Science & Mathematics

AI agents must decide when another search is worth its cost. Answer-success probabilities capture question difficulty, while this decision requires estimating the additional benefit of search. We introduce SearchHorizonBench, a benchmark for learning and evaluating search value from execution experience. Paired continuations and repeated executions separate expected search gain from generation noise and connect prediction quality to decision utility. The resource contains 45,632 executions across 3,168 questions. On 1,632 held-out questions, historical calibration reduces Brier error by 16.2%, with the improvement concentrated in success-level information. Search gains have a standard deviation of 21.4 percentage points, and tested historical estimators explain less than 2% of this variance in point estimates. Matched-configuration learning curves probe this distinction across training scales. At the reference token price, free selection observations yield a 2.49-point utility opportunity over the best constant action, driven primarily by gain information. Reusable histories, frozen splits, strong baselines and scoring tools provide a common foundation for developing and evaluating value-aware search agents.

Vision--language--action (VLA) models increasingly run on edge devices aboard robots, where inference latency constrains closed-loop robot control. However, it remains unclear how to design VLA models, their inference software and edge hardware so that robots can run VLA inference in real time. Most VLA models report only task success, while prior studies cover limited action-generation strategies, divide inference into at most three stages, and lack comparisons of inference optimizations on the same models.benchmarking 18 open VLA models across released and alternative backends on NVIDIA Jetson AGX Thor and AMD Ryzen AI Max+ PRO 395, with a seven-stage inference breakdown. Our study reveals three insights. First, action-generation strategy matters more than model size. It dominates 14 of 19 Jetson configurations, latency varies by $82\times$ on Jetson and $157\times$ on AMD, and parameter count explains at most 15% of latency variance. Models that generate long action chunks in a few flow-matching steps keep up with control, whereas token-by-token decoding does not. Second, kernel fusion with 4-bit weights helps most, running $π_{0.5}$ and GR00T N1.7 2.6--2.9$\times$ faster than PyTorch with \texttt{torch.compile}, but it exists for only these two models and only on the Jetson. Optimization also moves the time to the stages it shortens least, as CPU preprocessing grows from 12\% to 39\% of the GR00T N1.7 call. Third, after optimization, memory bandwidth limits only the stages that process few tokens on the Jetson, such as the denoising steps, whereas software limits all other stages, from unfused kernels to serial preprocessing on the host CPU. Overall, low-latency edge VLAs should use few action-generation iterations, inference software should extend kernel fusion and low-bit weights to more models and platforms, and edge hardware needs high memory bandwidth and a fast host CPU.

When a Mixture-of-Experts (MoE) model does not fit in GPU memory, batch-1 decoding is dominated by copying expert weights from host memory. Cache-aware routing reduces these copies by biasing the router toward resident experts, but combined with asynchronous prefetching it makes the model's outputs depend on transfer timing, so a harness that does not simulate that timing no longer evaluates the deployed model. We propose prefetch-steered routing (PSR): routing conditions on a deterministic logical cache, into which at most n experts predicted by the next layer's router are inserted per layer and token; physical copies affect only latency. On OLMoE-1B-7B with 50% and 25% of experts resident on an RTX 4090, PSR decodes 2.2-2.3x faster than exact routing with LRU and 25-30% faster than the Cache-Prior operating point, with mean benchmark differences of +0.1 and +0.5 points (95% lower bounds -0.8 and -0.3). In greedy generation its lead over Cache-Prior depends on the workload and reverses on HumanEval at 50% residency. On Moonlight-16B-A3B, PSR is 1.9x faster than exact routing with LRU at 37.5% residency and, at the latency of Cache-Prior, significantly better on three of eight benchmarks. On Qwen3.6-35B-A3B, where copies are cheap relative to computation, PSR keeps its advantage over Cache-Prior on the likelihood benchmarks but not its speed. Distillation-trained gates did not help. Speculative decoding with block-level expert selection was slower than PSR with a 1B autoregressive drafter on OLMoE, but with a block drafter on Qwen3.6 it was 2.1-2.9x faster than any autoregressive configuration with at most its GPU memory on the 64 prompts we timed.

Multi-agent systems (MAS) require both effective coordination and specialized task execution. Existing prompt optimization approaches typically couple these functions within agent-specific prompts. We introduce Coordinator-Executor Prompt Optimization (CEPO), a general framework for improving prompt optimization in MAS. CEPO separates system-level coordination from agent-level execution through a shared coordinator prompt and agent-specific executor prompts, and guides their updates according to their respective roles. This structure allows shared coordination rules to be refined consistently across agents while keeping execution instructions separately adjustable. CEPO provides the prompt structure and optimization guidance, while existing prompt optimizers perform the underlying search. Extensive experiments spanning 11 benchmarks across four domains demonstrate that applying CEPO to four existing prompt optimizers consistently improves their performance. Ablations and qualitative analyses support the proposed design, while controlled studies on a document-understanding MAS further demonstrate improved optimization budget efficiency and performance scaling with the number of agents.

Verifying quantum programs today requires specialized proof systems built on interactive theorem provers such as Rocq, Lean, Isabelle, or Why3, and even modest quantum algorithms cost substantial human effort to verify. Because quantum programs cannot be effectively checked by runtime testing, formal verification is the only viable path to correctness, making the cost of these bespoke systems the chief barrier to scalable quantum software assurance. We show that verifying a quantum program should be no harder than verifying a classical array program, a paradigm commonly used in parallel programming: the Qafny quantum proof system translates into Hoare logic with array theory completely (every Qafny proof becomes a valid Hoare triple) and soundly (a valid translated triple certifies only true quantum specifications), and we mechanize the completeness direction in Rocq. The embedding determinizes quantum proof obligations and exposes the full machinery of classical automated verifiers (e.g. Dafny and SMT-based separation logic) to quantum code, removing the need for dedicated quantum proof system infrastructure. In particular, the embedding supplies the correctness argument that the existing compilation of Qafny into Dafny lacks.

Long-horizon agents continuously encounter reusable procedural knowledge, yet existing systems typically rely either on model adaptation or external memory, trading persistent learning for rapid reuse. This limitation is especially important for Web agents that repeatedly execute related multi-step workflows. We propose HaMo, a Harness-Model collaborative framework that combines fast non-parametric experience accumulation with slow parametric adaptation. HaMo distills trajectories into strategy-level experience cards for immediate retrieval, while an experience-conditioned continual learning objective trains the model to interpret, ground, and execute retrieved strategies across changing contexts. A retrieval-coverage-based selection mechanism further identifies compact and complementary trajectories for parameter updates. Experiments show that HaMo outperforms harness-only and model-only adaptation under within-domain and cross-domain shifts, improving both retention and adaptation. These results highlight the importance of jointly preserving procedural experience and internalizing the competence to transfer and execute it across tasks.

Value-based Task Selection (VTS) asks whether a real-time system can select tasks whose total value reaches a required target while preserving every deadline. We study its threshold decision form for finite collections of independent sporadic tasks scheduled preemptively on one processor. We prove that this problem is $Σ_2^P$-complete for both constrained and arbitrary deadlines, even when all numerical parameters are encoded in unary. The existential choice is the selected task set, while the universal condition requires every processor-demand window to remain feasible. Our hardness reduction already uses constrained-deadline tasks. It represents universal assignments by Chinese Remainder Theorem timestamps and uses polynomial-weight local selector locks to enforce a valid existential assignment. The reduction produces polynomially many tasks, and every generated execution time, deadline, period, task value, and target is polynomially bounded in the source-instance size.

Skills provide LLM agents with reusable procedural knowledge, making appropriate skill selection essential for reliable agent behavior. As skill libraries expand, distinguishing semantically similar candidates becomes increasingly difficult. Existing methods encode or rewrite skill descriptions for relevance-based ranking, yet heterogeneous descriptions can obscure capability differences. To address this challenge, we propose SKALOS, a graph-native reranker that compares skills via explicit capability evidence. Its multi-hierarchy graph aligns equivalent expressions across skills while preserving skill-specific capability evidence. Since individually relevant capability elements may not jointly support a query, semiring-based capability witness learning scores each branch by its weakest evidence, checks joint support for the extracted requirements within one implementation, and selects the strongest alternative. To reduce the influence of extraction errors, we develop ranking-feedback graph calibration that adjusts unreliable evidence for candidate comparison. Experiments on SkillRet show that SKALOS improves NDCG@10 by 2.68 percentage points over a matched text-only reranker. Witness-guided skill selection also increases ALFWorld task success rate from 22% to 29%.

Semantic error control coding (SECC) has emerged as a promising paradigm for improving communication reliability by exploiting the inherent structure and semantic dependencies of transmitted information. This paper proposes a semantic successive cancellation list (SCL) decoder for polar codes carrying natural-language text. A fine-tuned byte-level language model provides next-byte priors, which are converted into conditional bit costs and incorporated directly into the SCL path metric before each pruning step. This allows the decoder to jointly exploit channel reliability and source structure while retaining the conventional polar coding framework. The language model can also be invoked only for bytes whose channel decisions are unreliable, trading decoding performance for latency. Simulations for length-512 polar codes show substantial block error rate (BLER) improvements over conventional CRC-aided SCL (CA-SCL) decoding. At rate R=0.75 and list size L=16, the proposed decoder achieves more than 1 dB of gain in BLER compared to CA-SCL with the same L. Moreover, with L=8, it outperforms conventional CA-SCL with L=256 in terms of BLER, while requiring about one quarter of the decoding time of the latter in the evaluated software implementation.

The canonical $\mathbb{Z}_{2^s}$-linear Hadamard codes are the Gray images of the $\mathbb{Z}_{2^s}$-additive codes. Their rank, the dimension of the binary linear span, was known in closed form only for $s\le 3$, and for $s \ge 4$ only through computer computations for small lengths. We give a method that computes the rank for every $s$. We show that the rank is a binomial polynomial in the type parameters, with coefficients $ρ_s(\boldsymbol{k})$ independent of the length. We compute all the coefficients for $s\le 5$, which yields explicit rank formulas over $\mathbb{Z}_4$ and $\mathbb{Z}_8$ (recovering the known ones), and over $\mathbb{Z}_{16}$ and $\mathbb{Z}_{32}$.

This paper presents Memory- and Communication-Aware Framework for Long-Context LLM Inference (MCF) that optimizes long-context inference through communication-aware and stage-adaptive KV-cache layout transitions. At the core of MCF is {Memory-Aware Replication and Scheduling} (MARS), a dynamic orchestration mechanism for chunked prefill. MARS exploits early-stage memory headroom by selectively replicating highly reused KV blocks, improving locality and hiding KV propagation latency behind attention computation. As prefill progresses and memory pressure increases, MARS progressively drains redundant replicas and converges the KV cache into a sharded decode-ready layout, avoiding expensive layout conversion at the prefill--decode boundary. For decoding, MCF adopts a query-centric execution model in which KV shards remain stationary while lightweight query tensors are routed across workers. This changes communication from KV-cache-dependent transfer to query-dependent transfer, reducing communication growth with context length. MCF further integrates a hardware-software co-designed execution layer based on an FPGA compute-NIC datapath, enabling metadata-driven KV routing and collective operations directly in the network path. Experiments show that MCF improves TTFT, throughput, and scalability over state-of-the-art distributed inference systems, with larger benefits as context length and cluster size increase.

Conversational AI systems are increasingly used to fulfill both functional and affective needs, yet how users perceive the AI-adoption effects varies. Existing research tends to treat these two needs as binary categories and to classify users by a single dominant need. However, this approach fails to capture users who engage with AI for both affective and functional purposes simultaneously, leaving it unclear how these needs combine in real-world AI use. To capture the multifaceted nature of users' AI engagement, we developed the AI-Support measurement with 1,090 participants, grounded in social support theory, which evaluates four dimensions of perceived AI support: emotional/esteem, social companionship, informational, and instrumental support. This study provided (1) an AI-Support Scale developed to measure the perceived support types from human-AI interaction, (2) evidence that users might adopt AI with a high level of both functional and affective use, and (3) exploratory indicators of users' real-life social support and well-being, enabling descriptive results aligning with theoretical expectations. Overall, this study highlights the importance of understanding individual differences in AI affordance use, enabling AI systems to be designed in alignment with different user adoption profiles rather than assuming uniform benefits across all users.

Frequency-domain information is beneficial for semantic segmentation. However, existing frequency-based methods often suffer from insufficient frequency utilization, inefficient fusion, and high computational complexity. To address these issues, we propose FGD2HF, a dual-domain fusion network that leverages frequency features to guide spatial-domain fusion and enhancement. Specifically, we introduce a novel Hybrid-Attention Encoder (NAHA) that jointly models local and global relationships through neighborhood and hypergraph attention, enhancing the network's ability to extract effective features. We further introduce a Wavelet Transform and Enhancement (WTE) module to extract and refine low- and high-frequency cues from preceding-stage features, followed by Frequency-guided Dual-domain Fusion (FGDF) modules that use frequency cues to condition current-stage spatial grouping and feature propagation. This design incorporates both global and local frequency guidance into spatial feature modulation, enhancement and fusion with low computational cost. Extensive comparisons with numerous state-of-the-art semantic segmentation methods demonstrate that our method is highly competitive. Our full FGD2HF-B model achieves 48.7% mIoU on ADE20K (+0.3% over SegMAN-B with 18% fewer GFLOPs), 83.2% mIoU on Cityscapes (comparable to SegMAN-B with 21% fewer GFLOPs), and 1.1% higher mIoU than VWFormer-T on SBD with 21% fewer GFLOPs, while also achieving significant improvements in boundary segmentation accuracy.

An agent that truly knows and acts on behalf of its user has been a long-standing promise of AI assistants. Computer-use agents bring this promise within reach by capably executing fully specified tasks across apps. However, real-world user requests are often underspecified. To act correctly for the particular user, an agent must infer some missing information from the user's activity history and ask for other information that only the user can provide, without making unsupported assumptions or burdening the user with unnecessary questions. We present iPrefBench, a benchmark for evaluating the infer and ask capabilities of personal mobile-use agents in an executable iOS environment with an interactive simulated user. Grounded in a coherent persona with 3,051 past activity events, iPrefBench comprises 130 underspecified user tasks across 26 apps. Each omitted user-dependent requirement is, by construction, either history-inferable or ask-only. iPrefBench evaluates personalized mobile agents on both whether agents complete tasks correctly for the user (i.e., personalized task completion) and whether they seek user input when necessary (i.e., user-agent interaction). Evaluation of 7 frontier agents reveals a substantial gap between execution and personalization. Agents achieve high action completion scores (up to 91.6%) but low full task success (up to 47.7%) because personalization lags behind. They especially struggle with ask-only information, failing to request enough necessary information. Our findings highlight a challenge for personalized agents in knowing what to infer and when to ask. We hope that iPrefBench can serve as a valuable testbed for building agents that know their user and work with them.

Materials with chemical identity Mn3X, where X is Sn, Ga, or Ge, are Heusler alloys with intriguing properties that are fundamentally and technologically interesting. They have topological properties with a non-trivial band structure that identify them as Weyl semimetals in the bulk. Their magnetic properties, which give rise to the topological band structure through spin-orbit coupling, are determined by their Kagome lattice structure, making them chiral non-collinear antiferromagnets. Among these materials, Mn3Ge is particularly interesting because it has a larger Berry curvature, leading to a large anomalous Hall conductivity and motivating potential spintronic applications. The desirable properties are only obtained when Mn3Ge has the D019 crystal structure, so producing an easily reproducible growth recipe is the main objective of this paper. Here we show that ultra-high-quality thin films of Mn3Ge with D019 symmetry can be grown from a Mn5Ge target on an Al2O3(0001) substrate with a Ru(0001) seed layer. The expected non-collinear antiferromagnetic order was confirmed by the finding of weak zero-field magnetisation (0.0086 +/- 0.0002) mu_B/Mn that does not saturate at high magnetic fields and has a Neel temperature of (291 +/- 2) K.