arXiv subjects

Xiaoxue Zhang

Publications and source records attributed to Xiaoxue Zhang.

At least 19 recordsLinked to original sources

arXiv2026

Semord: Learned Semantic-Preserving Placement and Low-Fanout Routing for Distributed Vector Search

Vector databases are increasingly deployed in distributed settings where different users, sites, or domains maintain vector data. Existing vector databases rely on a coordinator to record which shards store which parts of the vector space and to route each query to those shards. In a decentralized setting, peers may join, leave, or move data without a trusted node tracking every change, and outdated routing information can therefore send queries to the wrong peers or require contacting many peers, reducing vector retrieval recall and increasing network latency. We present Semord, a decentralized vector search overlay system that achieves high recall by routing each ANN query to a small set of relevant peers, without relying on a centralized coordinator. Semord addresses this problem by making semantic locality routable: 1) We propose VHash to place semantically related vectors near each other in the overlay key space while avoiding load imbalance, so that each query only needs to contact a small neighborhood of peers for distributed local ANN ranking. 2) We design VecDHT, a communication protocol that maintains decentralized routing, region metadata, churn resilience, and VHash updates under membership and workload changes. Our extensive experiments on a real testbed show that Semord improves recall by more than 15% and reduces contacted peers by over 60% compared with decentralized baselines. Semord also approaches the recall and latency of a centralized oracle baseline while reducing peak peer-local ANN index memory by more than 2X. Controlled large-scale simulations further show that Semord scales across real-world embedding workloads and remains robust under churn for scoped vector retrieval as a decentralized overlay.

cs.NI↗

arXiv2026

Edge and spectral conditions for rainbow pancyclicity in graph collections

Let $\mathbf{G}=\{G_1,\dots,G_{n}\}$ be a collection of not necessarily distinct $n$-vertex graphs with a common vertex set $V$. A cycle $C$ with $V(C)\subseteq V$ and $|E(C)|\leq n$ is called \emph{rainbow} in $\mathbf{G}$, if there exists an injection $ϕ\colon E(C)\to [n]$ such that $e\in E(G_{ϕ(e)})$ for each $e\in E(C)$. The graph collection $\mathbf{G}$ is said to be \emph{rainbow pancyclic} if it contains a rainbow cycle of every length from 3 to $n$. In this paper, we show that if $e(G_i)\ge \binom{n-1}{2}+1$ for each $i\in[n]$ with $n\ge 3$, then $\mathbf{G}$ is rainbow pancyclic, apart from three explicitly described exceptional graph collections. This answers Problem $1$ of [Discrete Math., \textbf {348}(2025), 114600] and strengthens the result from rainbow Hamiltonicity to rainbow pancyclicity. As a consequence, we obtain that if $ρ(G_i)>n-2$ for each $i\in[n]$, then $\mathbf{G}$ is rainbow pancyclic unless $G_1=G_2=\dots=G_n\cong K_1\vee(K_{n-2}\cup K_1)$, which improves Theorem $5$ of [Discrete Math., \textbf {348}(2025), 114600]. We also characterize all graph collections that are not rainbow pancyclic under the condition $ρ(G_i)\ge n-2$ for each $i\in[n]$.

math.CO↗

arXiv2026

Breaking the Horizontal Prior: From Long-Tailed Orientation Bias to Roll-Robust Monocular Depth Estimation

Despite recent advances in Monocular Depth Estimation, state-of-the-art depth foundation models remain vulnerable to robustness issues. Particularly, even slight camera rolls can result in substantial degradation in depth estimations. We attribute this problem to a previously overlooked phenomenon, termed the Horizontal Prior, which is a manifestation of long-tailed distribution bias: most training images are captured in approximately horizontal orientations due to human visual preferences and photographic habits. While intuitive remedies such as re-balanced data augmentation and horizon leveling provide partial improvements, they fail to fully address the issue. In this paper, we introduce Invariant Depth Constraint (ID-Constraint), a training-time supervision strategy that improves roll robustness by fine-tuning and jointly regularizing the depth backbone with a series of geometric and spatial reasoning tasks. These auxiliary objectives encourage the backbone to learn rotation-stable, depth-relevant representations, while the auxiliary prediction heads are discarded after training, leaving the original inference architecture unchanged. Extensive experiments on five benchmark datasets across four roll settings demonstrate the effectiveness of the proposed method.

cs.CV↗

arXiv2026

Two problems on booksize and triangular edges in Nosal graphs

A graph $G$ with $m$ edges is said to be a Nosal graph if $ρ(G)>\sqrt{m}$. For a graph $G$, we write $bk(G)$ for its maximum book size and $τ(G)$ for the number of edges contained in triangles. Li, Liu and Zhang [J. Combin. Theory Ser. B 179 (2026) 219--249] proved that every $m$-edge Nosal graph satisfies $bk(G)> \frac{1}{24}\sqrt{m}$ and $τ(G) > \frac{1}{12}\sqrt{m}$. Recently, two results on the booksize constant are proved: $\frac{1}{9}$ by Zhai, Li and Lou [arXiv:2601.10163v2], and $\frac{1}{4}$ by Chen, Li and Tang [arXiv:2607.16746v1]. In this paper, we establish the following result: Every $m$-edge graph $G$ with no isolated vertices and $ρ(G)\geq \sqrt{m}$ that is not isomorphic to any complete bipartite graph satisfies $bk(G)\geq\frac{ρ(G)}{3}$ and $τ(G)\geq ρ(G)$. As direct consequences, we answer a question of Li, Liu and Zhang [J. Combin. Theory Ser. B 179 (2026) 219--249] and confirm a conjecture of Li, Feng and Peng [J. Graph Theory 110 (4) (2025) 408--425].

math.CO↗

arXiv2026

Rainbow panconnectivity in a graph collection

Let $\mathbf{G}=\{G_1,\dots,G_{n-1}\}$ be a collection of not necessarily distinct $n$-vertex graphs with the same vertex set $V$. A path $P$ with $V(P)\subseteq V$ and $|E(P)|\leq n-1$ is called \emph{rainbow} in $\mathbf{G}$, if there exists an injection $ϕ\colon E(P)\to [n-1]$ such that $e\in E(G_{ϕ(e)})$ for each $e\in E(P)$. The graph collection $\mathbf{G}$ is said to be \emph{rainbow panconnected} if for every pair of vertices $x,y\in V$, there exists a rainbow path of $k$ vertices joining $x$ and $y$ in $\mathbf{G}$ for every integer $k\in \left[d_{\mathbf{G}}(x,y)+1, n\right]$, where $d_{\mathbf{G}}(x,y)$ is the length of a shortest rainbow path between $x$ and $y$ in $\mathbf{G}$. In this paper, we study the rainbow panconnectivity of $\mathbf{G}$ under the minimum degree condition. Our result improves upon the corresponding results of [J. Graph Theory, \textbf{104}(2)(2023), 341--359] and [Electron. J. Combin., \textbf{32}(4)(2025), \#P4.17].

math.CO↗

arXiv2026

The saturation number of $K^s_t$

For a given graph $F$, a graph $G$ is said to be $F$-saturated if $G$ contains no copy of $F$ but for any edge $uv\notin E(G)$, $G+uv$ contains a copy of $F$. The saturation number $sat(n,F)$ is defined as the minimum number of edges among all $n$-vertex $F$-saturated graphs. The virus graph $K^s_t$, where $s\geq0$ and $t\geq \max\{3,s\}$, is a graph of order $s+t$ constructed by attaching $s$ distinct leaves to $s$ different vertices of a complete graph $K_t$. Hua and Peng [Discrete Math. 349 (2026) 114674] determined $sat(n,K^2_3)$ and characterized its corresponding extremal graphs. In this paper, we determine $sat(n,K^3_3)$ and $sat(n,K^2_t)$ with $t\geq 4$, together with the structural descriptions of the related extremal saturated graphs.

math.CO↗

arXiv2026

AnchorSplat: Feed-Forward 3D Gaussian Splatting with 3D Geometric Priors

Recent feed-forward Gaussian reconstruction models adopt a pixel-aligned formulation that maps each 2D pixel to a 3D Gaussian, entangling Gaussian representations tightly with the input images. In this paper, we propose AnchorSplat, a novel feed-forward 3DGS framework for scene-level reconstruction that represents the scene directly in 3D space. AnchorSplat introduces an anchor-aligned Gaussian representation guided by 3D geometric priors (e.g., sparse point clouds, voxels, or RGB-D point clouds), enabling a more geometry-aware renderable 3D Gaussians that is independent of image resolution and number of views. This design substantially reduces the number of required Gaussians, improving computational efficiency while enhancing reconstruction fidelity. Beyond the anchor-aligned design, we utilize a Gaussian Refiner to adjust the intermediate Gaussiansy via merely a few forward passes. Experiments on the ScanNet++ v2 NVS benchmark demonstrate the SOTA performance, outperforming previous methods with more view-consistent and substantially fewer Gaussian primitives.

cs.CV↗

arXiv2026

Transversal and Hamiltonicity in a bipartite graph collection

Let $\mathbf{G}=\{G_1,\dots,G_{s}\}$ be a collection of $s$ bipartite graphs with the same bipartition $V=(X,Y)$. For a path $P$ with $V(P)=V$ and $|E(P)|=s$, if there exists an injection $ϕ$: $E(P)\rightarrow [s]$ such that $e\in E(G_{ϕ(e)})$ for each $e\in E(P)$, then we say that the Hamiltonian path $P$ is a $\mathbf{G}$-transversal. A bipartite graph collection $\mathbf{G}$ is called Hamiltonian connected if for any two vertices $x\in X$ and $y\in Y$, there exists a $\mathbf{G}$-transversal isomorphic to a Hamiltonian path between $x$ and $y$. In this paper, we give the minimum degree conditions that ensure the existence of a $\mathbf{G}$-transversal isomorphic to a Hamiltonian path and the Hamiltonian connectivity of a balanced bipartite graph collection $\mathbf{G}$, which improve the results of [Hu, Li, Li and Xu, Discrete Math., 2024]. Moreover, we also provide a minimum degree condition that guarantees a nearly balanced bipartite graph collection $\mathbf{G}$ contains a $\mathbf{G}$-transversal isomorphic to a Hamiltonian path.

math.CO↗

arXiv2026

PlanetServe: A Decentralized, Scalable, and Privacy-Preserving Overlay for Democratizing Large Language Model Serving

While significant progress has been made in research and development on open-source and cost-efficient large-language models (LLMs), serving scalability remains a critical challenge, particularly for small organizations and individuals seeking to deploy and test their LLM innovations. Inspired by peer-to-peer networks that leverage decentralized overlay nodes to increase throughput and availability, we propose GenTorrent, an LLM serving overlay that harnesses computing resources from decentralized contributors. We identify four key research problems inherent to enabling such a decentralized infrastructure: 1) overlay network organization; 2) LLM communication privacy; 3) overlay forwarding for resource efficiency; and 4) verification of serving quality. This work presents the first systematic study of these fundamental problems in the context of decentralized LLM serving. Evaluation results from a prototype implemented on a set of decentralized nodes demonstrate that GenTorrent achieves a latency reduction of over 50% compared to the baseline design without overlay forwarding. Furthermore, the security features introduce minimal overhead to serving latency and throughput. We believe this work pioneers a new direction for democratizing and scaling future AI serving capabilities.

cs.DC↗

arXiv2025

Saturation numbers of $K_{2}\vee P_{k}$

A graph $G$ is called $H$-saturated if $G$ contains no copy of $H$, but $G+e$ contains a copy of $H$ for any edge $e\in E(\overline{G})$. The saturation number of $H$ is the minimum number of edges in an $H$-saturated graph of order $n$, denoted by $sat(n,H)$. In this paper, we investigate $sat(n,K_{2}\vee P_{k})$, where $k\geq 3$. Let $a_k$ be an integer, defined as follows: $a_k=k$ for $3\leq k\leq 5$; $a_k=3\cdot 2^{t-1}-2$ for $k=2t\geq 6$; and $a_k=2^{t+1}-2$ for $k=2t+1\geq 7$. We show that $sat(n, K_{2}\vee P_{k})=2n-3+sat(n-2,P_{k})$ for $n\geq a_k+2$ and $k\geq 3$, characterize the $K_{2}\vee P_{k}$-saturated graphs with $sat(n,K_{2}\vee P_{k})$ edges, the $K_{1}\vee P_{k}$-saturated graphs with $sat(n,K_{1}\vee P_{k})$ edges for $3\leq k\leq5$ and the $P_{k}$-saturated graphs with $sat(n, P_{k})$ edges for $3\leq k\leq4$. Furthermore, we propose some questions for further research.

math.CO↗

arXiv2025

A Distributed Learned Hash Table

Distributed Hash Tables (DHTs) are pivotal in numerous high-impact key-value applications built on distributed networked systems, offering a decentralized architecture that avoids single points of failure and improves data availability. Despite their widespread utility, DHTs face substantial challenges in handling range queries, which are crucial for applications such as LLM serving, distributed storage, databases, content delivery networks, and blockchains. To address this limitation, we present LEAD, a novel system incorporating learned models within DHT structures to significantly optimize range query performance. LEAD utilizes a recursive machine learning model to map and retrieve data across a distributed system while preserving the inherent order of data. LEAD includes the designs to minimize range query latency and message cost while maintaining high scalability and resilience to network churn. Our comprehensive evaluations, conducted in both testbed implementation and simulations, demonstrate that LEAD achieves tremendous advantages in system efficiency compared to existing range query methods in large-scale distributed systems, reducing query latency and message cost by 80% to 90%+. Furthermore, LEAD exhibits remarkable scalability and robustness against system churn, providing a robust, scalable solution for efficient data retrieval in distributed key-value systems.

cs.NI↗

arXiv2025

Spatially inhomogeneous field induced harmonic radiation in solids

By theoretical derivation, we constructed an inhomogeneous coefficient equation to correctly describing harmonic radiation in solids induced by a spatially inhomogeneous field, where the widely used semiconductor Bloch equation fails. This equation has superiority over the semiconductor Bloch equation with good applicability to both homogeneous and inhomogeneous fields. Using graphene as an example, it is found that under inhomogeneous field driving, even-order harmonics occur with an enhancing tendency as the field inhomogeneity increases. As for the second-order harmonic, its intensity dependence is consistent with the prediction from the perturbation theory, and its wavelength dependence can use to directly distinguish the relative contribution of intraband and interband transitions. The inhomogeneous coefficient equation provides a direct theoretical analysis tool for elucidating the physical mechanism of inhomogeneous field induced harmonic radiation in solids.

physics.optics↗

arXiv2025

The connectedness of friends-and-strangers graphs about graph parameters and others

Let $X$ and $Y$ be two graphs of order $n$. The friends-and-strangers graph $\textup{FS}(X,Y)$ of $X$ and $Y$ is a graph whose vertex set consists of all bijections $σ: V(X)\rightarrow V(Y)$, in which two bijections $σ$ and $ σ'$ are adjacent if and only if they agree on all but two adjacent vertices of $X$ such that the corresponding images are adjacent in $Y$. The most fundamental question about these friends-and-strangers graphs is whether they are connected. In this paper, we provide a sufficient condition regarding the maximum degree $Δ(X)$ and vertex connectivity $κ(Y)$ that ensures the graph $\textup{FS}(X,Y)$ is $s$-connected. As a corollary, we improve upon a result by Bangachev and partially confirm a conjecture he proposed. Furthermore, we completely characterize the connectedness of $\textup{FS}(X,Y)$, where $X\in\textup{DL}_{n-k,k}$.

math.CO↗

arXiv2024

Towards Practical Overlay Networks for Decentralized Federated Learning

Decentralized federated learning (DFL) uses peer-to-peer communication to avoid the single point of failure problem in federated learning and has been considered an attractive solution for machine learning tasks on distributed devices. We provide the first solution to a fundamental network problem of DFL: what overlay network should DFL use to achieve fast training of highly accurate models, low communication, and decentralized construction and maintenance? Overlay topologies of DFL have been investigated, but no existing DFL topology includes decentralized protocols for network construction and topology maintenance. Without these protocols, DFL cannot run in practice. This work presents an overlay network, called FedLay, which provides fast training and low communication cost for practical DFL. FedLay is the first solution for constructing near-random regular topologies in a decentralized manner and maintaining the topologies under node joins and failures. Experiments based on prototype implementation and simulations show that FedLay achieves the fastest model convergence and highest accuracy on real datasets compared to existing DFL solutions while incurring small communication costs and being resilient to node joins and failures.

cs.DC↗

arXiv2024

Inflight Performance and Calibrations of the Lyman-alpha Solar Telescope on board the Advanced Space-based Solar Observatory

The Lyman-alpha Solar Telescope (LST) on board the Advanced Space-based Solar Observatory (ASO-S) is the first payload to image the full solar disk and the solar corona in both white-light (WL) and ultraviolet (UV) H I Lya, extending up to 2.5 solar radii (Rs). Since the launch of the ASO-S on 9 October 2022, LST has captured various significant solar activities including flares, prominences, coronal mass ejections (CMEs). LST covers different passbands of 121.6 nm, 360 nm and 700 nm. The Lya Solar Disk Imager (SDI) has a field of view (FOV) of 38.4 arcmin and a spatial resolution of around 9.5 arcsec, while the White-Light Solar Telescope (WST) has a FOV of 38.43 arcmin and a spatial resolution of around 3.0 arcsec. The FOV of the Lya Solar Corona Imager (SCI) reaches 81.1 arcmin and its spatial resolution is 4.3 arcsec. The stray-light level in the 700 nm waveband is about 7.8e-6 MSB (mean solar brightness) at 1.1 Rs and 7.6e-7 MSB at 2.5 Rs, and in the Lya waveband it is around 4.3e-3 MSB at 1.1 Rs and 4.1e-4 MSB at 2.5 Rs. This article will detail the results from on-orbit tests and calibrations.

astro-ph.SR↗

arXiv2024

Secure Decentralized Learning with Blockchain

Federated Learning (FL) is a well-known paradigm of distributed machine learning on mobile and IoT devices, which preserves data privacy and optimizes communication efficiency. To avoid the single point of failure problem in FL, decentralized federated learning (DFL) has been proposed to use peer-to-peer communication for model aggregation, which has been considered an attractive solution for machine learning tasks on distributed personal devices. However, this process is vulnerable to attackers who share false models and data. If there exists a group of malicious clients, they might harm the performance of the model by carrying out a poisoning attack. In addition, in DFL, clients often lack the incentives to contribute their computing powers to do model training. In this paper, we proposed Blockchain-based Decentralized Federated Learning (BDFL), which leverages a blockchain for decentralized model verification and auditing. BDFL includes an auditor committee for model verification, an incentive mechanism to encourage the participation of clients, a reputation model to evaluate the trustworthiness of clients, and a protocol suite for dynamic network updates. Evaluation results show that, with the reputation mechanism, BDFL achieves fast model convergence and high accuracy on real datasets even if there exist 30\% malicious clients in the system.

cs.CR↗

arXiv2022

A universal and improved mutation strategy for iterative wavefront shaping

Recent advances in iterative wavefront shaping (WFS) techniques have made it possible to manipulate the light focusing and transport in scattering media. To improve the optimization performance, various optimization algorithms and improved strategies have been utilized. Here, a novel guided mutation (GM) strategy is proposed to improve optimization efficiency for iterative WFS. For both phase modulation and binary amplitude modulation, considerable improvements in optimization effect and rate have been obtained using multiple GM-enhanced algorithms. Due of its improvements and universality, GM is beneficial for applications ranging from controlling the transmission of light through disordered media to optical manipulation behind them.

physics.optics↗

arXiv2022

High fidelity generation of complex optical field through scattering medium with iterative wavefront optimization

Light scattering within scattering media presents a substantial obstacle to optical transmission. A speckle pattern with random amplitude and phase distribution is observed when coherent light travels through strong scattering media. Fortunately, wavefront shaping has been successfully employed with a spatial light modulator to recover intensity targets after scattering media, such as a sharp focus point or specified two-dimensional patterns. There have, however, been few studies that attempted to separately manipulate the amplitude and phase of the focusing field. In this paper, we propose a feedback-based wavefront shaping method to generate complex optical fields through scattering medium. A reliable phase retrieval approach is introduced to provide the complex feedback information, i.e., the amplitude and phase of the focusing field. Accordingly, in order to modulate the speckle field into a desired complex structured optical field, a multi-objective genetic algorithm is used to find the best phase map. To demonstrate the proposed method's high performance, experimental tests have been carried out. High fidelity is demonstrated in the generation of diverse complex light fields, both in amplitude and phase. Our findings may facilitate the manipulation of light field through scattering medium, and are anticipated to further promote future applications such as optogenetics, vortex optical communication, and optical trapping through scattering media.

physics.optics↗