, the corresponding regular black holes differ weakly from the Schwarzschild black hole.

gr-qc↗
arXiv2026

How a minimal length scale modifies thermodynamics of RN AdS Black Holes?

We investigate the thermodynamic modifications of the Reissner-Nordstroem anti-de Sitter (RN AdS) black hole induced by a minimal length scale,which naturally emerges in $\kappa$-deformed space-time. By constructing the modified metric via phase-space commutation relations,we derive the deformed Hawking temperature, entropy, and enthalpy. We analyze the thermal stability through the heat capacity and study the P-V criticality, revealing that the black hole undergoes a small-to-large phase transition analogous to the Van der Waals system, albeit with a critical ratio slightly lowered by non-commutativity. Furthermore, we examine the Joule-Thomson expansion and find that the non-commutative (NC) parameter expands the cooling region in the temperature-pressure plane. Our results demonstrate that while the overall thermodynamic analogy with the Van der Waals fluid persists, the minimal length scale systematically deform the coexistence region and inversion curves, offering potential observational signatures for quantum gravity.

gr-qc↗
arXiv2026

Holographic Timelike Entanglement and Subregion Complexity in Localized AdS3*S3*T4 Black Holes

We study timelike entanglement entropy and timelike subregion complexity in localized black holes with asymptotic AdS3*S3*T4 geometry, focusing on the black-pole solution. Unlike the BTZ solution, the black pole exhibits a nontrivial dependence on the internal sphere through the functions $K_y(r,\theta)$ and $G(r,\theta)$. Both observables are constructed from spacelike and timelike Lorentzian branches, but they probe the geometry in different ways: timelike entanglement yields a complex lifted area, while timelike complexity gives a real, finite renormalized volume. We employ a localized timelike prescription in which the branch profile is built at an angular label $\theta_0$ and subsequently lifted over the physical internal angle $\theta$. In the large-$r$ regime, the leading angular dependence drops out, recovering the expected short-interval behaviour. In the exact black-pole geometry, the temporal families become non-monotonic, making a fixed-boundary-interval selection essential. As the boundary interval increases, the selected branches move inward and become sensitive to the localized cap-horizon transition region. These results demonstrate that timelike Lorentzian observables probe localized-geometry effects that are absent in BTZ and in the leading large-$r$ description.

hep-th↗
arXiv2026

Super-Generalist: Towards Comprehensive and Accurate Medical Image Understanding via Generalist-Specialist Synergy

Medical images require comprehensive and accurate interpretation to support the diagnosis of diverse clincial conditions. Recent vision-language generalist models offer broad task coverage and promising zero-shot capabilities, yet often lack fine-grained anatomical and lesion awareness for reliable diagnosis and spatial interpretability. In contrast, supervised specialist models achieve strong performance on specific tasks but typically lack generalization across diseases and anatomies. In this work, we present SuG, a Super-Generalist framework that unifies generalist vision-language learning with specialist objectives, enabling both broad generalization and specialist-level diagnostic capability. We perform specialist-enhanced vision-language alignment in SuG by incorporating spatial priors from multiple segmentation experts, including anatomy, class-specific lesion and class-agnostic lesion segmentors that captures lesions beyond anatomies annotated during training. To improve lesion grounding capability, we leverage lesion masks as spatial priors to calibrate text-conditioned visual attention, encouraging disease-related semantics to focus on clinically relevant regions. We evaluate SuG on extensive chest and abdominal CT benchmarks, including CT-RATE, Merlin, MedVL-CT69K, and several in-house tumor datasets. SuG achieves state-of-the-art performance across a wide range of disease diagnosis tasks and surpasses specialist models on several critical tumor diagnosis benchmarks. Furthermore, SuG demonstrates strong lesion grounding capability, including robust generalization to lesion types lacking class-specific supervision.

cs.CV↗
arXiv2026

Multiparameter Quantum Estimation and Degeneracy Structure in Three-Flavor Neutrino Oscillations

Achieving precision measurements of neutrino oscillation parameters and resolving parameter degeneracies remain central challenges in neutrino physics. This work presents a systematic investigation of three-flavor neutrino oscillations within the framework of quantum estimation theory using the quantum Fisher information matrix (QFIM). The behavior of all six independent elements of the QFIM associated with the parameters theta23, deltaCP, and Delta(m31)^2 is analyzed, and the impact of parameter correlations on the quantum Cram\'er-Rao bound is studied. Furthermore, we demonstrate that parameter degeneracies in neutrino oscillation probabilities do not necessarily imply indistinguishability of the underlying quantum states. By employing quantum fidelity and the QFIM, we show that degenerate parameter sets can exhibit distinct quantum-information characteristics that remain hidden at the probability level, revealing quantum-state differences between probability-degenerate solutions.

hep-ph↗
arXiv2026

Beyond Uniform Token-Level Trust Region in LLM Reinforcement Learning

Reinforcement learning with verifiable rewards (RLVR) has become standard for improving LLM reasoning. However, existing PPO-style trust-region mechanisms remain position-agnostic by enforcing uniform thresholds across all tokens independently. This pointwise treatment conflicts with autoregressive generation in two critical ways. First, uniform thresholds ignore autoregressive asymmetry. Early-stage deviations produce compounding sequence-level drift, causing static thresholds to under-regulate early divergence and excessively constrain late-stage exploration. Second, evaluating token-level divergence in isolation overlooks cumulative prefix drift, granting the same divergence allowance regardless of how far the conditioning history has already deviated from the rollout policy. To address this limitation, we propose CPPO (Cumulative Prefix-divergence Policy Optimization), a token-level masking rule that aligns updates with a finite-horizon policy-improvement bound via two coupled mechanisms. First, a position-weighted threshold imposes stricter limits at early positions whose effects persist longer, relaxing constraints for late-stage tokens. Second, a cumulative prefix budget tracks historical deviations, dynamically restricting further token-level deviation to prevent compounding errors along the prefix. Empirically, CPPO enhances training stability and significantly improves reasoning accuracy across various model scales.

cs.LG↗
arXiv2026

BreastGPT: A Multimodal Large Language Model for the Full Spectrum of Breast Cancer Clinical Routine

Breast cancer remains a leading cause of cancer-related mortality among women. Its clinical management requires multimodal reasoning across a clinical workflow that spans \textit{screening}, \textit{diagnosis} and \textit{treatment planning}, where each stage involves distinct imaging modalities, task objectives, and reasoning patterns. However, constrained by data scarcity and model versatility, existing medical MLLMs are typically evaluated on isolated modalities or narrow task families, limiting their ability to support workflow-level clinical reasoning. In this work, we first introduce \textbf{BreastStage}, a workflow-aligned breast imaging instruction corpus comprising 1.86M instruction-following pairs curated from 17 sub-datasets across 5 imaging modalities and 136 task templates. Its held-out split, \textbf{BreastStage-Bench}, provides a comprehensive benchmark for evaluating multimodal reasoning across the breast cancer care continuum. Building on this corpus, we propose \textbf{BreastGPT}, a unified MLLM equipped with a dual-branch visual encoder and concept-preserving token compression to bridge the scale gap between standard radiology and gigapixel pathology. On BreastStage-Bench, BreastGPT achieves 75.66\% closed-ended accuracy and 89.92\% open-ended score, outperforming both general-purpose and medical-specific MLLMs across clinical stages and task formats. These results suggest that workflow-aligned data and cross-scale visual modeling are critical for clinically grounded medical MLLMs. All data, code, and model checkpoints are released at https://yangyy-liu.github.io/BreastGPT.io.

cs.CV↗