Searcharxiv⌕ Search

arXiv subjects

Yongkun Li

Publications and source records attributed to Yongkun Li.

At least 19 recordsLinked to original sources

Besicovitch-like Pseudo Almost Automorphic Stochastic Processes: Theory and Applications to Semilinear SDEs

This paper introduces, for the first time, the concept of Besicovitch-like pseudo almost automorphic stochastic processes. As a meaningful application, we investigate a class of semilinear stochastic equations driven by Brownian motion in a separable Hilbert space. By combining the Banach fixed point theorem with refined inequality techniques, we establish sufficient conditions guaranteeing the existence, uniqueness, and global exponential stability of Besicovitch-like pseudo almost automorphic solutions in distribution for the considered equations. An example is provided to show the validity of the obtained results.

math.PR↗

Visual Cache Reuse for Efficient Multimodal Inference

Vision-language model (VLM) inference incurs substantial prefill latency and memory overhead due to the long visual token sequences of large visual inputs. Existing serving systems typically exploit exact cache reuse through prefix caching or position-independent caching, overlooking the potential of approximate caching in real-world workloads. However, directly reusing an approximate cache entry may introduce spurious visual information, which naive approaches that select a subset of important visual tokens for recomputation fail to adequately mitigate this risk. We present VisMemo, an efficient serving system that enables approximate KV cache reuse for modern VLMs. Its core is a system-algorithm co-design that combines an entropy-aware KV cache reuse algorithm with differentiated cache management. Specifically, VisMemo uses patch entropy to identify reuse candidates and warmup attention to prioritize recomputation based on the current query. To support the large cache capacity across GPU memory, CPU memory, and SSD storage, VisMemo applies distinct I/O and cache policies to exact and approximate hits: sparse patch-level I/O for approximate hits, prioritized retention for exact hits, and a reserved CPU victim cache for preset entries selected using offline workload priors and online reuse observations. VisMemo further optimizes the memory access path between the inference engine and the storage backend, eliminating redundant caching and unnecessary data movement. Experiments on multiple public datasets show that VisMemo preserves output quality under approximate reuse. On a real-world production trace, VisMemo increases the overall cache hit rate from 27.78% to 61.45% and reduces prefill latency by up to 48.3% at 8K input tokens.

cs.OS↗

gKSM: GPU Memory Deduplication at Driver Layer for Serverless LLM Inference

In serverless LLM inference, cloud providers commonly locate multiple instances on a single GPU to improve utilization. However, each instance independently maintains its model weights and runtime state, making duplicate GPU memory the primary bottleneck. Memory deduplication is a widely-used technique for eliminating duplicate data in CPU systems, but extending it to GPU memory faces fundamental challenges. Effective GPU memory deduplication must avoid redundant data movement between CPU and GPU while preserving write isolation for shared pages without native copy-on-write (COW) support. This paper presents gKSM, a transparent GPU memory deduplication framework for serverless LLM inference. gKSM addresses these challenges through CPU-GPU collaborative hash management that offloads hash computation to GPU kernels, zero-copy model weight deduplication that eliminates duplication before model weight transfer, and a cross-privilege page fault handler that combines a lightweight kernel hook and CUDA Virtual Memory Management (VMM) remapping in the user space. gKSM also designs a stable page scanning algorithm and a reuse-aware placement policy in cluster scale. Experiments show that gKSM reduces GPU memory usage by 59.8% while maintaining high throughput compared to ServerlessLLM. Under memory constrained workloads, gKSM improves aggregate throughput by 2.15X and supports five instances concurrently. In cluster level simulation, gKSM achieves a service rate of 99.97%, compared to 78.2% for SLInfer and 26.9% for ServerlessLLM.

cs.OS↗

Energy-Efficient LLM Serving via Disaggregated Attention--FFN and Flexible Frequency Scaling

Large language model (LLM) serving spans diverse applications with stringent service-level objectives (SLOs), often requiring GPUs to run at maximum frequencies and increasing energy consumption. Existing energy-management approaches adapt GPU frequencies only at the request or inference-phase level, overlooking operator-level differences in frequency sensitivity between Attention and feed-forward networks (FFNs). We find that the energy-optimal frequencies of Attention and FFN (A/F) differ and vary with the inference phase, workload, and system configurations. However, runtime variability and independent A/F frequency control create a large search space and high communication overhead. To address these challenges, we present AFlex, a framework that jointly optimizes resource provisioning and GPU frequency scaling for disaggregated A/F serving. AFlex introduces a global scheduler and a local operator-level dynamic voltage and frequency scaling (DVFS) controller to determine A/F resource allocations and frequencies. It further introduces an interleaved A/F pipeline with dynamic microbatch depth and adaptive request batching to reduce pipeline bubbles. We implement AFlex in SGLang and evaluate it on NVIDIA A800 GPUs using Qwen3-32B and Mixtral-8$\times$7B under production Conversation and Coding traces. \AFlex reduces energy per token by up to 49\% over state-of-the-art disaggregated serving and 48\% over frequency-scaling systems while satisfying TTFT and TPOT SLOs.

cs.DC↗

LUMEN: Coordinated Failure Recovery for Distributed LLM Serving

Modern large language model (LLM) serving clusters distribute inference requests across multiple worker processes on different GPUs, but failures are prevalent at scale. When a worker fails, the cluster simultaneously loses the failed worker's GPU-resident key-value (KV) caches and serving capacity, leaving surviving workers to absorb the redirected traffic while re-running interrupted requests from scratch. Existing fault-tolerant systems either restart interrupted requests from scratch or restore KV caches from checkpoints stored on a fixed neighboring worker, but both approaches route recovery work without considering current cluster load and leave the recovering worker idle during model reload. We present LUMEN, a fault-tolerant LLM serving system that treats recovery as a load-aware coordination problem across three decision points: checkpoint placement before failures, interrupted-request distribution at failure time, and serving capacity restoration during model reload. We evaluate LUMEN using both prototype experiments and large-scale simulations and demonstrate significant improvements in serving and recovery times.

cs.DC↗

Leveraging I/O Stalls for Efficient Scheduling in ANNS

Disk-based graph indexes for approximate nearest neighbor search (ANNS) must serve latency-sensitive queries and throughput-demanding updates concurrently. We observe that over 40% of search-thread CPU time is spent stalling on disk I/O; such idle cycles are invisible to thread-level scheduling yet available for other work. We present LIOS(Leverage I/O Stall), a framework that executes index updates inside search-side I/O stall windows. LIOS introduces three techniques: (i) splitting each update into resumable subtasks small enough to fit within a single stall window; (ii) bounding the expected overrun of update subtasks to a given threshold; and (iii) dynamically adjusting the fraction of idle time devoted to updates to drive end-to-end search latency degradation toward a user-specified target. We integrate LIOS into two update-optimized ANNS systems, FreshDiskANN and OdinANN. LIOS achieves speedups of up to 2.68$\times$ in insertion and 2.18$\times$ in deletion, with search latency degradation maintained near the user-specified target.

cs.DB↗

Varuna: Enabling Failure-Type Aware RDMA Failover

RDMA link failures can render connections temporarily unavailable, causing both performance degradation and significant recovery overhead. To tolerate such failures, production datacenters assign each primary link with a standby link and, upon failure, uniformly retransmit all in-flight RDMA request over the backup path. However, we observe that such blanket retransmission is unnecessary. In-flight requests can be split into pre-failure and post-failure categories depending on whether the responder has already executed. Retransmitting post-failure requests is not only redundant (consuming bandwidth), but also incorrect for non-idempotent operations, where duplicate execution can violate application semantics. We present Varuna, a failure-type-aware RDMA recovery mechanism that enables correct retransmission and us-level failover. Varuna piggybacks a lightweight completion log on every RDMA operation; after a link failure, this log deterministically reveals which in-flight requests were executed (post-failure) and which were lost (pre-failure). Varuna then retransmits only the pre-failure subset and fetches/recovers the return values for post-failure requests. Evaluated using synthetic microbenchmarks and end-to-end RDMA TPC-C transactions, Varuna incurs only 0.6-10% steady-state latency overhead in realistic applications, eliminates 65% of recovery retransmission time, preserves transactional consistency, and introduces zero connectivity rebuild overhead and negligible memory overhead during RDMA failover.

cs.DC↗

Weyl double-measure pseudo almost automorphic functions and Weyl double-measure pseudo almost automorphic solutions to a semilinear abstract differential equations

This paper first propose a concept of Weyl double-measure pseudo-almost automorphic functions and examines their fundamental characteristics. Subsequently, employing fixed point theorems, we systematically investigate the existence and uniqueness of both Weyl almost automorphic solutions and Weyl double-measure pseudo-almost automorphic solutions for a class of semilinear abstract differential equations. Finally, through the application of inequality-based analytical methods, we establish the global exponential stability of these solutions.

math.CA↗

FOCUS: Boosting Schema-aware Access for KV Stores via Hierarchical Data Management

Persistent key-value (KV) stores are critical infrastructure for data-intensive applications. Leveraging high-performance Non-Volatile Memory (NVM) to enhance KV stores has gained traction. However, previous work has primarily focused on optimizing KV stores themselves, without adequately addressing their integration into applications. Consequently, existing applications, represented by NewSQL databases, still resort to a flat mapping approach, which simply maps structured records into flat KV pairs to use KV stores. Such semantic mismatch may cause significant I/O amplification and I/O splitting under production workloads, harming the performance. To this end, we propose FOCUS, a log-structured KV store optimized for fine-grained hierarchical data organization and schema-aware access. FOCUS introduces a hierarchical KV model to provide native support for upper-layer structured data. We implemented FOCUS from scratch. Experiments show that FOCUS can increase throughput by 2.1-5.9x compared to mainstream NVM-backed KV stores under YCSB SQL workloads.

cs.DB↗

Probabilistic Crop Yields Forecasts With Spatio-Temporal Conditional Copula Using Extreme Weather Covariates

We introduce a novel forecasting model for crop yields that explicitly accounts for spatio-temporal dependence and the influence of extreme weather and climatic events. Our approach combines Bayesian Structural Time Series for modeling marginal crop yields, ensuring a more robust quantification of uncertainty given the typically short historical records. To capture dynamic dependencies between regions, we develop a time-varying conditional copula model, where the copula parameter evolves over time as a function of its previous lag and extreme weather covariates. Unlike traditional approaches that treat climatic factors as fixed inputs, we incorporate dynamic Generalized Extreme Value models to characterize extreme weather events, enabling a more accurate reflection of their impact on crop yields. Furthermore, to ensure scalability for large-scale applications, we build on the existing Partitioning Around Medoids clustering algorithm and introduce a novel dissimilarity measure that integrates both spatial and copula-based dependence, enabling an effective reduction of the dimensionality in the dependence structure.

stat.ME↗

Left fractional Sobolev space via Riemann$-$Liouville derivatives on time scales and its application to a fractional boundary value problem on time scales

We first prove the equivalence of two definitions of Riemann-Liouville fractional integral on time scales, then by the concept of fractional derivative of Riemann-Liouville on time scales, we introduce fractional Sobolev spaces, characterize them, define weak fractional derivatives, and show that they coincide with the Riemann-Liouville ones on time scales. Next, we prove equivalence of some norms in the introduced spaces and derive their completeness, reflexivity, separability and some imbeddings. Finally, as an application, by constructing an appropriate variational setting, using the mountain pass theorem and the genus properties, the existence of weak solutions for a class of Kirchhoff-type fractional p-Laplacian systems on time scales with boundary condition is studied, and three results of the existence of weak solutions for this problem is obtained.

math.CA↗

Right fractional Sobolev space via Riemann$-$Liouville derivatives on time scales and an application to fractional boundary value problem on time scales

Using the concept of fractional derivatives of Riemann$-$Liouville on time scales, we first introduce right fractional Sobolev spaces and characterize them. Then, we prove the equivalence of some norms in the introduced spaces, and obtain their completeness, reflexivity, separability and some imbeddings. Finally, as an application, we propose a recent method to study the existence of weak solutions of fractional boundary value problems on time scales by using variational method and critical point theory, and by constructing an appropriate variational setting, we obtain two existence results of the problem.

math.FA↗

Besicovitch almost periodic solutions of abstract semi-linear differential equations with delay

In this paper, we first use the Bohr property to give a definition of Besicovitch almost periodic functions, and study some basic properties of Besicovitch almost periodic functions, including the equivalence of the Bohr property and the Bochner property. Then, as an application, we use the contraction principal to obtain the existence and uniqueness of Besicovitch almost periodic solutions for a class of abstract semi-linear differential equations with delay.

math.FA↗

Deterministic Data Distribution for Efficient Recovery in Erasure-Coded Storage Systems

Due to individual unreliable commodity components, failures are common in large-scale distributed storage systems. Erasure codes are widely deployed in practical storage systems to provide fault tolerance with low storage overhead. However, random data distribution (RDD), commonly used in erasure-coded storage systems, induces heavy cross-rack traffic, load imbalance, and random access, which adversely affects failure recovery. In this paper, with orthogonal arrays, we define a Deterministic Data Distribution ($D^3$) to uniformly distribute data/parity blocks among nodes, and propose an efficient failure recovery approach based on $D^3$, which minimizes the cross-rack repair traffic against a single node failure. Thanks to the uniformity of $D^3$, the proposed recovery approach balances the repair traffic not only among nodes within a rack but also among racks. We implement $D^3$ over Reed-Solomon codes and Locally Repairable Codes in Hadoop Distributed File System (HDFS) with a cluster of 28 machines. Compared with RDD, our experiments show that $D^3$ significantly speeds up the failure recovery up to 2.49 times for RS codes and 1.38 times for LRCs. Moreover, $D^3$ supports front-end applications better than RDD in both of normal and recovery states.

cs.DC↗

Enabling Efficient Updates in KV Storage via Hashing: Design and Performance Evaluation

Persistent key-value (KV) stores mostly build on the Log-Structured Merge (LSM) tree for high write performance, yet the LSM-tree suffers from the inherently high I/O amplification. KV separation mitigates I/O amplification by storing only keys in the LSM-tree and values in separate storage. However, the current KV separation design remains inefficient under update-intensive workloads due to its high garbage collection (GC) overhead in value storage. We propose HashKV, which aims for high update performance atop KV separation under update-intensive workloads. HashKV uses hash-based data grouping, which deterministically maps values to storage space so as to make both updates and GC efficient. We further relax the restriction of such deterministic mappings via simple but useful design extensions. We extensively evaluate various design aspects of HashKV. We show that HashKV achieves 4.6x update throughput and 53.4% less write traffic compared to the current KV separation design. In addition, we demonstrate that we can integrate the design of HashKV with state-of-the-art KV stores and improve their respective performance.

cs.DB↗

Almost periodic functions on the quantum time scale and applications

In this paper, we first propose two types of concepts of almost periodic functions on the quantum time scale. Secondly, we study some basic properties of almost periodic functions on the quantum time scale. Thirdly, based on these, we study the existence and uniqueness of almost periodic solutions of dynamic equations on the quantum time scale by Lyapunov method. Then, we give a equivalent definition of almost periodic functions on the quantum time scale. Finally, as an application, we propose a class of high-order Hopfield neural networks on the quantum time scale and establish the existence of almost periodic solutions of this class of neural networks.

math.CA↗

Almost automorphic functions on the quantum time scale and applications

In this paper, we first propose two types of concepts of almost automorphic functions on the quantum time scale. Secondly, we study some basic properties of almost automorphic functions on the quantum time scale. Then, we introduce a transformation between functions defined on the quantum time scale and functions defined on the set of generalized integer numbers, by using this transformation we give equivalent definitions of almost automorphic functions on the quantum time scale. Finally, as an application of our results, we establish the existence of almost automorphic solutions of linear and semilinear dynamic equations on the quantum time scale.

math.CA↗

The existence and global exponential stability of almost periodic solutions for neutral type CNNs on time scales

In this paper, a class of neutral type competitive neural networks with mixed time-varying delays and leakage delays on time scales is proposed. Based on the exponential dichotomy of linear dynamic equations on time scales, Banach's fixed point theorem and the theory of calculus on time scales, some sufficient conditions that are independent of the backwards graininess function of the time scale are obtained for the existence and global exponential stability of almost periodic solutions for this class of neural networks. The obtained results are completely new and indicate that both the continuous time and the discrete time cases of the networks share the same dynamical behavior. Finally, an examples is given to show the effectiveness of the obtained results.

math.CA↗