Science & Mathematics

We prove that every nontrivial linear $[n,k]$ maximum distance separable (MDS) code over a finite field $\F_q$ of characteristic two satisfies $n\le q+1$, except in dimensions $k=3$ and $k=q-1$, where $n\le q+2$. This resolves the linear MDS conjecture in even characteristic.

$\mathrm{NiPS_3}$ has attracted considerable attention, yet the origin of its intriguing magnetic behaviors remains elusive due to limited insight into its low-energy anisotropy. Here, high-field torque magnetometry measurements are performed, revealing a previously unrecognized anisotropic spin response that reflects interaction terms beyond the standard XY or XXZ models. In combination with relativistic density functional theory calculations and magnetic force response theory, an effective low-energy spin Hamiltonian is constructed, elucidating not only the observed anomalous magnetic behavior but also other recent experiments. Remarkably, the magnetic dipole interaction and the associated spin-flop transition play a crucial role in driving the deviation from two-dimensional spin alignment, providing direct insight into the underlying anisotropic energy hierarchy. The generalized Hamiltonian can serve as the important fundamental ground for future study, and our approach adopted here will hopefully be relevant to many other related material systems.

Despite rapid progress in text-to-motion generation, existing methods are commonly designed for single-shot generation. When users are dissatisfied with specific motion details, they are often forced to regenerate the entire sequence from scratch, which is computationally inefficient and potentially destroys previously established kinematic context. A practical generation system should instead support iterative user feedback and refine only the affected spatio-temporal regions while preserving the remaining motion. Therefore, we introduce a conversational motion generation paradigm that requires the system to coordinate instruction understanding, motion generation, localized editing, and quality assessment. Recent advances in LLM-driven multi-agent systems provide a promising foundation for such role-specialized coordination. To tackle these challenges, a Self-evolving Multi-Agent RefinemenT Framework (SMART) is proposed for conversational motion generation. Within SMART, to preserve semantic consistency, we design a semantic-guided masked motion editing mechanism, which precisely grounds natural language feedback into spatiotemporal motion regions to enable localized refinements without altering unedited regions. To reduce redundant exploration, we propose a memory-based self-evolving strategy that dynamically accumulates motion editing knowledge from interaction history and reuses past successful experience. Extensive experiments demonstrate that SMART effectively supports multi-turn conversational motion generation and enables efficient motion refinement.

Applying South Africa's draft R3337 warning criteria to food-package photographs requires nutrition, ingredient and product-form evidence. This study presents a region-guided evidence extraction and abstention-aware screening workflow for draft R3337. A three-class detector localises nutrition information panels, ingredient lists and product-identity regions, after which PaddleOCR and structured extraction recover nutrient values, ingredient evidence and product-form information. Recovered evidence is aggregated and assessed using deterministic draft R3337 rules. When evidence is incomplete, uncertain or conflicting, the workflow abstains through Review or Insufficient evidence rather than forcing an automatic outcome. For qualifying unresolved cases, a selective vision-language model (VLM) review stage using Qwen3-VL 8B Instruct reassesses unresolved evidence. Accepted evidence is merged and the same deterministic rules are reapplied. The workflow was applied to 442 products represented by 3,929 images. Evaluation used a manual validation set of 65 products. Before selective VLM review, 31 of 60 decidable products received automatic outcomes and 30 were correct, giving 51.7% coverage and 96.8% accuracy among automatic outcomes (selective accuracy). Selective VLM review resolved eight additional products, increasing coverage to 65.0%. Of 39 automatic outcomes, 38 were correct, giving 97.4% selective accuracy. The single false warning was retained from the rules-only workflow, while no unsafe warning omissions occurred after selective review. Evidence-level evaluation nevertheless identified incorrect VLM-recovered nutrition, ingredient and product-form evidence. These findings show that selective VLM review can recover unresolved evidence, while deterministic screening and explicit abstention remain necessary for reliable image-based draft R3337 screening.

ExeBench is one of the most widely used benchmarks for evaluating LLM-based decompilers. We analyze ExeBench and identify nine issues that affect how faithfully a test measures a decompiler. Three of them prevent LLM decompilers from recovering C code from ExeBench's function. These are ambiguous parameter order when integer and floating-point types are mixed, unused parameters that optimized assembly may drop, and constants stored outside the function's assembly code. Six more affect test validity: non-void functions without a return, placeholder types, static or inline functions removed by the optimizer, functions named main, trivial bodies and truncated definitions. We present ExeBench-FHD, a backward-compatible revision of ExeBench that adds 16 new fields: call-site assembly, data sections, test wrappers, linked-binary disassembly and five warning flags. In ExeBench, 38.9% of functions have at least one of the five flags; dropping records with any of the four warning flags (all except unused_param, which marks functions that the new wrappers make evaluable) keeps 63.9% of the corpus. In an evaluation at -O2 on four architectures, the new fields raise GPT 6 Sol's re-execution rate by up to 26.5%, 19.1% and 21.0% on functions that read data sections, mix parameter types and have unused parameters, respectively.

Progressive neuronal loss is the hallmark of neurodegenerative disease, yet how loss of neurons shapes the pattern of cognitive decline remains poorly understood. Here we systematically ablate neurons (computational units) in language models spanning recurrent long short-term memory (LSTM) and Transformer architectures and apply limited fine-tuning as a simple proxy for compensatory plasticity. With whole-unit ablation, LSTMs degrade gradually with syntactic and semantic faculties declining in parallel across mild-to-moderate damage. By contrast, Transformers exhibit late threshold-like collapse under feedforward multi-layer perceptron (MLP) ablation and smoother sigmoid-like decline under attention-head ablation. Qualitatively, MLP-ablated models continue to make within-category semantic substitutions (semantic paraphasias) alongside rising non-responses, paralleling Wernicke-type aphasia, while attention-head ablation produces a trend toward loss of semantic specificity similar to the semantic variant of primary progressive aphasia (svPPA). These findings indicate that feedforward and attentional pathways have functionally distinct failure modes, suggesting that architecture-dependent failure profiles may offer new computational perspectives on aphasic syndromes and neurodegenerative diseases as an in silico model. Notably, early accuracy retention under Transformer attention-head ablation qualitatively parallels pre-symptomatic reserve before the 40--60% neuronal loss reported by Gomez-Isla et al. (1996) when clinical symptoms emerge in Alzheimer's disease, whereas the steadier LSTM decline parallels the symptomatic progression described by DeKosky and Scheff (1990) and Terry et al. (1991).

We study the parameterized complexity of finding an EF1 allocation of indivisible items in the group-sharing setting, which has received increasing attention in recent years. There, items are not allocated to individual agents but rather to one of two (or possibly more) groups, and agents within that group all share the assigned items. We consider both the "fixed" (EF1-Fix) and "variable" (EF1-Var) group cases, where in the former the assignments of agents to groups are prescribed. While both are NP-hard in the classical sense, we show that their algorithmic behavior differs significantly from the parameterized perspective. As our main results, we: (1) prove that both cases are NP-hard even when the number of agents is a small constant; (2) circumvent this hardness by establishing fixed-parameter tractability for EF1-Var w.r.t. the number of "subjective" items, that is, items which are viewed as chores or goods by different agents---but rule out an analogous result for EF1-Fix; and (3) establish fixed-parameter tractability for EF1-Fix w.r.t. the number of item types, but link the existence of an analogous fixed-parameter algorithm for EF1-Var to a fundamental open question in the setting of Integer Linear Programming.

Post-event building assessment requires linking observed physical damage to its consequences for safety, habitability, functionality, and recovery. Yet post-disaster evidence is often heterogeneous, incomplete, and distributed across images and reconnaissance reports, while recovery models typically require structured engineering inputs. This study develops a multimodal engineering Graph Retrieval-Augmented Generation (GraphRAG) framework that connects post-wind-event damage evidence to physical consequences, building-condition states, recovery actions, and downtime. Image- and text-based evidence is transformed into a provenance-preserving, component-level representation distinguishing confirmed damage, explicit negative evidence, unresolved conditions, and temporary measures. Graph-aligned retrieval combines comparable building scenarios with explicit engineering relationships, while compatibility, contradiction, evidence-precedence, and abstention mechanisms constrain inference under partial observability. Recovery actions are derived from supported damage-consequence relationships, and downtime uncertainty is propagated through an action-dependency network using Monte Carlo simulation. The framework is demonstrated for tornado-affected low-rise residential buildings using reconnaissance and image-based evidence. Comparisons with BaselineRAG and Structured Baseline RAG show improved agreement for key building-condition and recovery outcomes, together with traceable engineering reasoning and preserved evidence provenance. Ablation analyses demonstrate the contributions of engineering context, evidence safeguards, and graph-path traversal. The results support transparent, recovery-oriented assessment from incomplete multimodal post-event evidence.

Large language models can solve complex problems across a variety of subjects, such as mathematics and coding. For such complex analytical tasks, models with a long context perform better. Serving models with long context is currently mostly done on powerful systems with multiple GPUs, which are too expensive for most users. An alternative is local inference, or serving a model on a user's own machine, which removes per-token costs and rate limits, besides keeping sensitive data private and giving users full control over the models they run. The common understanding, however, is that local LLMs need compression (e.g., quantization) to run on a single GPU. In this paper, we set out to study the limitations that prevent serving long-context models for single users, besides the capacity issue. Rather than characterizing a single model on one GPU, we analyze how different models behave across several GPU generations. This lets us uncover not just individual bottlenecks but broader trends in where inefficiencies arise. Our analysis confirms that, as expected, self-attention dominates the runtime for long-context inference. More unexpectedly, our evaluation reveals that the current structure of self-attention leaves key GPU resources significantly underutilized, and the newest GPU is often the least well served. Many of these findings are fundamentally data management problems, from paging the KV cache to improving cache utilization and synchronization in the GPU, which is why we bring this study to the Data Systems community. To engage this community, we present our findings as a series of experimental observations and formulate recommendations for future work.

We consider the problem of model order reduction (MOR) in reproducing kernel Hilbert spaces (RKHSs). We prove generalizations of well-known necessary interpolatory conditions for norm-optimal reduced-order models, originally stated for rational functions in $H_2$ Hardy-Hilbert spaces. We construct an alternative to the iterative rational Krylov (IRKA) algorithm for RKHSs with differentiable kernels to obtain norm-optimal reduced-order models. Then, we introduce generalized Malmquist-Takenaka (GMT) functions that can be used to define orthonormal bases in an arbitrary RKHS. Using GMT functions, we propose unitary transformations between separable RKHSs. This allows us to extend our optimality conditions and subsequent algorithms to MOR problems posed in RKHSs with non-differentiable kernels. The potential and versatility of the proposed MOR framework is demonstrated in several numerical experiments. We perform norm-optimal MOR on various machine learning models such as support vector machines and deep ReLU networks. In addition, we show how the proposed MOR framework can be used to solve seemingly unrelated problems. In particular, we provide an example for fitting high degree polynomials to data and also obtain a FEM-like method for solving partial differential equations (PDEs) with an optimal kernel basis.

Direct Preference Optimization (DPO) minimizes empirical risk over observed preference comparisons, implicitly assuming that the training preference distribution matches the deployment preference distribution. In practice, this i.i.d. assumption often fails, as preference distribution shift exhibits a natural hierarchical structure across three levels: group-level shift, prompt-level shift, and preference-level shift. Such multi-level distributional mismatch poses a critical challenge for safe LLM deployment, yet existing robust DPO methods typically either model uncertainty over the joint preference distribution or focus on individual sources of shift, leaving the hierarchical structure of OOD preference shifts underexplored. We propose TPO, a tri-level distributionally robust framework for OOD preference alignment, which jointly models all three levels of distributional uncertainty through a hierarchically structured rectangular ambiguity set. By instantiating it with Wasserstein distance, we derive a tractable and interpretable objective that combines a closed-form preference-level margin penalty, prompt-level adversarial embedding perturbation, and worst-case group reweighting, each governed by an independent radius. We establish finite-sample convergence guarantees and empirically validate the effectiveness of TPO over existing robust DPO methods.

Developing radiation-resistant fusion reactor materials is constrained by the slow pace of conventional irradiation qualification methodologies, which usually require extensive one-at-a-time irradiation campaigns and post-irradiation characterization. At the current pace of development, we may not discover the materials necessary to deploy fusion power plants at desired rates. In this work, we present a high-throughput (HT) experimental framework for the accelerated down-selection of fusion-relevant, Cu-Cr-Ta RF antenna materials by integrating combinatorial synthesis, in situ ion irradiation, and multi-property screening within a unified workflow. A 2D compositional Cu-Cr-Ta gradient wafer was fabricated by physical vapor deposition and characterized using in situ ion irradiation transient grating spectroscopy I3TGS. Radiation-induced surface acoustic wave frequency fluctuations were used as a rapid proxy for primary defect accumulation, while thermal diffusivity measurements were obtained simultaneously during irradiation. Complementary HT characterization included EDS compositional mapping, four-point probe electrical resistivity measurements, and waste disposal rating calculations to evaluate long-term activation behavior. This integrated methodology enabled an acceleration with respect to conventional irradiation approaches by more than 100 times, from months to hours to analyze 78 new materials.

In the ARRIVAL problem, a token travels through a switch digraph: at every vertex, it leaves along one of two out-arcs, and the two are used alternately. The question is which of two terminals the token reaches first. ARRIVAL lies in $\mathsf{NP} \cap \mathsf{coNP}$, but no polynomial-time algorithm is known.The fastest known algorithms turn some vertices into terminals and search for consistent guesses of the numbers of tokens that these vertices send out. The first, due to Gärtner et al. [ICALP 2021], searches all vertices at once with an algorithm for fixed points of monotone maps, and the second, due to Haslebacher [ICALP 2025], searches one vertex at a time by binary search. Both search for a fixed point of the same map, which returns how many tokens arrive at the guessed vertices. We guess the vertices in blocks of $k$, between these two extremes, which needs a search in $k$ dimensions with few evaluations of this map. To do so, we build on the connection between ARRIVAL and $\ell_1$-contractions established in the second of these works. We search with $\ell_1$-centerpoints of the remaining guesses, which are guaranteed to exist by Haslebacher et al. [FOCS 2025], and which we approximate deterministically. In order to prove our results, we show how to find the least fixed points of maps from $\{0, \ldots, N\}^k$ to itself that have unit increments. For ARRIVAL on $n$-vertex switch digraphs, this gives run time $2^{\mathcal{O}(n^{1/3} \cdot \log^{2/3} n)}$, improving the previous time bound of $n^{\mathcal{O}(\sqrt{n})}$. We work with the more general \garrival problem, with many tokens and several terminals, and solve it in the same time when there are at most $2^n$ tokens.

Autonomous agents are moving from application-layer components to active participants in networked systems. They not only generate traffic but also perceive context, express intent, invoke services, coordinate with other agents, and act under task, policy, and resource constraints. This evolution exposes a mismatch between emerging agentic interactions and conventional endpoint- and traffic-centric network abstractions. This white paper introduces Agentic Communication Networks (ACN) as a network-centric paradigm for the time when agents become first-order network entities. At its core, ACN reshapes networking around three fundamental aspects. First, agents are represented as identifiable and governable network entities with explicit identity, authority, and lifecycle. Second, agentic services are expressed through intent, capability, context, policy, and task state, allowing the network to capture what an agent seeks to accomplish and under what conditions. Third, agentic execution becomes an explicit network concern as tasks involve multiple agents, services, tools, and distributed capabilities across heterogeneous infrastructures. Based on these shifts, the paper defines the scope, principles, architecture, and research directions of ACN, positioning it as a networking foundation for autonomous agent ecosystems.

One-to-many ultra-wideband (UWB) transport can disseminate shared audio blocks without replicating a downlink stream for every receiver. Yet common reception does not ensure aligned electrical playback, coincident acoustic arrival, or delivery to an entire receiver group. This paper develops a unified feasibility analysis of position-assisted UWB multicast audio that separates clock and actuation error, age-dependent acoustic propagation compensation, correlated block erasures, and reserved communication resources. We derive a conditional bound on one-listener arrival spread, a stationary expression for all-receiver block delivery under common and receiver-local losses, and complementary bounds for reliability-constrained admission. Reproducible Monte Carlo experiments examine 4,590 policy--run records across clock, localization, channel, and slot configurations. For eight receivers, affine clock correction produces a mean run-level 95th-percentile electronic span of $21.19\us$, while position assistance reduces the corresponding one-listener acoustic arrival span from $21.95\ms$ to $419.43\us$. Under the stated $30\us$ timing criterion, 98.86\% of compensated events still exceed the threshold. For 64 receivers, three block copies attain an all-receiver delivery probability of approximately 0.97478 under the modeled correlated channel. The analysis exposes the distinct failure mechanisms of position age, shared fades, retransmission limits, and schedule saturation, and establishes joint criteria for interpreting synchronous multicast audio performance.