Searcharxiv⌕ Search

arXiv subjects

Pengcheng Zhang

Publications and source records attributed to Pengcheng Zhang.

At least 19 recordsLinked to original sources
arXiv2026

Direct Satellite-to-Device Communications: From Cooperative Task Offloading to Non-Cooperative Access Monitoring

Direct satellite-to-device (DS2D) communication is emerging as a transformative paradigm for extending ubiquitous connectivity and edge computing capabilities to remote and underserved regions within 6G non-terrestrial networks. However, practical deployment faces dual critical challenges: i) dynamic satellite channel conditions (e.g., severe Doppler shifts, fast fading) and constrained satellite computing resources in cooperative scenarios; and ii) unauthorized satellite access introduces significant spectrum security threats in non-cooperative scenarios. To address these challenges, we propose a versatile DS2D system that supports cooperative task offloading and non-cooperative access monitoring. For cooperative DS2D communications, we integrate a channel estimation module with a dueling double deep Q-network (D3QN) to dynamically optimize task offloading strategy. For non-cooperative DS2D communications, we propose Transformer-based models to enable blind signal detection and automatic modulation classification (AMC). Simulation results show that: 1) The D3QN algorithm reduces average latency by up to 225\% compared to static association policies. 2) Our signal detection model achieves an average presence detection probability of 90.5\% for DS2D signals. 3) The proposed AMC algorithm achieves superior performance across different signal-to-noise ratios (SNRs), with a 9.4\% accuracy gain in low-SNR regimes compared to existing methods.

cs.IT↗
arXiv2026

TEngineDB-V: An OLAP-Native Vector Search System for Large-$k$ Workloads at Tencent

Vector search systems are essential infrastructure for modern data-driven applications. Large-$k$ analytical vector search, which retrieves $k=10^3$--$10^5$ results for analytics (e.g., aggregation, filtering, joins), is increasingly important for emerging workloads, including LLM data management and advertising analysis at Tencent. Existing systems remain inadequate: specialized vector databases often cap $k$ (e.g., $k \leq 10^4$) to satisfy tail-latency constraints and offer limited analytical support, while OLAP systems typically embed per-segment vector indexes as black boxes, causing severe read/compute amplification and preventing native query optimization. This paper presents TEngineDB-V, an OLAP-native vector search system for large-$k$ workloads. TEngineDB-V makes vector search a first-class analytical primitive in Tencent's OLAP engine through a global segment-decoupled index materialized as relational tables, eliminating scatter-gather execution, reducing amplification, and enabling native storage optimizations. It decomposes IVFPQ-based search into relational operators, integrates OLAP optimizations, and introduces DPPQ, which combines direction-aware quantization with hierarchical residual refinement to improve recall while preserving relational efficiency. TEngineDB-V further incorporates index-aware query rewriting and a distributed-aware cost model for efficient distributed execution. Experiments show that TEngineDB-V achieves up to a $145\times$ speedup over competitive systems such as StarRocks, and up to a $52\times$ improvement in 10-billion-scale production deployments.

cs.DB↗
arXiv2026

CLIP: Lightweight Cosine-Law-Based Inverted-List Pruning for IVF-Based Vector Search

Vector search has become a core component of modern multimodal retrieval systems. Among existing methods, inverted file (IVF)-based methods are widely adopted due to their scalability, efficient updates, and hardware friendliness. However, they are fundamentally limited by coarse-grained execution: each query typically probes many clusters and exhaustively scans all vectors within them, resulting in high query latency. Prior works mitigate this using pruning strategies, but they often incur substantial extra pruning overhead, lack cluster-level pruning, and compromise update efficiency due to heavy maintenance of pruning metadata. This paper proposes CLIP, a lightweight cosine-law-based pruning technique that supports both inter- and intra-cluster pruning, substantially reducing unnecessary cluster and vector accesses with negligible overhead. First, CLIP exploits the monotonicity of cosine-law-based lower bounds, enabling eliminating an undesirable cluster in O(1) time and filtering batches of irrelevant vectors in logarithmic time in the list size, with a tight analytical guarantee. Second, building on this, we develop two IVF variants: IVF-CLIP, which integrates CLIP into IVFFlat, and HIVF-CLIP, which extends it with a hierarchical structure for adaptive sub-cluster probing. Third, for dynamic workloads, we present LSM-IVF, an LSM-inspired design that supports fast updates by deferring index maintenance to background compaction, and enables efficient queries via CLIP-based optimizations that eliminate costly level-by-level searches. Extensive experiments show that CLIP variants achieve up to 78% pruning and 69% higher efficiency over static IVF baselines, while LSM-IVF improves throughput by up to 141% over dynamic IVF baselines with comparable update efficiency.

cs.DB↗
arXiv2026

TrainMover: An Interruption-Resilient Runtime for ML Training

Large-scale ML training jobs are frequently interrupted by hardware and software anomalies, failures, and management events. Existing solutions like checkpoint-restart or runtime reconfiguration suffer from long downtimes and degraded performance. We present TrainMover, a resilient LLM training runtime that leverages elastic and standby machines to handle interruptions with minimal downtime and zero memory overhead. To achieve these goals, TrainMover introduces three key techniques: two-phase, delta-based communication group setup; communication-free sandboxed warmup; and general standby design that enables failure recovery from any role. Our evaluation shows that TrainMover consistently achieves around 20 seconds of downtime when handling various interruptions at the 1024-GPU scale. TrainMover is projected to reduce wasted GPU hours by 55% compared to the best alternative, saving 1.4 million GPU-hours per week at the 64K-GPU scale.

cs.DC↗
arXiv2026

The complexity of Ford domains of $Γ_0(N)$

We investigate a particular choice of the Ford fundamental domain of the congruence subgroup $Γ_0(N)$ and define a notion of complexity $c(N)$ accordingly, which is a nonnegative integer and carries some information on the shape of the Ford domain. The property that $c(N)=0$ first appeared as a technical assumption in a paper by Pohl, which is closely related to a conjecture of Zagier on the "reduction theory" of $Γ_0(N)$. In this paper, we give a complete classification of positive integers $N$ with $c(N)=0$, and we also show that $c(N)$ goes to infinity if both the number of distinct prime factors of $N$ and the smallest prime factor of $N$ go to infinity.

math.NT↗
arXiv2026

EROICA: Online Performance Troubleshooting for Large-scale Model Training

Troubleshooting performance problems of large model training (LMT) is immensely challenging, due to unprecedented scales of modern GPU clusters, the complexity of software-hardware interactions, and the data intensity of the training process. Existing troubleshooting approaches designed for traditional distributed systems or datacenter networks fall short and can hardly apply to real-world training systems. In this paper, we present EROICA, the first online troubleshooting system that provides both fine-grained observation based on profiling, and coverage of all machines in GPU clusters, to diagnose performance issues in production, including both hardware and software problems (or the mixture of both). EROICA effectively summarizes runtime behavior patterns of LMT function executions via online profiling, and leverages differential observability to localize the root cause with minimal production impact. EROICA has been deployed as a production service for large-scale GPU clusters of ~100,000 GPUs for 1.5 years. It has diagnosed a variety of difficult performance issues with 97.5% success.

cs.DC↗
arXiv2026

Elliptic curves and Fourier coefficients of meromorphic modular forms

We discuss several congruences satisfied by the coefficients of meromorphic modular forms, or equivalently, the $p$-adic behaviors of meromorphic modular forms under the $U_p$ operator, that are summarized from numerical experiments. In the generic case, we observe the connection to symmetric powers of elliptic curves, while in the CM case, we furthermore observe the connection to the $p$-adic analogue of the Chowla--Selberg periods. Along with the discussions, we will provide some heuristic explanations for these congruences as well as prove some of them using hypergeometric functions and the Borcherds--Shimura lift.

math.NT↗
arXiv2026

Ordinary primes for $\mathrm{GL}_2$-type abelian varieties and weight $2$ modular forms

Let $A$ be a $g$-dimensional abelian variety defined over a number field $F$. It is conjectured that the set of ordinary primes of $A$ over $F$ has positive density, and this is known to be true when $g=1, 2$, or for certain abelian varieties with extra endomorphisms. In this paper, we extend the family of abelian varieties whose sets of ordinary primes have positive density. Specifically, we show that if the endomorphism algebra of $A$ contains a number field $K$ of degree $g$, then under certain conditions on the fields $F$ and $K$, the set of ordinary primes of $A$ over $F$ has positive density. This includes $\mathrm{GL}_2$-type abelian varieties over $\mathbb{Q}$ (resp. quadratic number fields) of dimension $q$ or $2q$ (resp. $q$) for any rational prime $q$. The proof is carried out in the general setting of compatible systems of Galois representations, and as a consequence, it also implies a positive density result for the sets of ordinary primes of certain modular forms of weight $2$.

math.NT↗
arXiv2025

Energy and Memory-Efficient Federated Learning With Ordered Layer Freezing

Federated Learning (FL) has emerged as a privacy-preserving paradigm for training machine learning models across distributed edge devices in the Internet of Things (IoT). By keeping data local and coordinating model training through a central server, FL effectively addresses privacy concerns and reduces communication overhead. However, the limited computational power, memory, and bandwidth of IoT edge devices pose significant challenges to the efficiency and scalability of FL, especially when training deep neural networks. Various FL frameworks have been proposed to reduce computation and communication overheads through dropout or layer freezing. However, these approaches often sacrifice accuracy or neglect memory constraints. To this end, in this work, we introduce Federated Learning with Ordered Layer Freezing (FedOLF). FedOLF consistently freezes layers in a predefined order before training, significantly mitigating computation and memory requirements. To further reduce communication and energy costs, we incorporate Tensor Operation Approximation (TOA), a lightweight alternative to conventional quantization that better preserves model accuracy. Experimental results demonstrate that over non-iid data, FedOLF achieves at least 0.3%, 6.4%, 5.81%, 4.4%, 6.27% and 1.29% higher accuracy than existing works respectively on EMNIST (with CNN), CIFAR-10 (with AlexNet), CIFAR-100 (with ResNet20 and ResNet44), and CINIC-10 (with ResNet20 and ResNet44), along with higher energy efficiency and lower memory footprint.

cs.LG↗
arXiv2025

On converse theorems for Hilbert modular forms assuming unramified twists

We prove two results on converse theorems for Hilbert modular forms over totally real fields of degree $r>1$. The first result recovers a Hilbert modular form (of some level) from an $L$-series satisfying functional equations twisted by all the unramified Hecke characters. The second result assumes both the 'unramified' functional equations and an Euler product, and recovers a Hilbert modular form of the expected level predicted by the shape of the functional equations. Our result generalizes the current converse theorems for $\mathrm{GL}_2$ in the case of Hilbert modular forms in that we completely remove the assumptions on ramified twists.

math.NT↗
arXiv2025

Proof-Carrying Fair Ordering: Asymmetric Verification for BFT via Incremental Graphs

Byzantine Fault-Tolerant (BFT) consensus protocols ensure agreement on transaction ordering despite malicious actors, but unconstrained ordering power enables sophisticated value extraction attacks like front running and sandwich attacks - a critical threat to blockchain systems. Order-fair consensus curbs adversarial value extraction by constraining how leaders may order transactions. While state-of-the-art protocols such as Themis attain strong guarantees through graph-based ordering, they ask every replica to re-run the leader's expensive ordering computation for validation - an inherently symmetric and redundant paradigm. We present AUTIG, a high-performance, pluggable order-fairness service that breaks this symmetry. Our key insight is that verifying a fair order does not require re-computing it. Instead, verification can be reduced to a stateless audit of succinct, verifiable assertions about the ordering graph's properties. AUTIG realizes this via an asymmetric architecture: the leader maintains a persistent Unconfirmed-Transaction Incremental Graph (UTIG) to amortize graph construction across rounds and emits a structured proof of fairness with each proposal; followers validate the proof without maintaining historical state. AUTIG introduces three critical innovations: (i) incremental graph maintenance driven by threshold-crossing events and state changes; (ii) a decoupled pipeline that overlaps leader-side collection/update/extraction with follower-side stateless verification; and (iii) a proof design covering all internal pairs in the finalized prefix plus a frontier completeness check to rule out hidden external dependencies. We implement AUTIG and evaluate it against symmetric graph-based baselines under partial synchrony. Experiments show higher throughput and lower end-to-end latency while preserving gamma-batch-order-fairness.

cs.DC↗
arXiv2025

BASFuzz: Towards Robustness Evaluation of LLM-based NLP Software via Automated Fuzz Testing

Fuzzing has shown great success in evaluating the robustness of intelligent natural language processing (NLP) software. As large language model (LLM)-based NLP software is widely deployed in critical industries, existing methods still face two main challenges: 1 testing methods are insufficiently coupled with the behavioral patterns of LLM-based NLP software; 2 fuzzing capability for the testing scenario of natural language generation (NLG) generally degrades. To address these issues, we propose BASFuzz, an efficient Fuzz testing method tailored for LLM-based NLP software. BASFuzz targets complete test inputs composed of prompts and examples, and uses a text consistency metric to guide mutations of the fuzzing loop, aligning with the behavioral patterns of LLM-based NLP software. A Beam-Annealing Search algorithm, which integrates beam search and simulated annealing, is employed to design an efficient fuzzing loop. In addition, information entropy-based adaptive adjustment and an elitism strategy further enhance fuzzing capability. We evaluate BASFuzz on six datasets in representative scenarios of NLG and natural language understanding (NLU). Experimental results demonstrate that BASFuzz achieves a testing effectiveness of 90.335% while reducing the average time overhead by 2,163.852 seconds compared to the current best baseline, enabling more effective robustness evaluation prior to software deployment.

cs.SE↗
arXiv2025

Efficient and Secure Sleepy Model for BFT Consensus

Byzantine Fault Tolerant (BFT) consensus protocols for dynamically available systems face a critical challenge: balancing latency and security in fluctuating node participation. Existing solutions often require multiple rounds of voting per decision, leading to high latency or limited resilience to adversarial behavior. This paper presents a BFT protocol integrating a pre-commit mechanism with publicly verifiable secret sharing (PVSS) into message transmission. By binding users' identities to their messages through PVSS, our approach reduces communication rounds. Compared to other state-of-the-art methods, our protocol typically requires only four network delays (4$Δ$) in common scenarios while being resilient to up to 1/2 adversarial participants. This integration enhances the efficiency and security of the protocol without compromising integrity. Theoretical analysis demonstrates the robustness of the protocol against Byzantine attacks. Experimental evaluations show that, compared to traditional BFT protocols, our protocol significantly prevents fork occurrences and improves chain stability. Furthermore, compared to longest-chain protocol, our protocol maintains stability and lower latency in scenarios with moderate participation fluctuations.

cs.DC↗
arXiv2025

TRIM: Accelerating High-Dimensional Vector Similarity Search with Enhanced Triangle-Inequality-Based Pruning

High-dimensional vector similarity search (HVSS) is critical for many data processing and AI applications. However, traditional HVSS methods often require extensive data access for distance calculations, leading to inefficiencies. Triangle-inequality-based lower bound pruning is a widely used technique to reduce the number of data access in low-dimensional spaces but becomes less effective in high-dimensional settings. This is attributed to the "distance concentration" phenomenon, where the lower bounds derived from the triangle inequality become too small to be useful. To address this, we propose TRIM, which enhances the effectiveness of traditional triangle-inequality-based pruning in high-dimensional vector similarity search using two key ways: (1) optimizing landmark vectors used to form the triangles, and (2) relaxing the lower bounds derived from the triangle inequality, with the relaxation degree adjustable according to user's needs. TRIM is a versatile operation that can be seamlessly integrated into both memory-based (e.g., HNSW, IVFPQ) and disk-based (e.g., DiskANN) HVSS methods, reducing distance calculations and disk access. Extensive experiments show that TRIM enhances memory-based methods, improving graph-based search by up to 90% and quantization-based search by up to 200%, while achieving a pruning ratio of up to 99%. It also reduces I/O costs by up to 58% and improves efficiency by 102% for disk-based methods, while preserving high query accuracy.

cs.DB↗
arXiv2025

Efficient Chambolle-Pock based algorithms for Convoltional sparse representation

Recently convolutional sparse representation (CSR), as a sparse representation technique, has attracted increasing attention in the field of image processing, due to its good characteristic of translate-invariance. The content of CSR usually consists of convolutional sparse coding (CSC) and convolutional dictionary learning (CDL), and many studies focus on how to solve the corresponding optimization problems. At present, the most efficient optimization scheme for CSC is based on the alternating direction method of multipliers (ADMM). However, the ADMM-based approach involves a penalty parameter that needs to be carefully selected, and improper parameter selection may result in either no convergence or very slow convergence. In this paper, a novel fast and efficient method using Chambolle-Pock(CP) framework is proposed, which does not require extra manual selection parameters in solving processing, and has faster convergence speed. Furthermore, we propose an anisotropic total variation penalty of the coefficient maps for CSC and apply the CP algorithm to solve it. In addition, we also apply the CP framework to solve the corresponding CDL problem. Experiments show that for noise-free image the proposed CSC algorithms can achieve rival results of the latest ADMM-based approach, while outperforms in removing noise from Gaussian noise pollution image.

cs.CV↗
arXiv2025

UNIFY: Unified Index for Range Filtered Approximate Nearest Neighbors Search

This paper presents an efficient and scalable framework for Range Filtered Approximate Nearest Neighbors Search (RF-ANNS) over high-dimensional vectors associated with attribute values. Given a query vector $q$ and a range $[l, h]$, RF-ANNS aims to find the approximate $k$ nearest neighbors of $q$ among data whose attribute values fall within $[l, h]$. Existing methods including pre-, post-, and hybrid filtering strategies that perform attribute range filtering before, after, or during the ANNS process, all suffer from significant performance degradation when query ranges shift. Though building dedicated indexes for each strategy and selecting the best one based on the query range can address this problem, it leads to index consistency and maintenance issues. Our framework, called UNIFY, constructs a unified Proximity Graph-based (PG-based) index that seamlessly supports all three strategies. In UNIFY, we introduce SIG, a novel Segmented Inclusive Graph, which segments the dataset by attribute values. It ensures the PG of objects from any segment combinations is a sub-graph of SIG, thereby enabling efficient hybrid filtering by reconstructing and searching a PG from relevant segments. Moreover, we present Hierarchical Segmented Inclusive Graph (HSIG), a variant of SIG which incorporates a hierarchical structure inspired by HNSW to achieve logarithmic hybrid filtering complexity. We also implement pre- and post-filtering for HSIG by fusing skip list connections and compressed HNSW edges into the hierarchical graph. Experimental results show that UNIFY delivers state-of-the-art RF-ANNS performance across small, mid, and large query ranges.

cs.DS↗
arXiv2025

Enhancing Large-Scale AI Training Efficiency: The C4 Solution for Real-Time Anomaly Detection and Communication Optimization

The emergence of Large Language Models (LLMs) has necessitated the adoption of distributed training techniques, involving the deployment of thousands of GPUs to train a single model. Unfortunately, the efficiency of large-scale distributed training systems is often suboptimal due to the increased likelihood of hardware errors in high-end GPU products and the heightened risk of network traffic collisions. Moreover, any local hardware failure can disrupt training tasks, and the inability to swiftly identify faulty components leads to a significant waste of GPU resources. And, prolonged communication due to traffic collisions can substantially increase GPU waiting times. To address these challenges, we propose a communication-driven solution, namely the C4. The key insights of C4 are twofold. First, the load in distributed training exhibits homogeneous characteristics and is divided into iterations through periodic synchronization, therefore hardware anomalies would incur certain syndrome in collective communication. By leveraging this feature, C4 can rapidly identify the faulty components, swiftly isolate the anomaly, and restart the task, thereby avoiding resource wastage caused by delays in anomaly detection. Second, the predictable communication model of collective communication, involving a limited number of long-lived flows, allows C4 to efficiently execute traffic planning, substantially reducing bandwidth competition among these flows. The C4 has been extensively deployed across real-world production systems in a hyperscale cloud provider, yielding a significant improvement in system efficiency, from 30% to 45%. This enhancement is attributed to a 30% reduction in error-induced overhead and a 15% reduction in communication costs.

cs.DC↗