Searcharxiv⌕ Search

arXiv subjects

Yongsheng Zhang

Publications and source records attributed to Yongsheng Zhang.

At least 19 recordsLinked to original sources
arXiv2026

Multi-Module Minimal Products: Exact Composition, Closed Profiles, and Gaussian Transfer

We establish an exact composition principle for spherical immersions coupled through coefficient profiles and norm-preserving bilinear maps. Under an explicit mixed-orthogonality condition, the spherical mean-curvature vector splits into orthogonal input and profile contributions. The composition is minimal if and only if every input is minimal and the profile is $W$-minimal for an explicit monomial weight. Using Hsiang--Lawson reduction and Kapouleas--McGrath gluing, we construct closed embedded $W$-minimal profiles, including families with unbounded intermediate Betti numbers. For sufficiently large comparable input dimensions, a Gaussian transfer developed in this paper produces interior profiles from closed embedded Gaussian seeds that are nondegenerate modulo rotations.

math.DG↗
arXiv2026

The Geometry and Dynamics of Spiral Minimal Products

We study the spiral product $G_γ(t,x,y)=(z_1(t)f_1(x),z_2(t)f_2(y))$, which couples two spherical $\mathscr C$-totally real immersed factors through a profile curve $γ=(z_1,z_2)\subset\mathbb S^3$. Its minimality is governed by a weighted-geodesic system: $G_γ$ is minimal precisely when both factors are minimal and $γ$ is an unparametrized geodesic of $|z_1|^{2k_1}|z_2|^{2k_2}g_{\mathbb S^3}$. This flow is Liouville integrable and, when $k_1+k_2>0$, its phase map has an open dense full-rank locus. Consequently, closed profiles of arbitrarily large primitive order occur densely. For compact minimal factors, Routh reduction and Sturm oscillation give $\text{Ind}(G_γ)\geq \text{Ind}(f_1)+\text{Ind}(f_2)+2m_γ-3$ for a closed oscillatory profile of primitive closing order $m_γ$. On the contact level, a factor-adapted choice of profiles produces, from any prescribed pair of compact connected embedded special Legendrians, embedded special Legendrian products of every sufficiently large prime closing order. Finally, canonical finite horizontal lifts and Hopf projection give Delaunay-type minimal Lagrangians in complex projective spaces. For compact embedded inputs and ordinarily closed profiles, the primitive spherical quotient is embedded, while its projective quotient is embedded exactly when the reduced relative winding is one.

math.DG↗
arXiv2026

IB-HFN: Information Bottleneck-Driven SAR-Optical Fusion Network for High-Fidelity Cloud Removal

Synthetic aperture radar (SAR)-assisted optical cloud removal aims to recover surface information obscured by clouds in optical remote sensing images by exploiting complementary SAR observations. Existing multimodal fusion methods typically rely on direct spatial concatenation and pixel-wise supervision, which can propagate SAR speckle noise into optical reconstruction and lead to over-smoothed results. To address these limitations, we propose an Information Bottleneck-driven High-Fidelity Network (IB-HFN) for SAR-assisted optical cloud removal. IB-HFN employs a dual-stream backbone to preserve modality-specific representations before deep semantic fusion, thereby mitigating premature cross-modal contamination. At the fusion stage, we introduce a Spatial Information Bottleneck Fusion module that compresses SAR features through a channel-wise variational information bottleneck to suppress unstructured speckle noise. In parallel, a local-global gating mechanism predicts clear-sky regions and routes reliable optical details through a Dirac-initialized skip connection, decoupling noise suppression from texture preservation. We further develop a joint optimization strategy that integrates feature-level bottleneck regularization with image-level constraints on reconstruction accuracy, structural consistency, spectral fidelity, and contrastive sharpness. A dynamic weighting schedule balances these objectives to stabilize training and reduce hazy artifacts. Experiments on the SEN12MS-CR dataset under challenging spatio-temporal splits demonstrate that IB-HFN achieves superior structural preservation and spectral fidelity over existing methods.

cs.CV↗
arXiv2026

Superconductivity in the pressure-amorphized topological insulator CrP$_4$

The interplay among superconductivity, magnetism, and nontrivial band topology represents one of the most compelling frontiers in condensed matter physics. The exploration of novel superconductivity in 3d transition-metal compounds, particularly the rare Cr-based systems containing strongly magnetic Cr ions, has long attracted attention owing to their unconventional pairing mechanisms that challenge conventional wisdom. Yet, Cr-based superconductors remain scarce, especially those possessing nontrivial topological character, underscoring the urgent need to uncover new members. Here we report the observation of superconductivity in pressure-amorphized Cr-based topological insulator CrP$_4$. Upon compression, CrP$_4$ undergoes an anomalous quantum phase transition from a metallic to a semiconducting-like state at around 15 GPa, driven by significant changes in the electronic structure. At approximately 70 GPa, re-metallization with superconductivity occurs alongside an irreversible amorphization. The superconducting transition temperature Tc increases monotonically with pressure, reaching 4.8 K at 141.3 GPa. Furthermore, theoretical calculations predict multiple topological phase transitions from a strong topological insulator to a trivial state and finally back to a strong topological state under pressure. Our study not only establishes CrP$_4$ as the first Cr-based amorphous superconductor but also opens a new paradigm for exploring superconducting and topological properties in amorphous materials.

cond-mat.supr-con↗
arXiv2026

Cones over minimal products cannot be calibrated by smooth calibrations

We extend a key result in [Zha26], by establishing the obstruction that the minimal product structure (for minimal submanifolds or stationary currents in spheres) automatically makes all cones over (non-trivial) minimal products fail to be calibrated by any global defined smooth calibration in Euclidean spaces.

math.DG↗
arXiv2026

Detect duality obstruction of calibrations in smooth category

This paper consists of three parts: (a) exhibit a new gluing result which can dramatically simplify extensions of calibration pairs; (b) observe that every Lawlor cone can support coflat calibrations singular only at the origin; (c) show that there exist many Lawlor cones which cannot support any smooth calibrations. As an application, we extend our previous work on detecting duality obstruction of calibrations in the smooth category.

math.DG↗
arXiv2026

Some configuration results for area-minimizing cones

We discover some very general configuration results for constructing area-minimizing cones. In particular, given any closed minimal submanifold in some Euclidean sphere, every cone over the minimal product of sufficiently many copies of the submanifold turns out to be area-minimizing; meanwhile every cone over the minimal product of the submanifold and a round sphere of sufficiently large dimension is also area-minimizing. Here no additional geometric assumption (e.g. on isometry group or second fundamental form) is required. Moreover, we establish that the category of regular area-minimizing cones in Euclidean spaces and that of closed minimal submanifolds in Euclidean spheres share the same cardinality.

math.DG↗
arXiv2026

RelayGR: Scaling Long-Sequence Generative Recommendation via Cross-Stage Relay-Race Inference

Real-time recommender systems execute multi-stage cascades (retrieval, pre-processing, fine-grained ranking) under strict tail-latency SLOs, leaving only tens of milliseconds for ranking. Generative recommendation (GR) models can improve quality by consuming long user-behavior sequences, but in production their online sequence length is tightly capped by the ranking-stage P99 budget. We observe that the majority of GR tokens encode user behaviors that are independent of the item candidates, suggesting an opportunity to pre-infer a user-behavior prefix once and reuse it during ranking rather than recomputing it on the critical path. Realizing this idea at industrial scale is non-trivial: the prefix cache must survive across multiple pipeline stages before the final ranking instance is determined, the user population implies cache footprints far beyond a single device, and indiscriminate pre-inference would overload shared resources under high QPS. We present RelayGR, a production system that enables in-HBM relay-race inference for GR. RelayGR selectively pre-infers long-term user prefixes, keeps their KV caches resident in HBM over the request lifecycle, and ensures the subsequent ranking can consume them without remote fetches. RelayGR combines three techniques: 1) a sequence-aware trigger that admits only at-risk requests under a bounded cache footprint and pre-inference load, 2) an affinity-aware router that co-locates cache production and consumption by routing both the auxiliary pre-infer signal and the ranking request to the same instance, and 3) a memory-aware expander that uses server-local DRAM to capture short-term cross-request reuse while avoiding redundant reloads. We implement RelayGR on Huawei Ascend NPUs and evaluate it with real queries. Under a fixed P99 SLO, RelayGR supports up to 1.5$\times$ longer sequences and improves SLO-compliant throughput by up to 3.6$\times$.

cs.DC↗
arXiv2025

P/D-Device: Disaggregated Large Language Model between Cloud and Devices

Serving disaggregated large language models has been widely adopted in industrial practice for enhanced performance. However, too many tokens generated in decoding phase, i.e., occupying the resources for a long time, essentially hamper the cloud from achieving a higher throughput. Meanwhile, due to limited on-device resources, the time to first token (TTFT), i.e., the latency of prefill phase, increases dramatically with the growth on prompt length. In order to concur with such a bottleneck on resources, i.e., long occupation in cloud and limited on-device computing capacity, we propose to separate large language model between cloud and devices. That is, the cloud helps a portion of the content for each device, only in its prefill phase. Specifically, after receiving the first token from the cloud, decoupling with its own prefill, the device responds to the user immediately for a lower TTFT. Then, the following tokens from cloud are presented via a speed controller for smoothed TPOT (the time per output token), until the device catches up with the progress. On-device prefill is then amortized using received tokens while the resource usage in cloud is controlled. Moreover, during cloud prefill, the prompt can be refined, using those intermediate data already generated, to further speed up on-device inference. We implement such a scheme P/D-Device, and confirm its superiority over other alternatives. We further propose an algorithm to decide the best settings. Real-trace experiments show that TTFT decreases at least 60%, maximum TPOT is about tens of milliseconds, and cloud throughput increases by up to 15x.

cs.DC↗
arXiv2025

Ultralow Lattice Thermal Conductivity Induced by Quasi-Chain Configuration in Rb2Se3

Alkali metal-based compounds have garnered significant attention due to their exceptionally low lattice thermal conductivity, which is crucial for applications in thermoelectric energy conversion and thermal barrier coatings. However, the fundamental mechanisms underlying such ultralow lattice thermal conductivity remain poorly understood. In this study, we investigate the intrinsic origins of the ultralow lattice thermal conductivity in the alkali metal-based ionic compound Rb2Se3, which exhibits a simple orthorhombic structure. By employing first-principles density functional theory (DFT) and solving the phonon Boltzmann transport equation (BTE), we reveal that Rb2Se3 achieves lattice thermal conductivity values below 0.2 W/mK along all crystallographic directions at 300 K. Our analysis uncovers a unique quasi-chain configuration within the crystal structure, characterized by strongly covalent Se-Se-Se trimers that act as localized rigid units, while Rb atoms occupy weakly bonded interstitial sites. This configuration induces pronounced anisotropy, weak bonding, and strong anharmonicity, leading to significant rattling-like behavior of all atoms and a dominance of low-frequency phonon modes. The interplay between the rigid Se trimers and the soft Rb matrix results in extreme phonon anharmonicity, as evidenced by large Gruneisen parameters and high atomic displacement parameters (ADPs). These findings provide a comprehensive understanding of the low lattice thermal conductivity in Rb2Se3 and establish a universal framework for designing low lattice thermal conductivity materials through the combination of rigid covalent clusters and soft ionic sublattices.

cond-mat.mtrl-sci↗
arXiv2025

SuperMapNet for Long-Range and High-Accuracy Vectorized HD Map Construction

Vectorized HD map is essential for autonomous driving. Remarkable work has been achieved in recent years, but there are still major issues: (1) in the generation of the BEV features, single modality-based methods are of limited perception capability, while direct concatenation-based multi-modal methods fail to capture synergies and disparities between different modalities, resulting in limited ranges with feature holes; (2) in the classification and localization of map elements, only point information is used without the consideration of element infor-mation and neglects the interaction between point information and element information, leading to erroneous shapes and element entanglement with low accuracy. To address above issues, we introduce SuperMapNet for long-range and high-accuracy vectorized HD map construction. It uses both camera images and LiDAR point clouds as input, and first tightly couple semantic information from camera images and geometric information from LiDAR point clouds by a cross-attention based synergy enhancement module and a flow-based disparity alignment module for long-range BEV feature generation. And then, local features from point queries and global features from element queries are tightly coupled by three-level interactions for high-accuracy classification and localization, where Point2Point interaction learns local geometric information between points of the same element and of each point, Element2Element interaction learns relation constraints between different elements and semantic information of each elements, and Point2Element interaction learns complement element information for its constituent points. Experiments on the nuScenes and Argoverse2 datasets demonstrate superior performances, surpassing SOTAs over 14.9/8.8 mAP and 18.5/3.1 mAP under hard/easy settings, respectively. The code is made publicly available1.

cs.CV↗
arXiv2024

The robustness of skyrmion numbers of structured optical fields in atmospheric turbulence

The development of vector optical fields has brought forth numerous applications. Among these optical fields, a particular class of vector vortex beams has emerged, leading to the emergence of intriguing optical skyrmion fields characterized by skyrmion numbers. The optical skyrmion fields are well-defined by their effective magnetization and possess topologically protected configurations. It is anticipated that this type of optical structure can be exploited for encoding information in optical communication, even under perturbations such as turbulent air, optical fibers, and even general random media. In this study, we numerically demonstrate that the skyrmion numbers of optical skyrmion fields exhibit a certain degree of robustness to atmospheric turbulence, even though their intensity, phase and polarization patterns are distorted. Intriguingly, it is also observed that a larger difference between the absolute values of two azimuthal indices of the vectorial structured light field can lead to a superior level of resilience. These properties not only enhance the versatility of skyrmion fields and their numbers, but also open up new possibilities for their use in various applications across noisy channels.

physics.optics↗
arXiv2024

Remarks on "Spiral Minimal Products"

This note aims to give a better understanding and some remarks about recent preprint ``Spiral Minimal Products". In particular, 1. it should be pointed out that a generalized Delaunay construction among minimal Lagrangians of complex projective spaces has been set up. This is a general structural result working for immersion and current situations. 2. uncountably many new regular (or irregular) special Lagrangian cones with finite density and ``regular" (or irregular) special Lagrangian cones with infinite density in complex Euclidean spaces can be found.

math.DG↗
arXiv2024

P/D-Serve: Serving Disaggregated Large Language Model at Scale

Serving disaggregated large language models (LLMs) over tens of thousands of xPU devices (GPUs or NPUs) with reliable performance faces multiple challenges. 1) Ignoring the diversity (various prefixes and tidal requests), treating all the prompts in a mixed pool is inadequate. To facilitate the similarity per scenario and minimize the inner mismatch on P/D (prefill and decoding) processing, fine-grained organization is required, dynamically adjusting P/D ratios for better performance. 2) Due to inaccurate estimation on workload (queue status or maintained connections), the global scheduler easily incurs unnecessary timeouts in prefill. 3) Block-fixed device-to-device (D2D) KVCache transfer over cluster-level RDMA (remote direct memory access) fails to achieve desired D2D utilization as expected. To overcome previous problems, this paper proposes an end-to-end system P/D-Serve, complying with the paradigm of MLOps (machine learning operations), which models end-to-end (E2E) P/D performance and enables: 1) fine-grained P/D organization, mapping the service with RoCE (RDMA over converged ethernet) as needed, to facilitate similar processing and dynamic adjustments on P/D ratios; 2) on-demand forwarding upon rejections for idle prefill, decoupling the scheduler from regular inaccurate reports and local queues, to avoid timeouts in prefill; and 3) efficient KVCache transfer via optimized D2D access. P/D-Serve is implemented upon Ascend and MindSpore, has been deployed over tens of thousands of NPUs for more than eight months in commercial use, and further achieves 60\%, 42\% and 46\% improvements on E2E throughput, time-to-first-token (TTFT) SLO (service level objective) and D2D transfer time. As the E2E system with optimizations, P/D-Serve achieves 6.7x increase on throughput, compared with aggregated LLMs.

cs.DC↗
arXiv2024

Delaunay hypersurfaces in spheres

We study Delaunay hypersurfaces in $\mathbb S^n$ with $n\geq 3$ and add a missing (flower) type of the category. Moreover, embedded Delaunay hypersurfaces of nonzero constant mean curvatures in $\mathbb S^n$ are found.

math.DG↗
arXiv2023

Symmetry of hypersurfaces with symmetric boundary

Let $G$ be a compact connected subgroup of $SO(n+1)$. In $\mathbb{R}^{n+1}$, we gain interior $G$-symmetry for minimal hypersurfaces and hypersurfaces of constant mean curvature (CMC) which have $G$-invariant boundaries and $G$-invariant contact angles along boundaries. The main ingredients of the proof are to build an associated Cauchy problem based on infinitesimal Lie group actions, and to apply Morrey's regularity theory and the Cauchy-Kovalevskaya Theorem. Moreover, we also investigate the same kind of symmetry inheritance from boundaries for hypersurfaces of constant higher order mean curvature and Helfrich-type hypersurfaces in $\mathbb{R}^{n+1}$.

math.DG↗