Science & Mathematics

Designing building energy systems is a high-dimensional, mixed-integer search problem: when each design is evaluated by a year-long simulation, black-box optimizers typically need hundreds of evaluations. This paper presents AutoEnergyDesign, a framework in which a Large Language Model (LLM) agent reasons over previous simulation results, proposes configuration changes and learns in context from each outcome, while a git-based mechanism automatically reverts unsuccessful changes. The agent couples a reasoning model with an explicit memory organisation (procedural, semantic, episodic and reflective), a constrained action interface and an environment-grounded keep/revert signal. The simulator covers eight technologies (photovoltaics, micro-wind, combined heat and power, solar thermal, gas boilers, batteries, heat pumps and inverters) and computes the 25-year Net Present Cost (NPC), including capital and operating costs and salvage value. On a case study with measured building load profiles, the agent, run with Claude Haiku 5.5, Sonnet 5.5 and Opus 5.5 (ten runs of 50 simulations each), is compared with random search, Tree-structured Parzen Estimator (TPE) Bayesian optimization and a $(1+1)$ evolution strategy sharing the agent's acceptance rule (300 evaluations, ten seeds). With all three models the agent reaches an NPC of about 3.53 MEUR, 1.20% below the best TPE result and 0.23% below the best evolution-strategy result. It comes within 1% of this design after a median of 18 (Opus) to 27.5 (Haiku) simulations, whereas among the baselines only the evolution strategy does so, in half of its seeds and after a median of 192 evaluations. Opus reaches the best design in all ten runs, Sonnet in six (two runs never come within 1%), and Haiku comes within 1% in every run but reaches it exactly only once. Every decision is recorded in an auditable trace.

Disseminating synthetic data in place of original ones is a promising way to enforce privacy while complying with open science principles, especially in the healthcare domain. To keep pace with the rate at which researchers collect such data, there is a need for fast and flexible synthesis procedures. A popular class of such procedures is based on models of conditional distributions induced by decision trees. Such models can be combined into a synthetic distribution following two main approaches: sequential specification and fully conditional specification. However, there is currently no recommendation on which specification to use to optimize the risk-utility tradeoffs. In this work, we use both specifications to synthesize medical datasets by combining tree-based models of conditional distributions. We compare the quality of the resulting synthetic data using several metrics of risk and utility. Our results indicate no general superiority of either specification over the other. Sequential synthesis seems sensitive to the ordering of the variables, while the fully conditional specification involves more hyperparameters and incurs higher computational costs, which may limit its suitability for off-the-shelf synthesis procedures.

Rotationally equivariant architectures improve the physical fidelity of machine-learned (ML) universal materials models based on graph neural networks (GNNs). However, their scalability is limited by the use of non-trivial, high-precision tensor operations during both training and inference. Adopting the reduced-precision formats of matrix multiplication units (MMUs), e.g., FP8 instead of FP32, provides massive increases in throughput, up to 30x (60x) on NVIDIA Hopper (Blackwell) GPUs, but may lead to drastic loss of accuracy. Here, we introduce tailored 'emulated' low-precision operations under the integer Ozaki scheme to accelerate the 'SO(2)-equivariant' processing layers of GNNs, which dominate the runtime of (1) many state-of-the-art ML interatomic potentials (MLIPs) and (2) most universal ML Hamiltonian (MLH) models. Our framework, FlashSO(2), outperforms TF32, the most accurate native low-precision MMU format, in throughput, accuracy, and memory footprint at once. At the SO(2) layer level, it reaches 40x (7x) the throughput of MALOQ's (UMA's) reference implementation, an MLH (MLIP) model, sustaining up to 236 (232) effective Tflop/s on an NVIDIA GH200, 3.5x the FP32 peak. At the model level, FlashSO(2) helps train MALOQ 3.9x faster than its pure-FP32 version while reducing peak memory by 58% at no loss in accuracy. In molecular dynamics simulations with UMA, the speedup factor is 3.3x, and the peak memory is reduced by 51%. FlashSO(2) thus pushes the speed-accuracy Pareto frontier of equivariant materials models beyond the native low-precision formats and substantially lowers the resources required to exploit their scaling laws.

Quadrotor controller comparisons commonly report a single tracking error, although the resulting ranking also depends on tuning effort, reference information, actuator prediction, optimizer fallback, and command timing. We quantify these effects for cascade PID, hover LQR, boundary-layer sliding-mode control (BL-SMC), linear MPC with LQR fallback (LMPC-F), a preview-enabled LMPC-F, and a geometric $SO(3)$ controller on a common nonlinear plant with rotor gyroscopic coupling, motor lag, and saturation. A primary study (556 episodes at 50~Hz) combines matched twelve-candidate tuning, 400 paired evaluation episodes, and a 48-cell MPC factorial; a confirmation study (1462 episodes) repeats the evaluation on 50 fresh realizations with tuning budgets of 12, 24, and 48 candidates. A matched twelve-candidate tuning protocol changed the mean-RMSE ordering obtained with the fixed baseline settings, moving BL-SMC from last to first; the new ordering recurred on the fresh realizations, and a 48-candidate budget left it unchanged. Reference preview substantially reduced MPC tracking error and, on the figure-eight task, removed fallback activity. On an abrupt step, the outcome depended on the interplay between fallback policy, predictor, and actuator authority. Command timing also changed the outcome: replaying measured command-call delays terminated three of four nominal MPC configurations on a desktop computer and all four on a Raspberry Pi~5, whereas capping the solver at 100 iterations removed every deadline miss in the tested nominal timing profiles. Controller rankings therefore depend not only on tracking accuracy but also on the tuning protocol, reference information, fallback behavior, and implementation timing. We propose a reporting checklist covering tracking accuracy, tuning effort, completion and failure rates, fallback activity, applied-input variation, and command timing.

Spatial transcriptomics provides spatially resolved molecular profiling, but its cost, limited throughput and experimental complexity hinder routine large-scale use. Predicting spatial gene expression from hematoxylin-and-eosin-stained histology images offers a scalable alternative, yet remains challenging because histology provides only an indirect morphological view of molecular activity. Existing methods typically decode expression from visual features in a single stage, even when enhanced with spatial aggregation, retrieval or multi-resolution fusion. This formulation is restrictive because spot-level expression reflects not only local morphology but also cellular identity, neighborhood composition and tissue-level organization. We propose BCARP, a Biology-guided Context-Aware Residual Propagation framework for histology-guided spatial transcriptomics prediction. Instead of directly predicting gene expression from histology embeddings, BCARP treats them as morphology visual priors and progressively refines spot representations in a transcriptomics-oriented latent space. Each refinement step integrates nuclei-guided local cellular context with prototype-guided global context propagation, enabling information exchange across both neighboring and spatially distant spots. A pattern-scale decoder further separates spatial expression into pattern prediction from raw expression with scale calibration. Evaluated on three ST datasets and ten HEST-1k cohorts, BCARP achieves the highest mean correlation-based performance against other baselines. Numerical results support biology-guided local-global residual propagation as an effective alternative to visual decoding for histology-to-transcriptomics prediction.

The value of language models for any community depends on whether a model actually understands the culture of the people it intends to serve. We use the Igbo language as a case study. Existing benchmarks test translated Western ideas, lacking depth. To address this, we introduce CAB-Igbo, a benchmark developed from indigenous cultural sources and verified by native speakers across six domains. We evaluated diverse models under varying conditions, incorporating regional personas. Performance was assessed using multiple-choice accuracy and human evaluations across Epistemic Fidelity, Representational Richness, Linguistic Quality, and Pragmatic Proficiency. Results reveal a stark dissociation between language fluency and cultural knowledge. Through statistical testing, we show that for most models, the severe lack of Epistemic Fidelity and Representational Richness is already present in English and does not significantly worsen when switching to Igbo. Instead, native language prompting mainly degrades Linguistic Quality and Pragmatic Proficiency. Gemini 2.0 Flash is the only model whose cultural knowledge scores improve under Igbo prompting. We find that current models struggle to demonstrate the knowledge needed for true alignment with Igbo, and multiple-choice formats allow pattern-matching to appear as understanding. Furthermore, automated judges agree with humans on linguistic quality but fail on epistemic fidelity. Ultimately, cultural evaluation must be built from community knowledge, establishing human evaluation as the gold standard to prevent misrepresenting low-resource communities globally.

Linear Temporal Logic over Finite Traces (LTLf) is widely used in Artificial Intelligence to specify desired temporal behavior and determine whether finite traces satisfy or violate it. Explaining these outcomes is essential for debugging and trust. However, existing approaches typically capture either sufficient reasons or counterfactual causes, without clarifying their relationship. We introduce a native characterization of abductive (AXps) and contrastive (CXps) explanations directly over finite traces, based on locking and unlocking time-indexed events. We prove its equivalence to the classifier-based formulation, and show that CXps coincide with a recently proposed temporal causality semantics for LTLf. This characterization enables an efficient implementation based on the enumeration of minimal unsatisfiable cores and minimal correction sets of ASP programs. Experiments on real-world event logs demonstrate the effectiveness of the approach.

A unified framework is developed for parameter estimation (PE) and channel estimation (CE) in monostatic multiple-input multiple-output (MIMO) direct-current-biased optical orthogonal frequency division multiplexing (DCO-OFDM)-aided visible light integrated sensing and communication (VL-ISAC) systems. A common optical waveform supports sensing and downlink communication while accounting for target reflections, clutter, Doppler effects, and the communication channel. For sensing PE, one-dimensional (1D) and two-dimensional (2D) on-grid and off-grid sparse representations are formulated. The proposed 1D off-grid Bayesian learning (1D-OffGBL) algorithm employs a joint delay-Doppler (DD) dictionary to exploit coupled DD information and achieve high estimation accuracy. In contrast, the proposed 2D off-grid Bayesian learning (2D-OffGBL) algorithm sequentially estimates the Doppler and delay parameters, thereby avoiding a high-dimensional joint dictionary and reducing computational complexity. Both algorithms iteratively refine the continuous-valued delay and Doppler parameters to mitigate off-grid basis mismatch. Furthermore, 1D and 2D sparse channel representations and corresponding CE procedures are developed, followed by range and velocity estimation for both sensing models. Closed-form Bayesian Cramer-Rao bounds (BCRBs) are derived for PE, CE, range, and velocity estimation. Simulations demonstrate that the proposed algorithms outperform conventional sparse recovery schemes in terms of normalized mean-squared error (NMSE). In particular, 1D-OffGBL achieves the highest estimation accuracy, whereas 2D-OffGBL provides a favourable performance-complexity trade-off for practical MIMO DCO-OFDM VL-ISAC systems.

Existing robot navigation methods primarily focus on isolated tasks, overlooking the need to accumulate knowledge and experience across repeated navigation episodes. Although recent lifelong navigation methods have explored cross-task knowledge accumulation, they mainly focus on scene knowledge while neglecting historical navigation experience and its connection to scene evidence, limiting the ability to reflect on past mistakes and improve future decisions. To address this limitation, this paper presents RetroNav, a training-free self-evolving navigation framework that integrates navigation, memory accumulation, and failure reflection. Specifically, we develop a unified memory system that connects geometric and semantic scene knowledge with historical navigation experience, distinguishing objective observations from task-dependent interpretations. After each navigation attempt, an independent evaluator determines whether a retry is necessary. If so, a Retrospection Agent revisits historical decisions using accumulated scene knowledge and navigation records, generating an evidence-grounded retry brief to guide the next attempt. RetroNav achieves state-of-the-art performance across five navigation benchmarks. Our method will be open-sourced at https://github.com/3dv-casia/RetroNav .

While large-scale models achieve strong results in remote sensing scene classification, deploying them on edge devices remains difficult due to limited compute, memory, and energy budgets. Compact CNN backbones are therefore a practical choice in many real-world settings, offering efficient inference with well-optimized implementations on embedded hardware. To push this efficiency further, we investigate how far a knowledge-distilled student model can be compressed while preserving high accuracy. We propose a Distillation-Aware Compression Loop (DAC-Loop) that integrates (i) knowledge distillation applied both before and after structured channel pruning, and (ii) an iterative compression schedule that alternates structured pruning with unstructured sparsification. Across EuroSAT and NWPU-RESISC45, the DAC-Loop consistently improves the accuracy-compression trade-off relative to standard one-shot pruning and single-stage KD recovery, maintaining strong accuracy even beyond 90% parameter reduction. These results demonstrate that combining distillation and pruning in a unified iterative loop offers an effective path toward compact remote sensing models.

Modulation on conjugate-reciprocal zeros (MOCZ) is a promising candidate for blind short-packet communication (SPC) in future wireless systems; however, its performance is sensitive to timing and frequency synchronization errors. In this paper, we propose a novel scheme named multi-radius binary MOCZ (MR-BMOCZ). This method introduces a new root placement strategy that enhances diversity in the autocorrelation domain. The resulting structure enhances the autocorrelation properties of the transmitted waveform, leading to improved timing offset (TO) estimation. Moreover, the multi-radius construction distributes the signal energy over a larger set of radii, hence reducing the peak-to-average power ratio (PAPR). The resulting waveform also exhibits a distinctive spectral structure that is exploited for carrier frequency offset (CFO) estimation and correction, yielding lower normalized mean square error (NMSE) compared with BMOCZ-based schemes. Simulation results over additive white Gaussian noise (AWGN) and fading channels demonstrate that MR-BMOCZ achieves superior synchronization performance while maintaining comparable bit error rate (BER) and block error rate (BLER) performance. Consequently, the proposed scheme offers a synchronization oriented extension of BMOCZ for robust non-coherent SPC.

Large language models (LLMs) are increasingly used for codebook-based content analysis in computational social science, where codebooks translate theoretical concepts into labeling schemes. While prior work reports strong predictive performance, it remains unclear whether these systems produce reliable and robust text annotations. We conduct a large-scale empirical failure-mode analysis of LLM-based coding pipelines across four datasets and five models by systematically evaluating their stability, robustness, and conceptual validity under controlled perturbations, domain shifts, and optimization procedures. We uncover multiple failure modes that are largely invisible under standard metrics, showing that high aggregate scores frequently coexist with severe label-level disparities, instability across runs, domain sensitivity, and shifts in label interpretation. Importantly, these failure modes persist under widely used optimization strategies and remain largely invisible under standard metrics such as accuracy and micro-F1. Our results demonstrate that current LLM-based coding pipelines cannot be treated as stable measurement tools without targeted validation beyond aggregate performance.

This paper extends the least-squares and certainty-equivalence minimum-variance framework of classical linear self-tuning regulators to a broad class of nonlinearly parameterized systems with compact parameter spaces, including commonly used deep neural network models. First, historical nonlinear least-squares fitting guarantees are converted into online one-step prediction bounds under a general modulus of continuity, with explicit regret rates under parameter Lipschitz continuity and finite-moment noise. Second, approximate NLS estimation combined with certainty-equivalence control achieves closed-loop stability and asymptotically optimal tracking under mild conditions without requiring excitation or parameter identifiability. Explicit closed-loop regret and tracking error rates are established in the parameter-Lipschitz case, and closed-loop examples show that the regret rates cannot, in general, be improved for this regulator. Finally, robustness to small unmodeled dynamics is established by showing that closed-loop stability is preserved and the tracking error bounds comprise a vanishing term and a residual term determined by the model discrepancy.

Large language models increasingly act through code, modifying repositories, executing programs, and interacting with services in agent form. Their actions introduce risks that cannot be assessed from generated responses alone. Safety evaluation must verify actual consequences, distinguish harm avoidance from task incapability, and cover realistic workflows. We introduce V$^3$AST-Bench, a benchmark for Verifiable results, Valid interpretation, and Versatile risk coverage of Agent Safety. It comprises 212 risk tasks spanning 18 categories. Dedicated runtime environments verify harm and task outcomes through executable checks over states, records, and artifact behavior, while paired capability twin tasks verify workflow competence. Our experiments span 51 combinations of 11 models and five harnesses, spanning 86,496 trials and over 9 billion tokens of model interaction. Results show that V$^3$AST-Bench remains challenging even for frontier coding agents, as Claude Opus 4.8 with Claude Code still causes verified harm on 20.9% of tasks. Across model-harness combinations, agents with comparable competence exhibit markedly different safety outcomes, showing that task capability alone does not guarantee safe execution. More importantly, even explicit recognition of risks does not reliably prevent harmful actions, revealing a critical gap between risk awareness and safe execution. These findings highlight the need to move beyond risk recognition toward enforcing safety throughout execution.

The carbon footprint of generative-AI inference depends not only on hardware efficiency and electricity supply, but also on whether the selected model is appropriate for the task. A frontier model may expend unnecessary computation on a routine request, whereas an under-capable model may increase energy use through failure, retries, extended reasoning and escalation. However, a standardized evaluation framework that accounts for task-specific capability requirements and the carbon emissions associated with successful task remains lacking. In this Perspective, we argue for a capability-constrained carbon labeling framework that shifts environmental assessment from carbon intensity per inference towards the expected carbon emissions per successfully completed task at the required capability level. Under this framework, models would first be assessed against task-specific capability requirements, and only models capable of meeting those requirements would be compared on their expected energy use and operational carbon emissions. By linking service quality to carbon cost, this approach could enable more informed matching of the right model to the right task, offering a demand-side pathway to reducing emissions associated with generative-AI token use. Realizing this vision will require standardized capability-based service assessments, comprehensive energy accounting and transparent carbon disclosures.