Papers

We show that one-way one-round quantum-LOCAL algorithms cannot $4$-color directed cycles with high probability. This is the first lower bound in the high-probability quantum LOCAL setting that goes beyond the non-signaling and bounded-dependence models, exploiting the structure of distributed quantum algorithms. Our proof establishes a bidirectional connection between distributed quantum computing and extremal combinatorics. We obtain our lower bound by proving a Mantel-type stability theorem for weighted matrix spaces.

Industrial fraud detection often relies on costly expert-crafted features that overlook graph-structured relational signals, while GNNs often do not meet the interpretability and deployment requirements of financial risk control. We propose GraphFAS (Graph Feature Automated Selection), a distributed feature selection procedure based on Boruta that bridges this gap through: (1) a non-parametric graph feature generation module that constructs explicit, interpretable structural features via multi-hop subgraph extraction and multi-scale aggregation without learned parameters; and (2) an automated distributed feature selection algorithm extending Boruta with median-based aggregation across partitions to robustly identify informative features at scale with minimal domain expertise. Compared with end-to-end GNN pipelines, GraphFAS decouples feature aggregation from model training, enabling direct integration with tabular models and direct compatibility with TreeSHAPbased explanations. Deployed in Alipay, GraphFAS delivers orderof-magnitude improvements in engineering efficiency while showing strong performance against expert-driven and graph-learning baselines on large-scale graphs.

Cloud-edge controllers coordinate service placement, replica scaling, and resource pre-warming to keep end-to-end latency within application deadlines. But evaluations often obscure the source of a reported gain: placement and scaling are studied separately; workload, connectivity, and calibration assumptions remain implicit; and metrics over completed tasks hide unfinished work. We present ContinuumBench, a benchmark that controls these factors. Its completion-aware accounting treats late, unfinished, and discarded tasks as deadline misses. A common protocol compares placement-only and scale-capable controllers under declared regimes and stressors. Built on the ECLYPSE simulator, ContinuumBench adds arrivals, worker elasticity, intermittent transport, buffering, and failures to close the control loop. We evaluate nine controllers across four scenarios and two regimes. The studied regimes are capacity-bound: elastic capacity, not placement sophistication, drives completion, and once capacity suffices, the choice of autoscaling policy decides how much of that work arrives on time. Placement re-planning has no measurable effect without relocation, while cost-free migration defines the observed exception. Consequently, scale-capable controllers approach an over-provisioned reference while placement-only controllers degrade with load; and placement quality separates controllers only once capacity is exhausted. Finally, the accounting choice itself changes the reported result: completion-only and completion-aware scoring can rank controllers differently.

Autonomous, cross-facility science requires capabilities that no individual project should have to build for itself: managed execution for long-lived services, versioned distribution of models to remote compute systems, governed access to large language models, a shared substrate for experimental data, and end-to-end provenance. The U.S. Department of Energy Genesis Mission platform, delivered through the American Science Cloud, provides these as reusable services. This paper reports how the Genesis platform enables cross-facility experiments and accelerates scientific discovery. We explore the plant phenotyping workflow of the Orchestrated Platform for Autonomous Laboratories as the exemplar: it couples Oak Ridge National Laboratory's Advanced Plant Phenotyping Laboratory with the Frontier supercomputer. In a 40-day nickel-treatment campaign, the resulting workflow replaced roughly twelve hours of manual analysis with interactive queries returning in seconds to minutes.

Distributed HPC and LLM workloads increasingly require efficient communication for scalability, yet growing data movement has become a major performance bottleneck. Communication compression can reduce this overhead and complement execution-level optimizations, but its benefits remain difficult to assess because existing benchmarks lack support for diverse backends, realistic datasets, application-specific accuracy metrics, and overlap-induced resource contention. We present CC-Bench, a lightweight, extensible, and application-oriented benchmark suite for evaluating communication compression under realistic execution conditions. CC-Bench uses declarative application-environment modeling to decouple profiling logic from communication libraries, datasets, and fidelity metrics, enabling portable cross-library evaluation. It further combines function-level interception and hardware counter monitoring to characterize per-phase latency, hardware utilization, numerical fidelity, and computation interference. With representative datasets from HPC and LLM workloads, CC-Bench evaluates three compression-enabled communication libraries on CPU and GPU clusters, revealing accuracy-performance trade-offs and bottlenecks to guide practical deployment and optimization.

The transition from traditional High Performance Computing (HPC) to the Computing Continuum emphasizes efficient resource management and sustainable practices across Multi-Scale hybrid architectures. This paper introduces a multidimensional metric framework to characterize these systems and guide deployment strategies for modern workloads. The framework combines Architectural Performance metrics (such as Throughput, Latency, Scalability), System Utilization, and key Sustainability and Accuracy indicators (such as Energy Efficiency and Power Consumption). Using a modular hybrid testbed, experiments reveal complex relationships among metrics, especially the trade-offs between accuracy and energy, and the efficiency of hybrid nodes. The guidelines help identify optimal operating points and lay the groundwork for improving orchestrators and schedulers (e.g., Kubernetes) to assign demanding applications, including AI and Quantum Computing, to suitable system modules, ensuring high performance and sustainability.

Sparse attention bounds downstream attention work by retaining a fixed-size subset of indexed tokens, but its standalone exact Top-$K$ stage must still process materialized score rows whose length grows with context. Production radix selectors discover their first actionable boundary only after a complete-row pass, forcing another row-scale traversal before exact refinement. We observe that locating a compact upper tail requires substantially less resolution than identifying the exact rank boundary, and that fixed-stride partial views of the current row remain calibrated to the corresponding complete-row rank across ragged lengths. We present HPC-Ops Top-K, a sample-guided exact selector for ragged sparse-attention score rows. A fixed-stride view proposes a row-local coarse boundary; the mandatory complete-row pass certifies its sufficiency, forms the admitted candidate set, and initializes exact FP32 refinement over the unresolved frontier. A nested secondary boundary and exact recovery handle underfilled proposals before any output is committed, so sampling controls common-path work but never correctness. The GPU implementation fuses complete-row certification and candidate formation, and combines persistent, KV-split, and direct-exact execution behind graph-capturable ragged-row dispatch. We evaluate HPC-Ops Top-K on indexer scores from Hy4-Preview. It outperforms the fastest verified external exact baseline by $1.29$–$1.75\times$ across 20 operator configurations, with a $1.55\times$ geometric-mean speedup. It further achieves $1.36\times$ and $1.48\times$ speedups on two framework-derived sparse-attention traces. The implementation is available in HPC-Ops, Tencent's open-source high-performance operator library for LLM inference, at https://github.com/Tencent/hpc-ops.

Large language models (LLMs) are increasingly used as backends for intelligent web services, but serving them across the edge continuum requires balancing quality, latency, model footprint, and energy. This paper presents a controlled measurement study of self-hosted LLM inference across edge and near-edge deployment nodes: an NVIDIA Jetson AGX Orin and a near-edge server with CPU-only and GPU-enabled inference modes. We evaluate multiple open-weight LLMs and quantization variants using a fixed question-answering workload, and compare them against GPT-4o as a cloud-hosted accuracy and latency reference. Our benchmarking pipeline reports accuracy, model footprint, per-token decoding latency, prefill latency, and overall execution energy. The results show that GPU-enabled server execution provides the lowest compute-side latency, while Jetson Orin shows lower measured energy, consistent with its lower platform power under our setup. CPU-only execution is consistently dominated in latency for our workload and shows higher measured energy. We also show that parameter count and downloaded weight-file size alone do not reliably predict observed accuracy or latency. Finally, using Pareto-frontier analysis, we study how deployment decisions may change under possible streamed-token delivery overheads, highlighting that compute-side inference metrics alone can lead to suboptimal placement for latency-sensitive interactive web services.

We study a resource-efficient approach for implementing logical fanout operations in fault-tolerant distributed quantum computing using transversal operations on quantum error-correcting code blocks. Logical fanout, comprising multiple controlled-NOT operations from a common control qubit to target qubits located at remote nodes, is an important primitive for distributed quantum computation but can require substantial non-local communication when implemented directly between encoded blocks. We exploit the structure of encoded blocks and the availability of transversal logical operations to construct distributed fanout circuits that reduce the required non-local operations while preserving the logical action of the fanout operation. The construction is developed for encoded quantum information and illustrated using Bivariate Bicycle (BB)-code blocks. We analyze the resulting physical gate, entanglement, circuit-depth, and ancilla requirements. The approach provides a systematic method for implementing large logical fanout operations across distributed error-corrected quantum processors. Also, we study a distributed implementation of the global gate GCZ involving logical qubits (encoded using BB-code blocks), exploiting the concurrency in transversal distributed fanouts.

Distributed-tensor systems specify where values reside, while collective systems optimize how requested operations execute. At a boundary between vendor runtimes that cannot share a native communicator, neither abstraction states what a remote consumer must observe. SemBridge fills this gap by compiling graph and runtime facts into a typed contract for the consumer-visible result and its delivery obligations. The contract captures provenance, substitutability, completion, authority, demand, and native-domain locality. A deterministic lowerer constructs backend-neutral communication plans, and a symbolic checker validates each plan before execution across CUDA/NCCL and CANN/HCCL. An independent layout-only planner handles all 72 structural transitions but establishes only 54 complete obligations; a byte-only minimizer proposes 40 semantically invalid candidates, all rejected by SemBridge. On nine real edges, SemBridge produces distinct observation-aware plans that reduce startups on all nine and payload bytes on the three result edges. A live CUDA/CANN run derives and executes full-logit reconstruction, source projection, and owner-token delivery from log-probability, token-only, and owner-scoped requests. On a measured two-host 1-GbE capacity-spillover deployment, source projection cuts result traffic by more than 99.97% and increases throughput by 8.92-80.20% across Dense, MoE, and MiniMax workloads. All 18 MiniMax restart pairs at concurrency 1, 8, and 16 favor source projection. A Qwen3-14B MLP slice additionally verifies bitwise activation-shard delivery and HCCL completion of row-parallel partials. These results establish consumer observation as a semantic layer between placement and collective execution.

Change-data capture (CDC) feeds downstream systems like caches, search indexes, and data warehouses from a database's log of committed row changes. When bootstrapping, adding a table, or repairing downstream data, a pipeline must also copy existing rows. Merging this copy with the active log introduces the copy-to-log handoff problem. Changes must not fall through a gap, and older copied state must not overwrite a newer logged update or resurrect a deleted row. DBLog, developed at Netflix, addressed this problem by reading tables in chunks and interleaving those reads with the live log. Watermarks identify the changes that overlap each read, and the log wins when a copied row is stale. Debezium and Flink CDC have since adapted this design. Earlier work proved that applying the original algorithm's copied rows and logged changes in their emitted order reconstructs the source's rows, including the effect of every logged insert, update, and delete processed. Generalized DBLog asks when the same result holds for variants of that design. We state the conditions the source and capture implementation must satisfy. Once copying and reconciliation are complete, we prove that the result holds across all selected tables and key ranges even when their rows were read at different times. A single database snapshot is not required for the copy. Further logged changes advance the reconstructed state one event at a time. We establish these guarantees for classic watermarking, Debezium's signal-table and read-only modes, Flink CDC's parallel chunks, reads and dumps tied to exact log positions, and engine-consistent backups whose log position lies within known bounds. The complete theory is machine-checked in Isabelle/HOL, its core independently verified in Lean 4, and the protocols are also examined by bounded model checking in TLA+.

The placement of microservices in the compute continuum plays a vital role in delivering services that comply with customers' needs, such as reduced latency, storage requirements, quality of service and availability. To achieve customers' needs in the geographically dispersed architecture of the compute continuum, Service Level Objectives (SLOs) have been largely used in decision-making to place microservices. However, because low-level SLOs increase the barrier to entry for continuum users, placement decisions based on high-level business vocabulary are required if the compute continuum is to be adopted at scale. This paper proposes an architecture for microservices placement decisions in the computing continuum utilizing high-level user intents described in natural language as input. The approach utilizes Generative Artificial Intelligence to translate the intents to low-level SLOs, which are used along with the infrastructure description to decide where different microservices that compose an application must be deployed so that SLOs are met. We implement and evaluate a prototype of the architecture to demonstrate the approach's feasibility.

We propose PKDB, the first interactive debugger for GPU and multithreaded low-level kernels written in Python. Python is widely used in high performance computing (HPC), with frameworks such as PyKokkos translating Python-embedded domain-specific languages to native code that runs across OpenMP-threaded CPUs and various GPUs. Yet interactive debugging support for such code is absent: developers resort to print statements, framework-specific assertions, or CPU-only execution, the last of which requires altering the program or its data and can mask device-specific bugs. PKDB enables standard interactive debugging like breakpoints, stepping, and variable inspection while preserving actual on-device execution without source modification. Beyond these fundamentals, PKDB introduces two advanced capabilities that exploit the dynamic nature of Python and PyKokkos: (i) Live code evaluation, which lets developers execute arbitrary Python expressions or entire kernels in the middle of a paused kernel without restarting the process; (ii) Kernel call site substitution, which allows an actively running kernel to be updated and reloaded on the fly, so only the kernel is recompiled and re-executed without restarting the application. Our performance evaluation on Intel, AMD, and NVIDIA CPUs, and NVIDIA and AMD GPUs shows that PKDB introduces limited overhead and is practical for everyday use while introducing critical debugging features to the Python HPC ecosystem.

Distributed applications (dApps) bring AI to the microsecond-to-millisecond band beside the 5G distributed unit (DU), but every public dApp framework realizes them the same way: an external process that receives an indication and returns a control message. That boundary is right for sensing and advisory workloads. It cannot express a neural receiver that must finish inside a slot, and it cannot hold a scheduling decision the MAC is waiting on. This paper asks what a dApp interface must deliver, in latency and bandwidth, for the AI-RAN use cases now filed under that name to be realized as dApps at all. It treats the coupling between a dApp and the DU as a design axis with two legitimate ends, a C ABI inside the DU process and a protocol-standard E3AP association over SCTP, and places the three OCUDU dApp classes on it. An audited corpus of 39 runtime AI-RAN use cases, sized by 5G NR timing, shows that more than half cannot cross the observer boundary: inline PHY work because an indication has no return path into the same slot, and bounded control because of the tail under load. Measuring the mechanisms each framework actually uses, on a quiet host and with a live cell on the air, shows that every carrier meets a 100 us control deadline on an idle host and that only the in-process paths still do once the DU is running. The asynchronous use cases remain feasible at every position, so the protocol-standard end is kept as a first-class option; what in-process placement adds for them is accountable staleness and an observation-to-decision path with no message on it. These measurements derive the released ABI, a stream for inline work, a validated call for bounded control, and a choice of lease, supervised ring, or portable E3 client for observation, and four dApps of all three classes are validated together on one over-the-air cell.

Kolmogorov-Arnold Networks (KANs) replace the fixed activation functions and linear weights of Multi-Layer Perceptrons (MLPs) with learnable univariate functions on network edges, offering improved interpretability and, in some settings, competitive parameter efficiency. While the approximation properties of KANs have received considerable attention, their behavior under distributed, multi-GPU training has not been systematically characterized. This paper presents an empirical scalability study of data-parallel KAN training on multi-node, multi-GPU high-performance computing (HPC) infrastructure, evaluated along four dimensions: strong scaling, weak scaling, communication overhead, and model-size scaling. Experiments were conducted on the FinisTerrae III supercomputer using up to 8 NVIDIA A100 GPUs across 4 nodes with PyTorch Distributed Data Parallel (DDP). KAN training reaches 74.7% parallel efficiency at 8 GPUs with a 5.97x speedup, consistent with conventional deep learning workloads. Weak scaling shows an initial single-to-multi-GPU throughput drop followed by strong stability. Communication overhead follows a non-monotonic pattern (1.3%-6.1%), driven primarily by All-Reduce algorithm selection and inter-node latency rather than KAN's edge-wise gradient structure. The parameter-to-memory ratio improves with model size even as training time scales unfavorably. These results indicate that operator-level and data-parallel optimizations for KAN are complementary. We provide deployment guidelines for GPU topology and model-size selection, and discuss the limitations of a synthetic-regression evaluation.

Providers of continuously produced, commercially valuable data – sensor streams, telemetry, and other feeds sold as machine-learning training material – cannot observe whether delivered data is actually used, while data that keeps flowing to inactive consumers enlarges the leakage surface without producing value. We present Attestream, a blockchain-based architecture for intermittently delivered dataset streams that couples continued delivery to verifiable usage reporting. Every lifecycle event – dataset preparation, dual-signed delivery, derivative creation (e.g., a model), and derivative distribution – is appended to an on-chain registry as a non-repudiable, mutually linked lifecycle record. The mechanism requires provable transfer, not tokenization: plain contract storage, ERC-721 tokens, and anchored off-chain receipts are interchangeable representations of the same protocol. A usage-aware gate suspends a consumer's stream when no derivative-creation record is registered within a reporting window; evaluated lazily inside the next delivery transaction, monitoring adds no dedicated transactions. A modality-pluggable fingerprinting layer binds any leaked copy to the dual-signed delivery record of the responsible consumer, instantiated for tabular/geospatial records, images, and documents. We implement the registry as a Solidity contract with EIP-712 dual signatures and evaluate it: a full lifecycle round costs 657k gas with plain records ($0.13 on rollups; ERC-721 tokenization adds ~30k gas per record), and over a 50-consumer pool leak attribution reaches 100% from 40 leaked table rows under moderate noise, survives JPEG recompression to quality 30, and tolerates paraphrase rates up to 30% for documents.

Multidimensional approximate agreement requires $n$ processes with inputs in $\mathbb{R}^d$ to output vectors close to each other, despite up to $t$ Byzantine faults. Under convex validity, outputs must lie in the convex hull of the correct inputs, which leads to resilience thresholds that grow with the dimension. We instead study Minimum Enclosing Ball (MEB) validity and its multiplicative relaxation $c$-MEB validity. Our first contribution is Adaptive MEB Contraction, a coordinate-free algorithm that, in the synchronous model with $n>(d+1)t$ contracts the correct MEB radius by $1/\sqrt2$ per round for $\alpha=1$ while satisfying $\sqrt2$-MEB validity. We also give an example showing that this bound is tight for our analysis. Our main technical contribution is a dimension-free inflation theorem: if every $\beta$ balls in a finite family of Euclidean balls have a common point, then inflating each radius by $\sqrt{\beta/(\beta-1)}$ guarantees a common intersection. Applying the theorem to the candidate balls defining the local MEB-safe areas results in a synchronous algorithm with optimal resilience $n>3t$, contraction factor $\sqrt3/2$ and $\sqrt6$-MEB validity. To the best of our knowledge, this is the first multidimensional approximate agreement algorithm with optimal resilience, constant $c$-MEB validity and coordinate-free contraction. We further extend the approach to the asynchronous setting. Without inflation we obtain resilience $n>(d+2)t$, contraction factor $\sqrt{2/3}$ and $\sqrt6$-MEB validity, whereas with inflation for $n>4t$ we get the contraction factor $\sqrt{15}/4$ and $2\sqrt{10}$-MEB validity. Finally, we compare our guarantees with existing algorithms, including Minimum-Diameter Averaging (MDA), for which we derive MEB-validity guarantees. Our algorithms achieve strictly better resilience while providing substantially stronger MEB-validity guarantees than MDA.

SNNs running on neuromorphic hardware use spikes to achieve sparse and energy-efficient communication over a mesh of cores. In turn, system performance heavily depends on the assignment of neurons to cores: the mapping. Since hardware features inter-core multicast and intra-core replication of spikes, we model SNNs as hypergraphs to exploit both opportunities for reducing communication traffic. Mapping thus comprises two NP-hard problems: hypergraph partitioning and placement on the lattice of cores. High-quality solutions to both are critical, yet increasingly difficult as networks scale to millions of neurons. Therefore, we propose a GPU-accelerated pipeline for SNN mapping: a multi-level partitioning scheme is devised around hardware constraints, while placement is initialized through recursive bisection, followed by refinement pulling together strongly connected cores through repeated swaps. Model-based experiments show upwards of 16% lower latency and 42% lower energy for spike movements over existing sequential tools, while our parallel mapper is on average 18-280x faster.

Fine-grained computation–communication overlap in distributed Mixture-of-Experts (MoE) inference allows communication to begin as partial compute results become ready. However, cooperative thread arrays (CTAs) performing computation and communication contend for finite residency capacity on streaming multiprocessors (SMs). Because a resident CTA generally retains its allocated SM resources until completion, CTAs that cannot be co-resident must wait for resources, resulting in wave-like execution. A fixed resource partition cannot adapt to changes in input size, routed expert load, and kernel configuration, potentially causing a communication backlog or reducing expert compute parallelism. We present a wave-quantized analytical model and launch-time resource manager for dependency-coupled overlap pipelines. Using routed-tile counts, kernel occupancy, GPU residency constraints, and split-level readiness dependencies, it selects the communication-CTA count and resource partition before each launch without candidate execution, per-workload profiling, or kernel recompilation. We integrate the method into the public COMET A100 implementation in FLUX. We evaluate three MoE models on four NVIDIA A100 GPUs under several parallelism strategies at the GEMM2+GatherRS operator, complete post-router MoE layer, and complete-model prefill levels. Across 15 real-p90 workloads, the analytical selector achieves 3.22 percent mean regret relative to the measured oracle with a mean solver overhead of 0.157 microseconds. Over COMET, our method achieves geometric-mean speedups of 2.528x at the GEMM2+GatherRS operator, 1.771x at the complete post-router MoE layer, and 1.185x for complete-model prefill, with maxima of 4.218x, 2.584x, and 1.439x, respectively. At every feasible TP=2/EP=2 sequence length of at least 4,096, our implementation outperforms COMET, Megatron core-TE, and FastMoE TP+NCCL.

Byzantine Reliable Broadcast (BRB) and Message-Adversary-Tolerant Byzantine Reliable Broadcast (MBRB) are reliable-dissemination abstractions for fault-tolerant distributed systems. Yet their operational behavior is shaped not only by specifications and asymptotic communication bounds, but also by serialization, cryptography, buffering, orchestration, deployment environment, and fault-injection semantics. This paper implements and evaluates Bracha [Information and Computation, 1987], AFRT by Albouy et al. [TCS, 2023], and Coded MBRB by Albouy et al. [OPODIS, 2024]. We implement the algorithms in a shared Go codebase with common orchestration, instrumentation, parser-based specification checks, fault injection, and an open-source reproducibility artifact. The evaluation uses single-shot broadcasts in the Shadow network simulator, native profiling, a Google Cloud Platform deployment, and a distributed FABRIC testbed, covering controlled experiments up to 30 nodes, payloads up to 40 MB, 92,190 runs, and 2,361,600 parser-checked entries. The results show that Coded MBRB reduces transmitted data and improves latency in the evaluated cloud setting for larger payloads, but shifts cost to cryptographic and coding computation. Bracha and AFRT incur lower CPU costs at smaller payloads, but their full-payload dissemination increases processing, allocation, and network costs as payloads grow. Across the tested configurations, the parser found no duplicate deliveries, conflicting deliveries, or deliveries of values different from the sender's payload. The paper contributes implementation-level evidence and an extensible artifact for benchmarking BRB and MBRB as executable distributed-system components, exposing bottlenecks and operational trade-offs that are hidden by algorithmic descriptions alone.

Healthcare Internet of Things (HIoT) technology is revolutionising the healthcare industry by enabling real-time data collection and analysis for personalised patient care. However, the rapid expansion of HIoT technology introduces challenges such as increased latency and higher energy consumption in fog computing environments, particularly when managing battery-operated devices. To address these issues, this work proposes a novel scheduling strategy that optimises both power consumption and latency through task-oriented scheduling for HIoT tasks. The proposed strategy, named PLATOS (Power and Latency Aware Task Oriented Scheduling), is implemented in four sequential phases. In the first phase, HIoT tasks are categorised into three groups: priority-oriented, storage-oriented, and computational-oriented. The second phase focuses on latency optimisation by identifying the fog computing resources that yield the lowest execution delay for each task category. In the third phase, power optimisation is achieved by selecting the resources that minimise energy consumption. Finally, in the decision-making phase, high-performance fog resources are allocated to high-priority tasks while the remaining tasks are scheduled based on a mapped list derived from the latency and power optimisation phases. Simulation experiments conducted in iFogSim2 demonstrate that PLATOS reduces energy consumption by 18.72% and latency by 8.65% when compared to the state-of-the-art. These improvements enhance the efficiency and responsiveness of HIoT systems and contribute to more effective patient care and proactive healthcare service delivery.

This paper proposes a robust decentralized personalized federated learning method R-DPFL, that enables clients to reduce the impact of Byzantine attacks via robust neighborhood direction estimation and history-based update trend prediction, rather than purely aggregating client models as in the existing work. In R-DPFL, each client first computes the current-round model update by aggregating the received neighborhood update vectors. It then predicts what this update should be based on its historical values and local model changes. Finally, R-DPFL computes the difference between these two quantities, adaptively clips this difference, and adds it to the local update. We prove convergence of the learning process through rigorous analysis and show that honest clients maintain stable personalized descent dynamics under Byzantine neighbor perturbations without requiring consensus among neighboring models. Extensive experiments on CIFAR-10 demonstrate that RDPFL consistently outperforms state-of-the-art decentralized and personalized federated learning baselines under heterogeneous and adversarial settings.

Given a graph $G=(V,E)$, let $S$ ($\subseteq V$) be a set of vertices. Two vertices are mutually visible if there exists a shortest path in $G$ between them that does not contain any other vertex of $S$. A set $S$ is a Mutual Visibility Set (\MVS) if every pair of vertices in $S$ is mutually visible. The concept of \MVS s in graphs has attracted significant attention since its introduction, as it provides an important structural property of graphs. However, determining a maximum \MVS\ in general graphs is computationally intractable; the decision problem of whether a graph admits an \MVS\ of size at least $k$ has been shown to be NP-complete. Thus, prior work has focused on finding maximal \MVS s or restricting attention to specific graph classes. Cactus graphs form a fundamental low-treewidth class, yet the maximum \MVS\ problem for this class remains open. In this paper, we first determine the size of maximum \MVS~in cactus graphs, and introduce two self-stabilizing algorithms that construct such sets. The first algorithm uses a single BFS tree and stabilizes in $O(D)$ rounds with $O(\log n)$ bits per process on average; the second one uses parallel BFS trees and stabilizes in $O(|C_{\max}|+|T_{\max}|)$ rounds, which we show to be asymptotically tight as a function of these two parameters, even on graphs where $|C_{\max}|+|T_{\max}| = o(D)$.

Computed Tomography (CT) is an essential 3D imaging technology widely used in medical diagnostics and scientific research. However, performing CT imaging on edge devices is challenging due to limitations in computational power, memory capacity, and energy budget. This paper presents an efficient CT reconstruction framework, called edgeFBP, designed for Nvidia Jetson System-on-Chip (SoC) devices. edgeFBP adopts an end-to-end pipeline design for efficient out-of-core image reconstruction under tight power and memory constraints. edgeFBP utilizes a mixed-precision strategy leveraging half-precision Tensor Cores (TCs) to accelerate the bottleneck back-projection (BP) kernel. edgeFBP achieves a 1.83x speedup over the widely used RTK library on Jetson Nano and a 2.56x speedup on Jetson AGX. Under a strict 25-Watt power budget, edgeFBP on Jetson Nano achieves up to 5-48x higher energy efficiency than an Nvidia DGX A100, enabling datacenter-scale imaging on constrained edge devices.

Selecting a parallelism strategy - the configuration of data, tensor, and pipeline parallelism degrees together with micro- and global-batch sizes - largely determines the training efficiency of large language models. State-of-the-art methods search for a parallelism strategy offline and select the single strategy that minimizes per-iteration time. But we find that they neglect the target validation perplexity and time-to-perplexity (TTP). In particular, our analysis reveals that the best strategy yielding the fastest perplexity improvement changes multiple times during training. As a result, state-of-the-art methods are 1.8-11.4x slower in TTP than the strategy sequence that selects the best strategy at each iteration. This paper proposes CONA, a new training method that introduces online strategy chaining. Instead of a single strategy selected offline, CONA ranks candidate strategies during training using a surrogate metric built from compute throughput and gradient statistics, and switches the current strategy to a new strategy with a higher metric. In our evaluation with GPT-3 1.3B, BERT-Large, and Llama-3.2-1B, CONA reaches the target validation perplexity 1.4-9.6x faster than state-of-the-art methods. Moreover, CONA closely tracks the perplexity achieved by the sequence that selects the best strategy at each iteration, within 2.6%.

This paper propose a robust decentralized federated distillation method that enables clients with heterogeneous models to collaborate through predictions on shared unlabeled public data. In the proposed method, each client first evaluates the received predictions in three modalities of class prediction, boundary decision, and prediction correlation. It then filters unreliable clients, assigns reliability-based weights to the retained clients, and constructs a teacher for each type of knowledge. Finally, the corresponding distillation gradients are validated using a supervised gradient computed from private data. Conflicting prediction and boundary gradients are removed, and conflicting relation gradients are suppressed before the final model update. We prove the convergence of the proposed method by showing stable local optimization for honest clients under Byzantine distillation. Particularly, we show that our method ensures a bounded Byzantine influence on both distillation gradients and individual client private gradients after cross-modality fusion, thereby enabling stable local optimization for honest clienunder Byzantine distillation. Extensive experiments on CIFAR-10 and CIFAR-100 demonstrate that the proposed method improves the prediction accuracy of heterogeneous models of clients under non-IID data and Byzantine attacks. As the booming demands of federated learning in decentralized environments such as edge computing and mission-oriented UAV collaborations, our method has a great potential for adoption of DFL in unreliable real-world scenarios where clients are exposed to receiver-specific Byzantine messages of malicious predictions.

The minimum cut problem for an undirected edge-weighted graph asks us to divide its set of nodes into two blocks while minimizing the weighted sum of the cut edges. Over the last years, we engineered a range of fast algorithms for this problem. Our fastest exact algorithm uses an inexact algorithm to obtain a better bound for the problem, reductions that depend on this bound, improved data structures and parallel contraction routines. It is available in the open-source package VieCut and, on real-world instances, outperformed the previously fastest solvers by a factor of up to 2.5 sequentially and up to 12.9 when run in parallel. We improve this algorithm using agentic algorithm engineering (AAE), a methodology that we introduce here, in which autonomous large language model agents run the algorithm engineering cycle on an existing code base: they form hypotheses about where running time is lost, implement them, benchmark the result on a fixed instance set and keep or discard the change. Even though we had already tuned our algorithm by hand extensively, the agent finds significant optimizations, in particular on the DIMACS core instances: factors of 1.28 (sequential) and 1.63 (32 threads) on real-world k-cores, and 6.26 and 127 on the DIMACS core instances.

Speculative decoding accelerates rollout generation, which dominates the cost of reinforcement learning (RL) post-training. Online co-training can further increase the draft's accuracy, yielding greater speedups. However, scaling this approach to co-training on large models with long contexts poses two obstacles: (1) branch attention is unsupported by standard causal context-parallel (CP) implementations, and (2) target features span across pipeline-parallel (PP) stages. We address both with an end-to-end system for large-scale online draft co-training. For CP, we extend packed, load-balanced zigzag ring attention by merging rank-local branch attention with causal main-sequence attention. For PP, TapChannel transports intermediate target features across stages via a separate path, leaving the pipeline schedule unaffected. Experiments demonstrate that co-trained drafts closely track the policy baseline while delivering substantial rollout and end-to-end speedups across model scales up to 122B. Our CP design achieves strong scaling at 256K tokens with significant memory savings over prior work, and our PP transport incurs modest overhead. Code can be found at https://github.com/NVIDIA-NeMo/RL/issues/3698.

By default, Apache Spark's tree aggregation primitives place the tree root on the driver, requiring the driver to participate in the same aggregation computation over intermediate aggregation state as executor nodes. For large aggregates, this can expose the single coordinator to substantial computation and memory requirements. Recent Spark versions optionally move the root to an executor, but the completed aggregate must still be returned to and materialized on the driver. We demonstrate this limitation using exact quantile computation and heavy-hitter identification, where the intermediate aggregation state can be substantially larger than the desired final result. We propose TreeRedux, a minimal extension that adds a terminal finalize operation, executed on an executor, that maps the aggregation state U to a compact result V, so that V rather than the potentially large U is materialized on the driver. For exact quantile computation, applying TreeRedux to GK Select removes the driver's epsilon-n memory term, reducing driver memory requirements to the same asymptotic order as Spark's GK Sketch. In our experiments, the default GK Select implementation encountered a driver out-of-memory error at 2.5 billion elements. Spark's executor-side final aggregation option extended this limit to approximately 16-18 billion elements but still required the final aggregation state to be materialized on the driver. Redux Select completed through 28 billion elements without a driver out-of-memory error. TreeRedux allowed Space-Saving sketches with up to 32x the capacity of the largest configuration that materializes a full sketch on the driver.

Large language model (LLM) inference increasingly spans models that differ in size, capability, price, and provider. This shift creates two costs for developers. One is the integration cost of choosing among and switching between many models. The other is the inference cost of rebuilding a KV cache when it is unavailable or incompatible with the selected model. We define and analyze the Unified AI Gateway as a system setting for an edge-deployed AI traffic hub. It coordinates model routing, KV cache management, and compute placement across end devices, edge resources, and cloud model services. At request time, the gateway jointly selects a target model, an execution site, and a KV cache action under task-quality, latency, cost, and resource constraints. In parallel, background cache-management actions optimize KV cache placement, replication, retrieval, and lifecycle decisions for subsequent requests. We synthesize existing evidence on KV cache reuse, compression, cross-model mapping, distributed storage, and transfer, and discuss the remaining challenges of integrating these capabilities into one system. Across eight typical workload profiles, our workload-level analytical simulation reports TTFT speedups of 1.25$\times$–13.28$\times$ and input-cost benefits of 1.20$\times$–6.16$\times$.