arXiv subjects

H. Zhang

Publications and source records attributed to H. Zhang.

At least 19 recordsLinked to original sources

arXiv2026

Towards quantum elliptic cohomology from GLSMs

This paper briefly outlines a proposal for a GLSM Coulomb-branch interpretation of the quantum elliptic cohomology rings currently being developed in mathematics by Bouaziz-Huq-Kuruvilla-Lee. The proposal is modeled on the established GLSM computations of quantum cohomology and quantum K theory of Fano toric varieties, where ring relations arise as critical loci of twisted effective superpotentials. In the case of quantum elliptic cohomology, the natural generalization would be to encode ring relations in Coulomb branch computations in two-dimensional theories obtained by KK reductions from four-dimensional theories, describing OPEs of parallel surface defects, which we review. However, in some basic examples, this interpretation is obstructed by four-dimensional gauge anomalies. In such cases, we provide a classical superpotential whose critical loci generate the expected ring relations, but which is not obtained by KK reduction from four dimensions, though it shares many characteristics with such KK reductions. We describe some simple examples, and make some predictions for the case of gerbes.

hep-th↗

arXiv2026

Apodex 1.1: Scaling Agentic Intelligence for Complex Work

General-purpose language models can reason and synthesize knowledge, but complex work also requires sustained interaction with files, information sources, and executable code, together with state maintenance, failure recovery, and verifiable delivery. We call this \emph{working capability}: sustained, verifiable progress toward a real-world objective. Apodex 1.1 develops this capability along two complementary dimensions. \emph{Environment Scaling} expands the diversity and verifiability of executable file, search, and code environments, while \emph{Agentic Coordination Scaling} trains agents to decompose long-horizon tasks, delegate parallel work, integrate asynchronous results, and replan. A shared execution harness and AgentOS maintain task state and provenance across tools and agents, and training turns environment trajectories and coordination traces into reliable behavior. Across complex professional work, finance, scientific research, mathematics, coding, and search, Apodex 1.1 reaches the leading performance band despite using a substantially smaller model than many frontier systems. The 35B-parameter Apodex 1.1 Mini further retains strong working capability in a locally deployable form. These results ground agentic intelligence in useful, verifiable work completed over time and advance our goal of building a \emph{Heavy-Duty Solver} for ambitious, long-running tasks.

cs.AI↗

arXiv2026

CoLoRe-2LPT: Lyman-$\alpha$ mock catalogues for the validation of DESI cosmological analyses

The Lyman-$\alpha$ (Ly$\alpha$) forest has become a crucial probe for studying the large-scale structure of the universe at high redshift ($z > 2$), providing powerful constraints on Baryon Acoustic Oscillations (BAO) and the full-shape (FS) clustering of matter. As a key ingredient for upcoming BAO and FS analyses, we present a new generation of fast cosmological Ly$\alpha$ mocks based on second-order Lagrangian perturbation theory (2LPT). These new mocks significantly improve upon previous log-normal approaches, both at accurately capturing small scale clustering and at recovering the non-linear broadening of the BAO peak. They are able to reproduce Ly$\alpha$ statistics within $10\%$ of the latest DESI measurement; including the Ly$\alpha$ bias and the redshift-space distortion $\beta$ parameter, mean transmitted flux, and 1D power spectrum. The corresponding quasar (QSO) clustering is also improved with respect to previous approaches, calibrated against high-resolution Abacus simulations, recovering the observational QSO linear bias to less than $5\%$ and improving redshift-space distortions via 2LPT velocities and the addition of Fingers-of-God effects. Furthermore, these mocks incorporate high column density systems and metal lines, allowing us to explore the effects and systematics induced by these astrophysical contaminants. This new set of mocks has been key for enhancing the modeling and validation of the DESI DR2 Ly$\alpha$ full shape cosmological analysis. This work provides a physically motivated and computationally efficient tool for simulating current and next-generation Ly$\alpha$ surveys and validating FS and BAO analysis.

astro-ph.CO↗

arXiv2026

Lyman-$\alpha$ forest holography: 3D predictions from 1D measurements

Cosmological analyses of Lyman-$\alpha$ forest clustering rely on either one-dimensional correlations along individual sightlines or three-dimensional correlations between different sightlines. Because these observables probe the matter distribution on very different scales, they have traditionally been analyzed independently. In this work, we bridge this gap using ForestFlow, an emulator trained on a suite of cosmological hydrodynamical simulations that provides a unified description of Lyman-$\alpha$ forest clustering from linear to nonlinear scales. This framework enables us to determine the range of three-dimensional clustering models compatible with the DESI one-dimensional flux power spectrum ($P_{\rm 1D}$). The resulting predictions successfully reproduce the large-scale clustering measured by the DESI BAO analysis and provide physically motivated priors on nonlinear clustering that are used in a companion paper presenting the full-shape analysis of the DESI DR2 Lyman-$\alpha$ forest. We validate our methodology using the large-volume, high-resolution hydrodynamical simulation ACCEL-2, demonstrating excellent agreement across the full range of scales considered. Finally, we combine constraints from the $P_{\rm 1D}$ and BAO analyses on the parameter combinations $b_\delta \sigma_8$ and $b_\eta f \sigma_8$, finding that the two probes provide comparable constraining power while exhibiting complementary parameter degeneracies. Our results establish a direct connection between one- and three-dimensional Lyman-$\alpha$ forest measurements through ForestFlow, an approach we term Lyman-$\alpha$ holography by analogy with the reconstruction of higher-dimensional structure from lower-dimensional information.

astro-ph.CO↗

arXiv2026

High Multiplicity Trigger for Long-Lived Particles in CMS detector

Searches for long-lived particles (LLPs) at the CMS experiment often involve unconventional event topologies that are difficult to efficiently select using standard trigger strategies. To improve sensitivity to such signatures during LHC Run~3 operation, a dedicated High Multiplicity Trigger (HMT) has been developed and deployed in the CMS trigger system. The trigger targets events containing unusually large numbers of hits in the CMS cathode strip chamber (CSC) muon detectors, a characteristic signature of several LLP scenarios involving displaced decays in the muon system. The HMT implementation, trigger logic, rate dependence with pileup, and operational stability are described. Optimized hit multiplicity thresholds are used to maintain acceptable trigger rates under high-luminosity and high-pileup conditions while preserving high efficiency across a broad range of LLP lifetimes and kinematic regimes. The trigger performance is evaluated using both simulated event samples and proton-proton collision data collected during Run~3 of the LHC. The HMT substantially extends the CMS sensitivity to non-standard signatures associated with LLP decays and provides a flexible platform for future searches for physics beyond the Standard Model.

hep-ex↗

arXiv2026

Mortality of ultra-thin LGADs and PiN diodes from high energy deposition

Low Gain Avalanche Diodes are prime candidates for high-resolution timing applications in High Energy Physics, Nuclear science, and several other fields. Operating these devices in high-radiation environments presents various hazards, including the risk of their permanent degradation or destruction caused by effects such as Single Event Burnout. Studies using minimum ionizing particles found a greatly reduced Single Event Burnout risk by operating below a bias voltage corresponding to an average electric field of 12 V/$\mu$m - however, as high energy particle colliders produce a wide energy spectrum of radiation, it is crucial to understand this phenomenon and other possible damage mechanisms at energy deposition levels greater than those of minimum ionizing particles. This was achieved by pre-irradiating LGADs and PiN diodes with active thicknesses of 20, 30, and 50 $\mu$m up to 1.5 $\times$ 10$^{15}$ $\mathrm{n_{eq}/cm^2}$, and exposing them to beams of protons and heavy ions (C, O, Fe, Au) at the BNL Tandem van de Graaff accelerator. Several mortality categories were observed, defined by different electrical and mechanical damage signatures. This furthers our understanding of permanent radiation damage of silicon devices, crucial towards mitigating Single Event Burnout and other damage mechanisms to safely operate future detectors.

physics.ins-det↗

arXiv2026

From Virtual Environments to Real-World Trials: Emerging Trends in Autonomous Driving

Autonomous driving technologies have achieved significant advances in recent years, yet their real-world deployment remains constrained by data scarcity, safety requirements, and the need for generalization across diverse environments. In response, synthetic data and virtual environments have emerged as powerful enablers, offering scalable, controllable, and richly annotated scenarios for training and evaluation. This survey presents a comprehensive review of recent developments at the intersection of autonomous driving, simulation technologies, and synthetic datasets. We organize the landscape across three core dimensions: (i) the use of synthetic data for perception and planning, (ii) digital twin-based simulation for system validation, and (iii) domain adaptation strategies bridging synthetic and real-world data. We also highlight the role of vision-language models and simulation realism in enhancing scene understanding and generalization. A detailed taxonomy of datasets, tools, and simulation platforms is provided, alongside an analysis of trends in benchmark design. Finally, we discuss critical challenges and open research directions, including Sim2Real transfer, scalable safety validation, cooperative autonomy, and simulation-driven policy learning, that must be addressed to accelerate the path toward safe, generalizable, and globally deployable autonomous driving systems.

cs.AI↗

arXiv2026

Flux pumping and bifurcated relaxations of helical core in 3D magnetohydrodynamic modelling of ASDEX Upgrade plasmas

Flux pumping was achieved in recent hybrid scenario experiments in the ASDEX Upgrade (AUG) tokamak, which is characterized by a sawtooth-free helical quiescent state and the anomalous radial redistribution of toroidal current density and poloidal magnetic flux. In this article, the self-regulation mechanism of the AUG core plasma during flux pumping is investigated at realistic parameters using the JOREK code based on the two-temperature, nonlinear, full magnetohydrodynamic (MHD) model. A key milestone in AUG flux pumping modelling is achieved by quantitatively reproducing the clamped current density and safety factor profiles in the plasma core, demonstrating the effectiveness of the dynamo effect in sustaining the flux pumping state. The dynamo term, that is of particular interest, is primarily generated by the pressure-gradient driven m/n = 1/1 quasi-interchange-like MHD instability. The work systematically extrapolates the parameter regimes of flux pumping from the above AUG base case by scanning dissipation coefficients and plasma beta. The simulation results reveal bifurcated plasma behaviours at different Hartmann numbers, including distinct states such as flux pumping (helical core with a flat current density), sawteeth (periodic kink-cycling), single crash (without subsequent cycle), and quasi-stationary magnetic island (peaked current density). Transitions from marginal flux pumping state to sawteeth are observed in long-term simulations. The relationships between system dissipation, plasma beta, and different plasma states are carefully analyzed. For practical purposes, the potential operational window for flux pumping, as determined by plasma density and temperature, is estimated. The modelling efforts advance the understanding of flux pumping and facilitate the development of a fast surrogate model for efficient evaluation of flux pumping.

physics.plasm-ph↗

arXiv2026

First observation of multi-phonon $\gamma$-vibrations in an odd-odd nuclear system

The identification of the first multi-phonon $\gamma$-vibrational bands in an odd-odd neutron-rich nucleus of the nuclear chart is presented. These high spin structures of hard to access $^{104}_{41}$Nb$_{63}$, produced in fission, were studied by combining a spectrometer with isotopic resolution coupled to a $\gamma$-ray tracking array and independently high-fold $\gamma$ coincidence measurements. Triaxial Projected Shell Model calculations for the high-spin states are in good agreement with the measured observables for the yrast, one-phonon and two-phonon $\gamma$ bands. The possibility of an oblate shape of an isomeric state and coexistence of triaxial and oblate configurations are investigated from the decay of the 141 keV isomer. The present work illustrates the robustness of vibration excitations in the presence of odd valence proton and neutron as well as the possibly coexisting shapes beyond the $N=60$ transitional region.

nucl-ex↗

arXiv2026

POLAMI Multi-Wavelength Polarization Study of AGN Jets: A Millimeter-Optical Comparison

Millimeter-band polarimetry offers a powerful probe of AGN jets, accessing regions less affected by opacity and Faraday rotation than at longer radio wavelengths. As part of the POLAMI program, we have conducted 14 years of 1 mm and 3 mm polarization monitoring of a sample of gamma-ray-bright blazars with the IRAM 30-m telescope, complemented here with long-term optical polarimetric observations from multiple facilities. We aim to test whether current models of parsec-scale jet physics are consistent with observed multi-band polarization behavior. Using a Bayesian framework, we derive intrinsic mean flux densities and modulation indices for total flux and fractional polarization, and characterize EVPA variability using circular statistics. We then examine how these quantities reflecting variability properties across millimeter and optical bands relate to synchrotron peak frequency, jet orientation, and radio/gamma-ray luminosities. BL Lac objects exhibit, on average, higher fractional polarization and lower EVPA variability than FSRQs at all wavelengths. Fractional polarization increases with frequency, consistent with increasingly ordered magnetic fields at shorter wavelengths. BL Lacs also show more frequent alignment of EVPAs between optical and millimeter bands, whereas FSRQs display weaker coherence. EVPA variability correlates positively with radio and gamma-ray luminosities and negatively with synchrotron peak frequency, most strongly in the optical. We further find a positive correlation between EVPA spread and fractional polarization variability, suggesting a direct link between magnetic-field structure and polarization dynamics.

astro-ph.HE↗

arXiv2025

Transfer Learning for Analysis of Collective and Non-Collective Thomson Scattering Spectra

Thomson scattering (TS) diagnostics provide reliable, minimally perturbative measurements of fundamental plasma parameters, such as electron density ($n_e$) and electron temperature ($T_e$). Deep neural networks can provide accurate estimates of $n_e$ and $T_e$ when conventional fitting algorithms may fail, such as when TS spectra are dominated by noise, or when fast analysis is required for real-time operation. Although deep neural networks typically require large training sets, transfer learning can improve model performance on a target task with limited data by leveraging pre-trained models from related source tasks, where select hidden layers are further trained using target data. We present five architecturally diverse deep neural networks, pre-trained on synthetic TS data and adapted for experimentally measured TS data, to evaluate the efficacy of transfer learning in estimating $n_e$ and $T_e$ in both the collective and non-collective scattering regimes. We evaluate errors in $n_e$ and $T_e$ estimates as a function of training set size for models trained with and without transfer learning, and we observe decreases in model error from transfer learning when the training set contains $\lessapprox$ 200 experimentally measured spectra.

physics.plasm-ph↗

arXiv2025

DeepSeek-V3.2: Pushing the Frontier of Open Large Language Models

We introduce DeepSeek-V3.2, a model that harmonizes high computational efficiency with superior reasoning and agent performance. The key technical breakthroughs of DeepSeek-V3.2 are as follows: (1) DeepSeek Sparse Attention (DSA): We introduce DSA, an efficient attention mechanism that substantially reduces computational complexity while preserving model performance in long-context scenarios. (2) Scalable Reinforcement Learning Framework: By implementing a robust reinforcement learning protocol and scaling post-training compute, DeepSeek-V3.2 performs comparably to GPT-5. Notably, our high-compute variant, DeepSeek-V3.2-Speciale, surpasses GPT-5 and exhibits reasoning proficiency on par with Gemini-3.0-Pro, achieving gold-medal performance in both the 2025 International Mathematical Olympiad (IMO) and the International Olympiad in Informatics (IOI). (3) Large-Scale Agentic Task Synthesis Pipeline: To integrate reasoning into tool-use scenarios, we developed a novel synthesis pipeline that systematically generates training data at scale. This methodology facilitates scalable agentic post-training, yielding substantial improvements in generalization and instruction-following robustness within complex, interactive environments.

cs.CL↗

arXiv2025

A Novel and Practical Universal Adversarial Perturbations against Deep Reinforcement Learning based Intrusion Detection Systems

Intrusion Detection Systems (IDS) play a vital role in defending modern cyber physical systems against increasingly sophisticated cyber threats. Deep Reinforcement Learning-based IDS, have shown promise due to their adaptive and generalization capabilities. However, recent studies reveal their vulnerability to adversarial attacks, including Universal Adversarial Perturbations (UAPs), which can deceive models with a single, input-agnostic perturbation. In this work, we propose a novel UAP attack against Deep Reinforcement Learning (DRL)-based IDS under the domain-specific constraints derived from network data rules and feature relationships. To the best of our knowledge, there is no existing study that has explored UAP generation for the DRL-based IDS. In addition, this is the first work that focuses on developing a UAP against a DRL-based IDS under realistic domain constraints based on not only the basic domain rules but also mathematical relations between the features. Furthermore, we enhance the evasion performance of the proposed UAP, by introducing a customized loss function based on the Pearson Correlation Coefficient, and we denote it as Customized UAP. To the best of our knowledge, this is also the first work using the PCC value in the UAP generation, even in the broader context. Four additional established UAP baselines are implemented for a comprehensive comparison. Experimental results demonstrate that our proposed Customized UAP outperforms two input-dependent attacks including Fast Gradient Sign Method (FGSM), Basic Iterative Method (BIM), and four UAP baselines, highlighting its effectiveness for real-world adversarial scenarios.

cs.CR↗

arXiv2025

Investigating the mysterious nature of 1LHAASO J1740+0948u through deep XMM-Newton observations

1LHAASO J1740+0948u is a very-high-energy (VHE) source reported by LHAASO, with no counterpart at other wavelengths. It is located at 0.2{\deg} from PSR J1740+1000, a radio and gamma-ray pulsar placed well above the Galactic plane, which displays an X-ray tail. Despite the offset, the association between the two sources is likely. We aim to study the diffuse X-ray emission around PSR J1740+1000 and its tail to investigate the origin of 1LHAASO J1740+0948u through a multi-wavelength SED fitting, testing different scenarios. We analysed ~500 ks of XMM-Newton observations and studied for the first time the diffuse emission around the pulsar. We also analysed the tail and how its emission evolves as a function of distance. We then performed a fit of the SED, including the spectrum of 1LHAASO J1740+0948u and the X-ray data obtained from either the analysis of the tail or the diffuse emission, to understand whether one of the two X-ray sources could be related to the TeV emission and attempt a source classification. The diffuse X-ray emission analysis resulted in upper limits in the 0.5-10 keV range. The tail is best fitted with a power law with $\Gamma=1.76\pm0.06$ in 0.5-8 keV, with no significant detection of spectral variations with distance. We do not find a good SED fit that can explain both the X-ray tail and the LHAASO spectrum with reasonable parameters, suggesting that the TeV emission likely comes from an older X-ray-faint electron population. We then performed an SED fitting of the VHE spectrum combined with the upper limits on the diffuse emission, constraining the magnetic field to be as low as $B\leq1.2 \mu$G. We suggest that 1LHAASO J1740+0948u could represent either the relic PWN of PSR J1740+1000 or its pulsar halo. Our energy density results hint at a halo-like nature for 1LHAASO J1740+0948u, but deeper multi-wavelength observations are required to confirm this hypothesis.

astro-ph.HE↗

arXiv2025

Comparison of MHD and gyrokinetic simulations of linear instabilities at the q = 1 surface

Accurate modeling of core instabilities in tokamak plasmas is essential to understand the underlying physical mechanisms and their impact on plasma confinement. The ideal stability of the internal kink mode and the m = 1 collisionless tearing mode are analyzed numerically both with gyrokinetic and MHD codes. We compare the different models implemented in the codes and show that the gyrokinetic equations without collisions inherently contain the ideal MHD limit. The simulation results show that the stability of the internal kink mode strongly depends on the choice of several setup parameters like the inclusion of parallel magnetic field fluctuations, the tokamak aspect ratio, the drift- or gyrokinetic treatment of the ions and the electron mass. Furthermore, we demonstrate the stabilization of the instabilities by diamagnetic effects. Our results indicate that gyrokinetic and MHD models can be reconciled in the description of the internal kink mode by careful consideration of the simulation setup and model assumptions, but instabilities like the collisionless tearing mode require a more advanced treatment beyond MHD.

physics.plasm-ph↗

arXiv2025

Measurement of $^{3,4}$He($K^-, \pi^0$)$^{3,4}_\Lambda$H reaction cross section and evaluation of hypertriton $\Lambda$ binding energy

Light $s$-shell hypernuclei ($^{3,4}_{\Lambda}\text{H}$) and their ground-state properties are crucial benchmarks in hypernuclear physics. In particular, comparing the production cross sections of $^{3}_{\Lambda}\text{H}$ and $^{4}_{\Lambda}\text{H}$ provides insights into the $\Lambda N$ interaction in different isospin configurations, which can help address recent discrepancies in the reported $\Lambda$ binding energy of hypertriton. We present the first measurement of the production cross sections for $^{3}_{\Lambda}\text{H}$ and $^{4}_{\Lambda}\text{H}$ using the in-flight $(K^-, \pi^0)$ reaction at a beam momentum of 1.0 GeV/$c$ with an identical experimental setup. The production cross sections in the laboratory frame, for the angular range from 0$^{\circ}$ to 20$^{\circ}$, are measured to be $15.0~\pm~2.6~(\text{stat.})~^{+2.4}_{-2.8}~(\text{syst.})~\mu \text{b} $ and $49.9~\pm~2.1~(\text{stat.})~^{+7.8}_{-8.0}~(\text{syst.})~\mu \text{b}$ for the ground-state of $^{3}_{\Lambda}\text{H}$ and $^{4}_{\Lambda}\text{H}$, respectively. Using the ratio of these cross sections and comparing it with theoretical calculations, we evaluate the $\Lambda$ binding energy of hypertriton, yielding a value consistent with the picture of a loosely bound system.

nucl-ex↗

arXiv2025

STRRINGS: STReams in Residual Images of Nearby GalaxieS

Tidal features from galaxy mergers, particularly stellar streams, offer valuable insights into galaxy assembly and dark matter halo properties. This paper aims to identify a large sample of nearby stellar streams suitable for detailed modelling and comparison with simulations to enable population-level constraints on halo properties. We visually inspect and compile a tidal feature catalogue for $19,387$ galaxies with redshift $z \leq 0.02$ from the Siena Galaxy Atlas 2020 using original, model, and residual images from the DESI Legacy Imaging Surveys. Residual images, produced by subtracting models of all sources, enhance the detectability of faint asymmetries such as tidal features. We find that $11.9 \pm 0.2\%$ of galaxies host detectable tidal features, more frequently around early-type than late-type galaxies. The tidal feature fraction increases with stellar mass, from $2.4 \pm 0.4\%$ at $\sim10^8$M$_\odot$ to $36.5 \pm 1.2\%$ at $\sim 5\times10^{11}$M$_\odot$. From this, we present the first release of STRRINGS: STReams in Residual Images of Nearby GalaxieS, a subsample of 35 galaxies with long, narrow streams suitable for modelling. We segment these streams and derive their geometry, surface brightness, colours, and stellar masses. The median $g$-band surface brightness is 26.8 mag$\,$arcsec$^{-2}$, reaching 27.5 mag$\,$arcsec$^{-2}$ for the faintest stream. Mass ratios are consistent with minor mergers, and we identify five potential dwarf galaxy progenitors. Our streams are typically longer (median 124 kpc) than the literature, with comparable widths. Stream mass correlates with length and colour, and wider streams lie at larger galactocentric radii. STRRINGS will be expanded and used to constrain halo properties in future work.

astro-ph.GA↗

arXiv2025

Imaging-spectroscopy diagnosis of the giant sloshing spiral in the Virgo cluster with the Einstein Probe Follow-up X-ray Telescope

We performed deep X-ray observations of the Virgo cluster using the Einstein Probe Follow-up X-ray Telescope (EP-FXT) with a total exposure of 295 ks. Leveraging the large field of view (FoV) and low particle background of EP-FXT, the image reveals a giant spiral feature connecting the cold fronts in the northwest and southeast, forming a coherent structure consistent with earlier results from XMM-Newton and Suzaku. We also present two-dimensional maps of temperature, metallicity, and entropy across the Virgo Cluster, covering a FoV of approximately 28.5 arcmin. These maps clearly show a spiral structure with high density, low temperature, high metallicity, and low entropy. The results support a scenario where the spiral morphology arises from gas sloshing driven by a minor merger. Additionally, EP-FXT temperature measurements agree well with XMM-Newton data within uncertainties.

astro-ph.HE↗