arXiv subjects

Yang Xue

Publications and source records attributed to Yang Xue.

At least 19 recordsLinked to original sources

arXiv2026

Sample-Guided Exact Top-K Selection for Long-Context Sparse Attention

Sparse attention bounds downstream attention work by retaining a fixed-size subset of indexed tokens, but its standalone exact Top-$K$ stage must still process materialized score rows whose length grows with context. Production radix selectors discover their first actionable boundary only after a complete-row pass, forcing another row-scale traversal before exact refinement. We observe that locating a compact upper tail requires substantially less resolution than identifying the exact rank boundary, and that fixed-stride partial views of the current row remain calibrated to the corresponding complete-row rank across ragged lengths. We present HPC-Ops Top-K, a sample-guided exact selector for ragged sparse-attention score rows. A fixed-stride view proposes a row-local coarse boundary; the mandatory complete-row pass certifies its sufficiency, forms the admitted candidate set, and initializes exact FP32 refinement over the unresolved frontier. A nested secondary boundary and exact recovery handle underfilled proposals before any output is committed, so sampling controls common-path work but never correctness. The GPU implementation fuses complete-row certification and candidate formation, and combines persistent, KV-split, and direct-exact execution behind graph-capturable ragged-row dispatch. We evaluate HPC-Ops Top-K on indexer scores from Hy4-Preview. It outperforms the fastest verified external exact baseline by $1.29$--$1.75\times$ across 20 operator configurations, with a $1.55\times$ geometric-mean speedup. It further achieves $1.36\times$ and $1.48\times$ speedups on two framework-derived sparse-attention traces. The implementation is available in HPC-Ops, Tencent's open-source high-performance operator library for LLM inference, at https://github.com/Tencent/hpc-ops.

cs.DC↗

arXiv2026

Photon Anomalous Blockade in Waveguide Cavity QED with Atomic Mirrors

Waveguide cavity quantum electrodynamics (QED) with atomic mirrors is a growing research area of quantum optics and can be applied to quantum information processing. We here study the photon statistics of output fields from a waveguide cavity QED system, in which the waveguide is coupled to quantized mirror atoms and one driven medium atom. Our results show that the photon blockade can occur even for a bad atom cavity with large dissipation and small coupling between the medium atom and the cavity, in contrast to the small dissipation and the strong coupling of the medium atom to the cavity field for the conventional photon blockade or the quantum interference for the unconventional photon blockade in the cavity QED system. Utilizing both the master equation and scattering theories, we derive the condition under which the photon blockade occurs in weakly driven systems. We find that such photon anomalous blockade is due to the quantum Zeno effect and is robust against variations of the medium atom's position within the cavity. Our study paves a way to exploit the photon blockade and single-photon devices via the waveguide cavity QED.

quant-ph↗

arXiv2026

3D imaging of the biphoton spatiotemporal wave packet

Photons are among the most important carriers of quantum information owing to their rich degrees of freedom (DoFs), including various spatiotemporal structures. The ability to characterize these DoFs, as well as the hidden correlations among them, directly determines whether they can be exploited for quantum tasks. While various methods have been developed for measuring the spatiotemporal structure of classical light fields, owing to the technical challenges posed by weak photon flux, there have so far been no reports of observing such structures in their quantum counterparts, except for a few studies limited to correlations within individual DoFs. Here, we propose and experimentally demonstrate a self-referenced, high-efficiency, and all-optical method, termed 3D imaging of photonic wave packets, for comprehensive characterization of the spatiotemporal structure of a quantum light field, i.e., the biphoton spatiotemporal wave packet. Benefiting from this developed method, we successfully observe the spatial-spatial, spectral-spectral, and spatiotemporal correlations of biphotons generated via spontaneous parametric down-conversion, revealing rich local and nonlocal spatiotemporal structure in quantum light fields. This method will further advance the understanding of the dynamics in nonlinear quantum optics and expand the potential of photons for applications in quantum communication and quantum computing.

quant-ph↗

arXiv2025

Engineering photonic dispersion relation and atomic dynamics in waveguide QED setup via long-range hoppings

Non-trivial dispersion relations engineered in photonic waveguide for the precise control of atomic dynamics has recently attracted considerable attention. Here, we study a system in which atoms are coupled to one-dimensional coupled-resonator waveguides with long-range hoppings. By carefully engineering the jth-order nearest neighbor (JNN) hoppings between resonators, we construct linear dispersion relations with the chiral characteristic. To quantify the degree of linearity, we analyze the propagation fidelities of Gaussian wave packets in these waveguides. Furthermore, we demonstrate that such coupled-resonator waveguides can serve as versatile platforms for enabling directional atomic radiation and absorption. Beyond linear dispersion relations, more general forms, including quadratic and cubic relations, can also be achieved through tailored JNN-hoppings. Our study thus provides a unified framework for simulating atom-environment couplings with arbitrary dispersion relations.

quant-ph↗

arXiv2025

MosaicDoc: A Large-Scale Bilingual Benchmark for Visually Rich Document Understanding

Despite the rapid progress of Vision-Language Models (VLMs), their capabilities are inadequately assessed by existing benchmarks, which are predominantly English-centric, feature simplistic layouts, and support limited tasks. Consequently, they fail to evaluate model performance for Visually Rich Document Understanding (VRDU), a critical challenge involving complex layouts and dense text. To address this, we introduce DocWeaver, a novel multi-agent pipeline that leverages Large Language Models to automatically generate a new benchmark. The result is MosaicDoc, a large-scale, bilingual (Chinese and English) resource designed to push the boundaries of VRDU. Sourced from newspapers and magazines, MosaicDoc features diverse and complex layouts (including multi-column and non-Manhattan), rich stylistic variety from 196 publishers, and comprehensive multi-task annotations (OCR, VQA, reading order, and localization). With 72K images and over 600K QA pairs, MosaicDoc serves as a definitive benchmark for the field. Our extensive evaluation of state-of-the-art models on this benchmark reveals their current limitations in handling real-world document complexity and charts a clear path for future research.

cs.CV↗

arXiv2025

Reshaping Biomolecular Structure Prediction through Strategic Conformational Exploration with HelixFold-S1

Generating large ensembles of candidate conformations is standard for improving biomolecular structure prediction. Yet aimless sampling is inefficient and costly, producing many redundant conformations with limited diversity, particularly for complex multimeric assemblies. Here, we present HelixFold-S1, a guided planning approach specifically designed to enhance the structural prediction of biomolecular complexes by strategically targeting the most informative regions of conformational space to produce accurate conformations. For each complex, predicted inter-chain contact probabilities serve as a blueprint of the conformational space, guiding computational effort toward higher-probability, low-redundancy contacts that constrain structure generation. Across diverse biomolecular complex benchmarks, HelixFold-S1 achieves markedly higher structural accuracy than traditional unguided methods while reducing sampling requirements by an order of magnitude. Predicted contact probabilities also provide a rough indicator of prediction difficulty and sampling utility. These results demonstrate that guided planning reshapes conformational exploration and enables more efficient and accurate structural inference.

q-bio.BM↗

arXiv2025

Observation and research on cosmic ray muons and solar modulation effect based on plastic scintillator detector

Cosmic rays, originating from stars, supernovae, and other astrophysical sources, are composed of high-energy particles that enter Earths atmosphere. Upon interaction with atmospheric nuclei, these primary cosmic rays generate secondary particles, including neutrons, electrons, and muons, with muons constituting a dominant component at ground level. Muons, due to their relative abundance, stability, and well-characterized energy loss mechanisms, serve as critical probes for investigating the fundamental properties of cosmic rays. Studies of muon energy distribution, diurnal anisotropy, and their modulation by solar activity provide critical insights into the mechanism of particle acceleration in cosmic ray sources and the effects of solar and atmospheric.This study aims to characterize the counting spectra and anisotropic properties of cosmic ray muons by using a plastic scintillator detector system. The experiment was conducted over a three-month period, from December 2023 to February 2024, leveraging long-bar plastic scintillator detectors equipped with dual-end photomultiplier tubes (PMTs) and a high-resolution digital data acquisition system. A dual-end coincidence measurement technique was used to enhance the signal-to-noise ratio by suppressing thermal noise and other background interferences. Diurnal variations in muon count rates exhibit a pronounced pattern, with a systematic reduction occurring between 8:00 AM and 1:00 PM. This phenomenon is attributed to the solar shielding effects, where enhanced solar activity during daytime hours modulates the flux of galactic cosmic rays reaching Earths surface. The study further corroborates these findings through cross-comparisons with data from the Yangbajing Cosmic Ray Observatory. These observations underscore the robustness of the plastic scintillator detector system for capturing detailed muon spectra and anisotropic patterns.

astro-ph.HE↗

arXiv2025

Anomalous Chern-Simons orbital magnetoelectric coupling of three-dimensional Chern insulators: gauge-discontinuity formalism and adiabatic pumping

Chern-Simons orbital magnetoelectric (OME) coupling is usually the hallmark of nontrivial band topology in three-dimensional (3D) crystalline insulators. However, if a 3D insulator exhibits nonzero Chern number within any two-dimensional plane of the Brillouin zone, then traditionally the Chern-Simons coupling becomes ill defined for such 3D Chern insulators due to topological obstructions. In this work, by employing a ``gauge-discontinuity" formalism, we resolve this long-standing issue and rigorously derive a quantized layer-resolved OME response in 3D Chern insulators. We demonstrate that the difference of the layer-resolved OME coupling between adjacent layers is universally quantized in unit of $-C e^2/h$, where $C$ is the Chern number. This quantization arises from an anomalous contribution to the Chern-Simons OME coupling, which is closely associated with the Berry curvature of the occupied bands and the hybrid Wannier centers along the direction of the Chern vector $(0,0, C)$. Furthermore, we demonstrate that the anomalous Chern-Simons coupling can be transported by an exact integer quantum from one unit cell to its neighboring cell through an adiabatic cyclic pumping process, accompanied by a quantized displacement of Wannier center along the direction of the Chern vector. Our work provides a rigorous theoretical framework for understanding magnetoelectric response in 3D Chern insulators and opens avenues for designing topological quantum phenomena in layered systems.

cond-mat.mes-hall↗

arXiv2025

Two-photon scattering in a waveguide by a giant atom

We study two-photon scattering by a two-level giant atom in a waveguide. We first study the case that the giant atom is coupled to the waveguide via two coupling points, and obtain Bethe ansatz eigenstates and eigenvalues in the Hilbert space of two-excitation. Then we derive bound states by subtracting the states corresponding to Bethe ansatz solutions from the entire two-excitation Hilbert space, and construct the two-photon scattering matrix (S-matrix) by using Bethe ansatz eigenstates and bound states. We further study the properties of output states, which include both the scattering and bound states, for arbitrarily incident two-photon states by using a concrete example. We find that the oscillation period of the scattering states and decay rates of the bound states strongly depend on the distance between two coupling points. Moreover, we find that the two-photon correlation in the bound states can be enhanced by changing such distance when the total energy of two incident photons equals to two times of single photon resonance energy. We also generalize our study to the case that the giant atom is coupled to the waveguide via $N$ coupling points. We obtain all the eigenstates and eigenvalues of the scattering matrix and construct the S-matrix. Comparing with the case of the two coupling points, we find the photon correlation can be further enhanced by increasing the number of the coupling points for the same incident states when the distance of any two nearest neighbor coupling points is half of the wavelength.

quant-ph↗

arXiv2024

Precise Antigen-Antibody Structure Predictions Enhance Antibody Development with HelixFold-Multimer

The accurate prediction of antigen-antibody structures is essential for advancing immunology and therapeutic development, as it helps elucidate molecular interactions that underlie immune responses. Despite recent progress with deep learning models like AlphaFold and RoseTTAFold, accurately modeling antigen-antibody complexes remains a challenge due to their unique evolutionary characteristics. HelixFold-Multimer, a specialized model developed for this purpose, builds on the framework of AlphaFold-Multimer and demonstrates improved precision for antigen-antibody structures. HelixFold-Multimer not only surpasses other models in accuracy but also provides essential insights into antibody development, enabling more precise identification of binding sites, improved interaction prediction, and enhanced design of therapeutic antibodies. These advances underscore HelixFold-Multimer's potential in supporting antibody research and therapeutic innovation.

q-bio.BM↗

arXiv2024

Technical Report of HelixFold3 for Biomolecular Structure Prediction

The AlphaFold series has transformed protein structure prediction with remarkable accuracy, often matching experimental methods. AlphaFold2, AlphaFold-Multimer, and the latest AlphaFold3 represent significant strides in predicting single protein chains, protein complexes, and biomolecular structures. While AlphaFold2 and AlphaFold-Multimer are open-sourced, facilitating rapid and reliable predictions, AlphaFold3 remains partially accessible through a limited online server and has not been open-sourced, restricting further development. To address these challenges, the PaddleHelix team is developing HelixFold3, aiming to replicate AlphaFold3's capabilities. Leveraging insights from previous models and extensive datasets, HelixFold3 achieves accuracy comparable to AlphaFold3 in predicting the structures of the conventional ligands, nucleic acids, and proteins. The initial release of HelixFold3 is available as open source on GitHub for academic research, promising to advance biomolecular research and accelerate discoveries. The latest version will be continuously updated on the HelixFold3 web server, providing both interactive visualization and API access.

q-bio.BM↗

arXiv2024

Simultaneous Verification of Genuine Multipartite Nonlocality and Full Network Nonlocality

Genuine multipartite nonlocality and nonlocality arising in networks composed of several independent sources have been separately investigated. While some genuinely entangled states cannot be verified by violating a single Bell-type inequality, a quantum network consisting of different sources allows for the certification of the non-classicality of all sources. In this paper, we propose the first method to verify both types of nonlocality simultaneously in a single experiment. We consider a quantum network comprising a bipartite source and a tripartite source. We demonstrate that there are quantum correlations cannot be simulated if the tripartite source distributes biseparable systems while the bipartite source distributes even stronger-than-quantum systems. These correlations can be used to verify both the genuine multipartite nonlocality of generalized Greenberger-Horne-Zeilinger states and the full network nonlocality that is stronger than all the existing results. Experimentally, we observe both types of nonlocality in a high fidelity photonic quantum network by violating a single network Bell inequality.

quant-ph↗

arXiv2024

Unifying Sequences, Structures, and Descriptions for Any-to-Any Protein Generation with the Large Multimodal Model HelixProtX

Proteins are fundamental components of biological systems and can be represented through various modalities, including sequences, structures, and textual descriptions. Despite the advances in deep learning and scientific large language models (LLMs) for protein research, current methodologies predominantly focus on limited specialized tasks -- often predicting one protein modality from another. These approaches restrict the understanding and generation of multimodal protein data. In contrast, large multimodal models have demonstrated potential capabilities in generating any-to-any content like text, images, and videos, thus enriching user interactions across various domains. Integrating these multimodal model technologies into protein research offers significant promise by potentially transforming how proteins are studied. To this end, we introduce HelixProtX, a system built upon the large multimodal model, aiming to offer a comprehensive solution to protein research by supporting any-to-any protein modality generation. Unlike existing methods, it allows for the transformation of any input protein modality into any desired protein modality. The experimental results affirm the advanced capabilities of HelixProtX, not only in generating functional descriptions from amino acid sequences but also in executing critical tasks such as designing protein sequences and structures from textual descriptions. Preliminary findings indicate that HelixProtX consistently achieves superior accuracy across a range of protein-related tasks, outperforming existing state-of-the-art models. By integrating multimodal large models into protein research, HelixProtX opens new avenues for understanding protein biology, thereby promising to accelerate scientific discovery.

cs.LG↗

arXiv2024

HelixFold-Multimer: Elevating Protein Complex Structure Prediction to New Heights

While monomer protein structure prediction tools boast impressive accuracy, the prediction of protein complex structures remains a daunting challenge in the field. This challenge is particularly pronounced in scenarios involving complexes with protein chains from different species, such as antigen-antibody interactions, where accuracy often falls short. Limited by the accuracy of complex prediction, tasks based on precise protein-protein interaction analysis also face obstacles. In this report, we highlight the ongoing advancements of our protein complex structure prediction model, HelixFold-Multimer, underscoring its enhanced performance. HelixFold-Multimer provides precise predictions for diverse protein complex structures, especially in therapeutic protein interactions. Notably, HelixFold-Multimer achieves remarkable success in antigen-antibody and peptide-protein structure prediction, greatly surpassing AlphaFold 3. HelixFold-Multimer is now available for public use on the PaddleHelix platform, offering both a general version and an antigen-antibody version. Researchers can conveniently access and utilize this service for their development needs.

q-bio.BM↗

arXiv2024

Two-dimensional topological effect in a transmon qubit array with tunable couplings

We investigate a square-lattice architecture of superconducting transmon qubits with inter-qubit interactions mediated by inductive couplers. Therein, the inductive couling between the qubit and couplers is suggested to be designed into the gradiometer form to intigimate the flux noise orginating from the environment. Via periodically modulating the couplers,the Abelian gauge potential, termed effective magnetic flux, can be synthesized artificially, making the system an excellent platform for simulating two-dimensional topological physics. In the simplest two-dimensional model, the double (or three-leg) ladder, the staggered vortex-Meissner phase transition different from that in the two-leg ladder can be found in the single-particle ground state as the effective magnetic flux varies. Besides, the large coupling ratio between the interleg and intraleg coupling strengths also makes the chiral current resemble squeezed sinusoidal functions. If the row number is further increased, the topological band structure anticipated at massive rows begins to occur even for a relatively small number of rows (ten or so for the considered parameters). This heralds a small circuit scale to observe the topological band. The edge state in the band gap is determined by the topological Chern number and can be calculated through integrating the Berry curvature with respect to the first Brillouin zone. Besides, we present a systematic method on how to measure the topological band structure based on time- and space-domain Frourier transformation of the wave function after properly excited. The result offers an avenue for simulating two-dimensional topological physics on the state-of-the-art superconducting quantum chips.

quant-ph↗

arXiv2023

Layer-Locked Anomalous Valley Hall Effect in Two-Dimensional A-Type Tetragonal Antiferromagnetic Insulator

Antiferromagnetic (AFM) spintronics provides a route towards energy-efficient and ultrafast device applications. Achieving anomalous valley Hall effect (AVHE) in AFM monolayers is thus of considerable interest for both fundamental condensed matter physics and device enginering. Here we propose a route to achieve AVHE in A-type AFM insulator composed of vertically-stacked monolayer quantum anomalous Hall insulator (QAHI) with strain and electric field modulations. Uniaxial strain and electric field generate valley polarization and spin splitting, respectively. Using first-principles calculations, $\mathrm{Fe_2BrMgP}$ monolayer is predicted to be a prototype A-type AFM hosting \emph{valley-polarized quantum spin Hall insulator} (VQSHI) in which AVHE and quantum spin Hall effect (QSHE) are synergized in a single system. Our findings reveal a route to achieve multiple Hall effects in 2D tetragonal AFM monolayers.

cond-mat.mes-hall↗

arXiv2023

GridFormer: Towards Accurate Table Structure Recognition via Grid Prediction

All tables can be represented as grids. Based on this observation, we propose GridFormer, a novel approach for interpreting unconstrained table structures by predicting the vertex and edge of a grid. First, we propose a flexible table representation in the form of an MXN grid. In this representation, the vertexes and edges of the grid store the localization and adjacency information of the table. Then, we introduce a DETR-style table structure recognizer to efficiently predict this multi-objective information of the grid in a single shot. Specifically, given a set of learned row and column queries, the recognizer directly outputs the vertexes and edges information of the corresponding rows and columns. Extensive experiments on five challenging benchmarks which include wired, wireless, multi-merge-cell, oriented, and distorted tables demonstrate the competitive performance of our model over other methods.

cs.CV↗

arXiv2023

Online Map Vectorization for Autonomous Driving: A Rasterization Perspective

Vectorized high-definition (HD) map is essential for autonomous driving, providing detailed and precise environmental information for advanced perception and planning. However, current map vectorization methods often exhibit deviations, and the existing evaluation metric for map vectorization lacks sufficient sensitivity to detect these deviations. To address these limitations, we propose integrating the philosophy of rasterization into map vectorization. Specifically, we introduce a new rasterization-based evaluation metric, which has superior sensitivity and is better suited to real-world autonomous driving scenarios. Furthermore, we propose MapVR (Map Vectorization via Rasterization), a novel framework that applies differentiable rasterization to vectorized outputs and then performs precise and geometry-aware supervision on rasterized HD maps. Notably, MapVR designs tailored rasterization strategies for various geometric shapes, enabling effective adaptation to a wide range of map elements. Experiments show that incorporating rasterization into map vectorization greatly enhances performance with no extra computational cost during inference, leading to more accurate map perception and ultimately promoting safer autonomous driving.

cs.CV↗