Searcharxiv⌕ Search

arXiv subjects

Xin Cheng

Publications and source records attributed to Xin Cheng.

At least 19 recordsLinked to original sources
arXiv2026

Electron acceleration by turbulent reconnection in solar flares

Solar flares can release magnetic energy explosively in the corona and produce high-energy particles on short timescales. However, how and where these particles are accelerated remains an open question. Here, we investigate the acceleration and transport of electrons during self-developed three-dimensional turbulent reconnection of solar flares by solving Parker's transport equation in the framework of high-resolution MHD simulations. We find that thermal electrons at both the flare current sheet and loop top are rapidly accelerated up to ~90keV, with energy spectra exhibiting a power-law feature. Although the population of accelerated electrons at the flare loop top is larger than that at the current sheet, their spectral indices are similar, close to the values usually observed. More importantly, the acceleration is achieved by turbulence-driven compression structures of various scales rather than the supposed termination shock, particularly at the flare loop top. A portion of compression structures even forms shocks. These results highlight the critical role of turbulent reconnection in accelerating electrons, thereby shedding new light on the acceleration and transport of particles in other high-energy phenomena.

astro-ph.SR↗
arXiv2026

On the Tur\'an number of the expansion of the book

The book with $t$ pages is the graph on $t+2$ vertices consisting of $t$ triangles which intersect at exactly one common edge. For a given graph $F$, the $r$-expansion $F^r$ of $F$ is the $r$-uniform hypergraph obtained from $F$ by adding $r-2$ distinct new vertices to each edge of $F$. We determine the Tur\'an number of the 3-expansion of the book graph for sufficiently large $n$.

math.CO↗
arXiv2026

FATE: Frame-Level Audio-Visual Temporal Embedding

When a dog opens its mouth and barks, humans naturally recognize what the sound is and when it occurs. Building audio-visual models with this same ability requires representations that capture both semantic and temporal alignment. Current approaches fall short on one side or the other: embedding models match semantic but lose temporal information; synchronization models capture temporal offsets but lack semantic understanding. To bridge this gap, we propose FATE, Frame-level Audio-visual Temporal Embedding. Unlike prior embedding models that pool each modality into a single embedding and discard temporal information, FATE retains frame-level sequences, aligns them on the physical timeline, and computes similarity over strictly aligned frame pairs. Unlike synchronization models that output only an offset prediction, FATE encodes synchronization in a reusable embedding space, trained with a joint objective combining cross-video semantic and within-video temporal contrastive learning to capture both what sounds and when it occurs. Across three tasks, FATE surpasses the strongest baseline on temporal and semantic retrieval by a large margin, matches fully supervised methods on event localization in a zero-shot setting, and achieves the best correlation with human judgments as a generation evaluation metric. The source code can be found at \texttt{https://github.com/guankaisi/FATE}.

cs.MM↗
arXiv2026

Back Reaction of the Untwisting Solar Corona Scars Sunspots

The evolution of magnetic fields in the tenuous solar corona is predominantly governed by the motions of the underlying dense photosphere. Despite, coronal magnetic restructuring driven by magnetic reconnection between interacting coronal fields can sometimes react backwards to change photospheric magnetic fields. However, the mechanism of reactions remains undetermined. Here, we report the discovery of a back-reaction phenomenon: the untwisting of coronal loops that become twisted during reconnection in an eruption results in enhanced currents at the boundary of their footpoint away from the eruption, manifesting as the growth of a sunspot scar. It is revealed to arise from the Alfvenic reverse transfer of magnetic twist from the corona to the lower atmosphere, thanks to joint space observations and a magnetohydrodynamics simulation. These findings provide a viable and quantitative interpretation for the majority of puzzling photospheric changes associated with coronal mass ejections and/or flares and warn for unexpected magnetic field evolutions in sunspots and starspots.

astro-ph.SR↗
arXiv2026

Joint Channel Estimation and Data Detection for Multi-LEO-Satellite Cell-Free OTFS Uplinks

Cell-free networks formed by multiple low Earth orbit (LEO) satellites offer a promising architecture for ubiquitous connectivity, but their cooperative reception is challenged by link-dependent residual delays and Doppler shifts. This paper investigates joint channel estimation and data detection (JCEDD) for multi-LEO-satellite cell-free orthogonal time frequency space (OTFS) uplinks. The JCEDD problem is formulated as a structured bilinear inference problem involving link-specific sparse beam--delay--Doppler channels and a multiuser data vector. We develop a low-complexity hierarchical JCEDD receiver in which all satellites first perform local JCEDD, and their observations and local estimates are then aggregated at a central satellite for cooperative refinement. Computational complexity is reduced by restricting channel estimation to coarse-information-aided local beam--delay--Doppler regions and evaluating the required forward and adjoint operations in a matrix-free manner. Simulation results validate the channel-estimation accuracy and data-detection reliability of the proposed JCEDD receiver.

eess.SP↗
arXiv2026

DSpark: Confidence-Scheduled Speculative Decoding with Semi-Autoregressive Generation

Speculative decoding accelerates Large Language Model (LLM) inference by decoupling draft generation from target verification. While recent parallel drafters efficiently propose long token sequences in a single forward pass, they suffer from rapid acceptance decay due to a lack of inter-token dependencies. Furthermore, indiscriminately verifying these extended blocks wastes critical batch capacity on tokens with high rejection risks, severely degrading throughput in high-concurrency serving systems. We introduce DSpark, a speculative decoding framework that unifies high-throughput parallel generation with adaptive, load-aware verification. To maintain draft quality, DSpark utilizes a semi-autoregressive architecture, coupling a parallel backbone with a lightweight sequential module, to introduce intra-block dependency modeling and mitigate suffix decay. To optimize system efficiency, DSpark employs confidence-scheduled verification, dynamically tailoring the verification length for each request based on estimated prefix survival probabilities and engine-specific throughput profiles. On offline benchmarks across diverse domains, DSpark substantially improves the accepted length over state-of-the-art autoregressive and parallel drafters. When deployed within the DeepSeek-V4 serving system under live user traffic, DSpark successfully mitigates verification waste. Compared to the established production baseline (MTP-1), DSpark accelerates per-user generation speeds by 60 to 85 percent at matched throughput levels. More importantly, by preventing severe throughput degradation under strict interactivity constraints, it enables performance tiers that were previously unattainable, shifting the Pareto frontier of our serving system.

cs.AI↗
arXiv2026

AVOC: Enhancing Hour-Level Audio-Video Understanding in Omni-Modal LLMs via Retrieval-Inspired Token Compression

Multimodal Large Language Models have achieved remarkable progress in short-form audio-video understanding, yet long-form audio-video comprehension remains challenged by limited context windows and severe information redundancy. To address these bottlenecks, we propose AVOC, a framework for long-form audio-video understanding in Omni-modal Large Language Models. AVOC introduces a learnable token compression module between the modality encoders and the LLM backbone. We reframe multimodal token compression as a top-$K$ retrieval problem: given a fixed context budget, the module must retrieve a compact subset of tokens that best supports answering the user query. We draw inspiration from three classical Information Retrieval criteria for selecting informative units from a large candidate pool: relevance, importance, and diversity. AVOC instantiates each criterion as a tailored mechanism for audio-video understanding, and integrates them into a unified retrieval-style compression pipeline. Experiments show that AVOC achieves state-of-the-art performance on long-form audio-video benchmarks, surpassing the second-best model by 4.9 and 5.5 points in average accuracy on OmniVideoBench and LVOmniBench, respectively. Moreover, AVOC maintains robust performance on Audio-Video Needle-in-a-Haystack task at durations up to one hour.

cs.CL↗
arXiv2026

SeFi-Image: A Text-to-Image Foundation Model with Semantic-First Diffusion

Training image generation foundation models consumes substantial resources. Previous methods have attempted to leverage semantic guidance to accelerate the training process, yet their experiments were only conducted on simple datasets such as ImageNet, at low resolutions, and with small-scale models. In this paper, we propose SeFi-Image, a text-to-image foundation model built upon semantic-first diffusion, a novel latent diffusion modeling paradigm. We instantiate SeFi-Image at three model scales, 1B, 2B, and 5B parameters, enabling systematic study of scaling behavior and flexible deployment under varying compute budgets. Notably, our largest 5B model was trained with merely 125K A800 GPU hours, corresponding to roughly 10-20% of the training compute used by Z-Image. However, it achieves results comparable to or even superior to Qwen-Image and Z-Image. Despite this modest training compute, SeFi-Image achieves strong performance on a wide range of benchmarks, including GenEval, DPG, LongTextBench, OneIG, and CVTG-2K. Moreover, we provide DMD2-distilled few-step turbo variants for each model scale to accommodate diverse hardware constraints and latency requirements. We publicly release our code, weights and hope this work offers the community useful insights into semantic-guided diffusion modeling for T2I generation, while also providing practical and readily deployable model options.

cs.CV↗
arXiv2026

AgenticOS: An Intent-Oriented Secure Operating System Architecture for Autonomous AI Agents

Traditional OS security models based on "resource exposure plus permission checks" face structural challenges as LLM-driven autonomous agents acquire capabilities for planning, tool use, network access, and code execution. Once an agent runtime is compromised through prompt injection or malicious tool outputs, an attacker can compose POSIX-style resource primitives into behaviors far beyond the user's task authorization. To address this, we propose AgenticOS, an intent-oriented secure OS architecture that consolidates delegable, auditable software capabilities into OS-native ones rather than replacing all applications. The core insight is to reframe the OS from a "resource manager" into an "intent filter": instead of requesting low-level resources directly, agents submit structured intent declarations, from which the system synthesizes a least-privilege environment with mandatory mediation, auditing, and information-flow constraints. At the implementation level, we introduce a four-layer architecture -- Ghost Kernel, Logic Shutter, Agent Capsule, and Semantic Boundary Gateway -- together with the Intent ABI, Manifest-Only Runtime, Weaver-based capability generation, and an admission model for AgenticOS-native Skills.

cs.CR↗
arXiv2026

Unified Synthesis of Compositional Speech and Sound from Free-Form Text Prompts

Audio generation has made significant progress, yet synthesizing unified audio where speech and sounds are naturally composited remains a challenge. Current methods either rely on disjoint pipelines, which fail to capture fine-grained interactions, or require structured inputs and external text rewriting, which limits the flexibility of free-form text prompts. In this paper, we introduce a new task: Free-Form-Text-Prompt-to-Unified-Audio generation, which aims to directly synthesize unified audio containing speech, sound, and their composites from unconstrained natural language. To address this task, we propose PlanAudio, a unified, autoregressive LLM-based framework. First, it simplifies the model architecture by leveraging intrinsic LLM reasoning capability instead of traditional text encoders. Second, it introduces a semantic latent chain-of-thought mechanism, an implicit planning mechanism that bridges high-level semantic understanding and low-level acoustic synthesis. Furthermore, we create PlanAudio-Bench, a specialized benchmark for evaluating composite audio scenarios. We perform evaluations in the scenarios of speech, sound, and their composites. The results demonstrate that PlanAudio generally outperforms the existing pipeline and unified baselines, while staying competitive with models designed for a single scenario. Our analysis further reveals the superiority of semantic latent CoT over other CoT mechanisms and highlights the importance of continuous multi-scenario training curricula.

cs.SD↗
arXiv2026

Beyond Trajectory-Level Attribution: Graph-Based Credit Assignment for Agentic Reinforcement Learning

Group-based reinforcement learning (RL) methods have achieved remarkable success in improving the performance of large language models (LLMs) and have been rapidly extended to agentic tasks. However, their credit assignment relies heavily on coarse-grained trajectory-level attribution according to final outcomes, making it difficult to capture the contribution of individual steps, such as valuable steps obscured within failed trajectories. To uncover latent information and enable more faithful step-level credit assignment, we propose Graph-based Group Policy Optimization (GraphGPO), which first aggregates all rollout trajectories into a unified state-transition graph and then estimates the distance from each state to the task goal using the global information encoded in the graph. Finally, GraphGPO assigns credit to each edge by estimating a graph-based advantage, based on how much the transition reduces the distance to the task goal. In this way, GraphGPO significantly improves training efficiency and achieves state-of-the-art performance across a range of challenging benchmarks.

cs.LG↗
arXiv2026

On the Tur\'an number of blow-ups of $\mathcal{F}_5$

Let $\mathcal{F}_5$ denote the $3$-uniform hypergraph on the vertex set $\{f_1,f_2,\dots,f_5\}$ with hyperedges $\{f_1f_2f_3,f_1f_2f_4,f_3f_4f_5\}$. Recently, Balogh, Clemen and Luo determined the Tur\'an number of a one-vertex blow-up of $\mathcal{F}_5$, more specifically, they blow up the vertex $f_5$ to $t$ vertices, the resulting hypergraph is denoted by $\mathcal{F}_5(f_5;t)$. They show that for infinitely many $t$, $\mathcal{F}_5(f_5;t)$ has exponentially many extremal constructions and positive Tur\'an density. In this paper, we determine the exact Tur\'an number of the hypergraph obtained by blowing up $f_3$ of $\mathcal{F}_5$ to $t$ vertices and show that it also has exponentially many extremal constructions. We also give a general upper bound and lower bound of the Tur\'an number of every blow-up of $\mathcal{F}_5$. For some special blow-ups of $\mathcal{F}_5$, for example, $t$-disjoint copies of $\mathcal{F}_5$, we determine the exact Tur\'an number. We construct a hypergraph $\mathcal{F}_{sim}(t)$ which is a subgraph of a blow-up of $\mathcal{F}_5$, and is contained in the hypergraph obtained by adding any new hyperedge to the Tur\'an hypergraph (the balanced complete $3$-partite hypergraph), but its extremal construction is not the Tur\'an hypergraph. We also determine the exact Tur\'an number of $\mathcal{F}_{sim}(t)$.

math.CO↗
arXiv2026

SyncDPO: Enhancing Temporal Synchronization in Video-Audio Joint Generation via Preference Learning

Recent advancements in video-audio joint generation have achieved remarkable success in semantic correspondence. However, achieving precise temporal synchronization, which requires fine-grained alignment between audio events and their visual triggers, remains a challenging problem. The post-training method for joint generation is largely dominated by Supervised Fine-Tuning, but the commonly used Mean Squared Error loss provides insufficient penalties for subtle temporal misalignments. Direct Preference Optimization offers an alternative by introducing explicit misaligned counterparts to better improve temporal sensitivity. In this paper we propose a post-training framework SyncDPO, leveraging DPO to improve the temporal sensitivity of V-A joint generation. Conventional DPO pipelines typically depend on costly sampling-and-ranking procedures to construct preference pairs, resulting in substantial computational cost. To improve efficiency, we introduce a suite of on-the-fly rule-based negative construction strategies that distort temporal structures without incurring additional annotation or sampling. We demonstrate that the temporal alignment capability can be effectively reinforced by providing explicit negative supervision through temporally distorted V-A pairs. Accordingly, we implement a curriculum learning strategy that progressively increases the difficulty of negative samples, transitioning from coarse misalignment to subtle inconsistencies. Extensive objective and subjective experiments across four diverse benchmarks, ranging from ambient sound videos to human speech videos, demonstrate that SyncDPO significantly outperforms other methods in improving model's temporal alignment capability. It also demonstrates superior generalization on out-of-distribution benchmark by capturing intrinsic motion-sound dynamics. Demo and code is available in https://syncdpo.github.io/syncdpo/.

cs.CV↗
arXiv2026

VertMark: A Unified Training-Free Robust Watermarking Framework for Vertical Domain Pre-trained Language Models

With the application of vertical domain pre-trained language models (VPLMs) in specialized fields such as medical, finance, and law, model parameters and inference capabilities have become important digital assets. Achieving traceable copyright verification for VPLMs has become an urgent challenge. Existing copyright verification methods primarily rely on embedding backdoor watermarks into models. However, most of these methods require additional training, suffer from inefficient watermark embedding, and lack scalable designs for multiple vertical domains. To address these limitations, we propose VertMark, the first unified training-free and robust watermarking framework for copyright verification across multiple vertical domain VPLMs. The framework embeds ownership-encoded watermarks by establishing a hidden semantic equivalence between low-frequency trigger tokens and high-frequency domain-relevant words via a training-free parameter replacement strategy. Experiments demonstrate that VertMark can achieve efficient watermark embedding and reliable watermark verification for both text understanding and text generation downstream tasks in the medical, financial, and legal domains, with negligible impact on model performance. Moreover, VertMark exhibits strong robustness against various attacks (e.g., pruning and quantization), highlighting its practical value and providing strong protection for the copyright security of VPLMs.

cs.CR↗
arXiv2026

Generalized Tur\'an problems for Berge hypergraphs

Let $\mathcal{H}$ be a hypergraph and $F$ be a graph. If there exists a bijection between the hyperedges of $\mathcal{H}$ and the edges of $F$ such that each hyperedge contains its image, then we say that $\mathcal{H}$ is a \textit{Berge copy} of $F$, and the collection of Berge copies of $F$ is denoted by Berge-$F$. Given $r$-graphs $\mathcal{F}$ and $\mathcal{H}$, the generalized hyper-Tur\'{a}n number $\text{ex}_r(n, \mathcal{H}, \mathcal{F})$ is the maximum number of copies of $\mathcal{H}$ in $n$-vertex $\mathcal{F}$-free $r$-graphs. We study $\text{ex}_r(n, \mathcal{H}, \text{Berge-}F)$. For general $\mathcal{H}$, we connect this problem to counting copies of the shadow graph of $\mathcal{H}$ in $F$-free graphs and obtain several exact results. In particular, we show that for any hypergraph $\mathcal{H}$, if $k$ is sufficiently large, then $\text{ex}_r(n, \mathcal{H}, \text{Berge-}K_k)$ is achieved by the balanced complete $(k-1)$-partite $r$-graph, generalizing a result of Morrison, Nir, Norin, Rza{\.z}ewski and Wesolek [\textit{Journal of Combinatorial Theory, Series B}, 162 (2023) 231--243] to the case of hypergraphs. We show that $\text{ex}_r(n,K_s^r,\text{Berge-}F)\le \text{ex}_s(n,\text{Berge-}F)$ and present sufficient conditions for equality. We also consider the connected generalized Tur\'{a}n number for Berge paths.

math.CO↗
arXiv2026

The Tur\'{a}n number of the Cartesian product of a star and an edge

Let $C_k$ denote the cycle of length $k$, $S_t$ be a star with $t$ edges. And let $B_t$ be the graph consisting of $t$ copies of $C_4$ sharing one fixed edge. Equivalently, $B_t=K_2 \mathbin{\square} S_t$, which is the Cartesian product of a star with $t$ edges and an edge. Recently, Gao, Janzer, Liu and Xu [\textit{Israel J. Math. 269(2025)}] proved that the Tur\'an number of $K_2\mathbin{\square} C_{2l}$ is $\Theta(n^{\frac{3}{2}})$ for every $l\ge 4$. In this paper, we obtain upper and lower estimates for the Tur\'an number of $B_t$ in both the general and bipartite settings for every $t\geq 2$. For the lower bound, we use random construction based on the extremal structure of $C_4$. These results imply that $\frac{1}{2\sqrt{2}}\leq \lim_{t\to \infty} \frac{\mathrm{ex}(n,B_t)}{\sqrt{t}}\leq \frac{1}{2}$, and $\frac{1}{4}\leq \lim_{t\to \infty} \frac{\mathrm{ex}_{bip}(n,B_t)}{\sqrt{t}}\leq \frac{1}{2\sqrt{2}}.$ In the case of $B_2$, we obtain sharper estimates. We show that the Tur\'an number of $B_2$ is approximately between $(0.518+o(1))n^{\frac{3}{2}}$ and $(0.603+o(1))n^{\frac{3}{2}}$. And in the bipartite setting, it is approximately between $(0.385+o(1))n^{\frac{3}{2}}$ and $(0.468+o(1))n^{\frac{3}{2}}$. Moreover, in the bipartite setting, we give a more general result, which shows that for every tree $T$ with $t$ edges, the bipartite Tur\'an number of $K_2\mathbin{\square}T$ is at most $\frac{\sqrt{t}}{2\sqrt{2}}(1+o(1))n^{\frac{3}{2}}$.

math.CO↗
arXiv2026

Forbidding matching as trace in uniform hypergraphs

We say a hypergraph $\mathcal{H}$ contains a hypergraph $\mathcal{G}$ as trace if there exists a vertex subset $S \subseteq V(\mathcal{H})$ such that $|S| = |V(\mathcal{G})|$ and $\{e \cap S: e \in E(\mathcal{H})\}$ contains $\mathcal{G}$ as a sub-hypergraph. We use $\mathrm{ex}_r(n, \mathrm{Tr}_r(\mathcal{G}))$ to denote the maximum number of hyperedges in an $r$-uniform hypergraph on $n$ vertices not containing $\mathcal{G}$ as a trace. The study of Tur\'{a}n numbers for traces was initiated by Mubayi and Zhao who studied the case when $\mathcal{G}$ is a complete graph. Let $M_{s+1}$ denote the graph of a matching with $s+1$ edges. In this paper, we give the upper bound of $\mathrm{ex}_r(n, \mathrm{Tr}_r(M_{s+1}))$ which is sharp asymptotically. When $r=3$, we give the exact value of $\mathrm{ex}_3 (n, \mathrm{Tr}_3 (M_{s+1}))$. We also consider the generalized Tur\'{a}n number in the case of matching. That is, the maximum number of copies of clique $\mathcal{K}_t^r$ in hypergraphs forbidding $\mathrm{Tr}_r (M_{s+1})$ as a trace. We give an upper bound which is sharp asymptotically and when $r=3$, we give the exact value. The Tur\'{a}n number of forbidding a matching and the other graph is another well studied topic initiated by Alon and Frankl. We also consider an analogue problem for the trace version, i.e., forbidding trace of matching and trace of complete graph as subgraphs.

math.CO↗
arXiv2026

CUTEv2: Unified and Configurable Matrix Extension for Diverse CPU Architectures with Minimal Design Overhead

Matrix extensions have emerged as an essential feature in modern CPUs to address the surging demands of AI workloads. However, existing designs often incur substantial hardware and software design overhead. Tight coupling with the CPU pipeline complicates integration across diverse CPUs, while fine-grained synchronous instructions hinder the development of high-performance kernels. This paper proposes a unified and configurable CPU matrix extension architecture. By decoupling matrix units from the CPU pipeline, the design enables low-overhead integration while maintaining close coordination with existing compute and memory resources. The configurable matrix unit supports mixed-precision operations and adapts to diverse compute demands and memory bandwidth constraints. An asynchronous matrix multiplication abstraction with flexible granularity conceals hardware details, simplifies matrix-vector overlap, and supports a unified software stack. The architecture is integrated into four open-source CPU RTL platforms and evaluated on representative AI models. Matrix unit utilization under GEMM workloads exceeds 90% across all platforms. When configured with compute throughput and memory bandwidth comparable to Intel AMX, our design achieves speedups of 1.57x, 1.57x, and 2.31x on ResNet, BERT, and Llama3, with over 30% of the gains attributed to overlapped matrix-vector execution. A 4 TOPS@2GHz matrix unit occupies only 0.53 mm\textsuperscript{2} in 14nm CMOS. These results demonstrate strong cross-platform adaptability and effective hardware-software co-optimization, offering a practical matrix extension for the open-source community.

cs.AR↗