Peer Revenue Returns subject to peer review

Papers

Fresh from arXiv, typeset properly. Highlighted words trade on the market.

All of arXiv

Tuesday 1 September

Analysis of PDEs

Uniqueness of Finite-Time Varifold Limits for the Möbius-Invariant Willmore Flow

We prove uniqueness of finite-time geometric endpoints for the Möbius-invariant Willmore flow in $\mathbb{S}^3$ under uniform quantitative nonumbilicity. The multiplicity-counting varifolds converge, without reparametrization or Möbius renormalization, to a unique integral two-varifold. An intrinsic transport estimate gives quantitative total-variation convergence of the induced area measures on the fixed domain and bounded-Lipschitz Cauchy control of their pushforwards. Together with Allard compactness and rectifiability, this upgrades subsequential compactness to full-trajectory varifold convergence. The limit has generalized Euclidean mean curvature in $L^2$ with the natural endpoint lower-semicontinuity bound. For finite maximal trajectories with initial energy at most $8\pi$, Jakob's subsequential alternative becomes sequence independent: the limit is zero, or it has unit density and embedded Lipschitz support of genus zero or one. For Hopf-torus trajectories under the same energy bound, nonumbilicity is automatic and the anchored constant-speed profiles converge weakly in $W^{2,2}$ and strongly in $W^{1,2}$ and $C^{1,\alpha}$ for every $\alpha<\frac{1}{2}$. At infinite time, the same method yields a unique limit under an additional finite-dissipation-length condition.
Read the full abstract

Human-Computer Interaction

UniScale: Exploring Unimanual Gesture Mapping Strategies for Gaze+Pinch-based Scaling Interaction

Object scaling serves as a fundamental spatial manipulation that enables complex and productive tasks in XR environments. This paper investigates unimanual scaling techniques for XR using gaze and hand interactions. We propose UniScale, a set of unimanual alternatives to the standard bimanual pinch, allowing users to scale objects while preserving hand availability for concurrent spatial manipulations. We design five distinct mapping strategies based on physical metaphors, exploring unimanual control that varies depth, angle, micro-gestures, and finger-distance input. We then compare these techniques against a standard bimanual baseline, in which users adjust the inter-hand distance via a bimanual pinch gesture. In a user study, we evaluate their effectiveness in a 3D object scaling task under both clutching and clutching-free conditions. The results indicate that while bimanual scaling relies on clutching for stable control, unimanual techniques excel in clutching-free conditions, significantly reducing physical hand movement. From the results, we derive valuable design implications for developing efficient 3D multimodal interactions in XR.
Read the full abstract

Monday 31 August

Machine Learning

Not All Data Leakage Is Equal: A Controlled Benchmark of Evaluation Inflation in Tabular Classification

Data leakage is widely recognized as a threat to trustworthy machine-learning evaluation, yet the phrase covers mechanisms with very different statistical consequences. We present a controlled empirical benchmark that measures the inflation in held-out ROC-AUC caused by five common leakage mechanisms: global standardization, global median imputation, duplicate contamination, supervised feature selection before splitting, and target encoding before splitting. The benchmark spans six public datasets and two synthetic stress tests, three model families, and five repeated stratified train/test splits, yielding 600 paired clean-versus-leaked comparisons. To avoid pseudo-replication, inference is performed on dataset-level effects with cluster bootstrap confidence intervals and Holm-corrected one-sided Wilcoxon tests. Label-dependent or direct data-reuse mechanisms produced systematic optimism: global target encoding increased ROC-AUC by 2.42 percentage points on average (95% cluster-bootstrap CI 0.62–4.89), global supervised feature selection by 1.93 points (0.63–3.39), and duplicate contamination by 1.52 points (0.40–2.80). All three remained significant after multiplicity correction. In contrast, global scaling and global median imputation had near-zero average effects under the i.i.d. settings studied here. A controlled target-encoding stress test further shows that leakage becomes sharply worse as categories become sparse. The results support a severity-aware view of leakage: evaluation pipelines should prevent all held-out-data dependence, but label-dependent transformations and observation reuse deserve especially aggressive auditing.
Read the full abstract

Artificial Intelligence

Does the Ranking Hold? A Practitioner-Facing Audit of Hallucination Detectors on Small, Quantized, On-Device LLMs

Hallucination detection for large language models is dominated by methods evaluated on 7B-70B research models with abundant sampling budgets and exact-match or ROUGE correctness labels. The small models practitioners actually ship on-device, Phi-3-mini, Llama-3.2-3B-Instruct, Qwen2.5-3B-Instruct, and Gemma-2-2B-Instruct, typically at 4-bit precision, sit outside this evidence base. We present a reproducible one-GPU audit of four detectors spanning three paradigms: length-normalized token-margin calibration, a lexical n-gram baseline, and two sampling-based consistency methods (semantic entropy and SelfCheckGPT-NLI) on two small open instruction-tuned models (Llama-3.2-3B-Instruct and Qwen2.5-3B-Instruct) loaded at 4-bit NF4 quantization on a single Colab T4 GPU, evaluated on closed-book TriviaQA. Cross-model ranking transfer is poor: Spearman rho is -0.80 and Kendall tau-b is -0.667, meaning the detector ranking actively inverts between the two models, and the cheap lexical baseline is statistically indistinguishable from the sampling-based consistency detectors on both models under bootstrap 95 percent confidence intervals. As a second, methodological contribution, we show that the standard token-overlap-to-multiple-choice correctness labeler for TruthfulQA-Generation produces near-chance AUROC (0.46-0.59) for all four detectors on both models, a labeler failure, not a detector failure, since the same detectors reach 0.69-0.89 AUROC on TriviaQA under the identical pipeline. We release the harness, prompts, and per-item scores and frame the detector choice explicitly as a fairness-of-error-rates allocation problem for practitioners deploying on resource-constrained, on-device hardware.
Read the full abstract

Human-Computer Interaction

Puzzles to Paradigms: A Survey of Human–Agent Collaboration in Scientific Innovation

Agentic AI systems are increasingly embedded in scientific workflows, including literature synthesis, hypothesis generation, tool use, writing, and peer review. However, existing surveys and evaluations often emphasize task performance while underexamining how humans and agents coordinate under different epistemic conditions of scientific work. This survey introduces CSCW$\times$Kuhn, a coordination-aware framework that connects computer-supported cooperative work dimensions with distinct scientific contexts. Using this framework, we synthesize recent human–agent systems and benchmarks to examine how coordination demands and failure risks may vary across contexts. Our synthesis identifies five recurring failure modes: procedural drift, grounding substitution, evaluative flattening, proxy optimization, and epistemic black-boxing, and argues for evaluations that move beyond task accuracy and toward grounding, provenance, disagreement preservation, and long-horizon scientific reliability. Our corpus and coding artifact are available at https://github.com/jenlia024/CSCWxKuhn-artifact.
Read the full abstract

All authors Zhenni Liang, Haoxuan Zhang, Haihua Chen, Ting Xiao, Junhua Ding, Yang Zhang

Econometrics

Electoral Intelligence: A Neural Zero-Inflated Negative Binomial Model for Forecasting Swedish Municipal Elections

Forecasting local elections is difficult because electoral competition is geographically heterogeneous, highly overdispersed, and characterized by many party-municipality combinations with little or no electoral activity. This article develops a bounded neural zero-inflated negative binomial (ZINB) model for forecasting party-level vote counts in Swedish municipal elections. The model combines a count component for expected votes, a zero-inflation component for structural electoral absence, municipality-specific dispersion, and learned embeddings capturing latent party, municipality, and party-municipality structure. Model training relies on data from the 2014, 2018, and 2022 Swedish municipal elections and performance is evaluated on a held-out set of municipalities. Testing results show strong predictive accuracy in both vote counts and vote shares. Explanatory diagnostics indicate that latent party structure, local party-municipality relationships, electoral persistence, organizational capacity, and candidate-electorate representational similarity are among the most important determinants of predictive performance. Following model evaluation and diagnostics, the selected specification is refit on all historical data and applied to generate forecasts for the 2026 municipal elections. The proposed framework provides a flexible approach for studying local electoral competition in multiparty systems characterized by heterogeneous party presence and substantial geographic variation.
Read the full abstract

All authors Haodong Qi, Stefano M. Iacus, Pieter Bevelander, Derek S. Hutcheson

Operating Systems

Streaming OS Installation from Compressed Partitioned Images on Memory-Constrained Bare-Metal Hosts

Bare-metal server provisioning systems typically decompress OS images into RAM before writing them to disk. When servers have insufficient RAM to hold the decompressed image (increasingly common as OS images grow due to AI/ML toolkits, GPU drivers, and firmware stacks), provisioning fails entirely. This paper presents a streaming installation technique that combines partial decompression for metadata extraction with pipeline-based disk writing, enabling OS installation on hosts where available RAM is less than the decompressed image size. The method decompresses only the first 30MB of a compressed image to extract the GPT partition table, calculates the target filesystem offset, then streams the full image through a kernel pipe buffer (64KB) directly to disk at the calculated offset, never storing the decompressed data in memory or on disk. Deployed across a production fleet of memory-constrained bare-metal servers, the technique achieves 100% provisioning success where the standard approach fails entirely.
Read the full abstract

Group Theory

Detecting Cohomological Dimension Three in Free-by-Demuškin Pro-$p$ Groups

Let \[1\longrightarrow N\longrightarrow G\longrightarrow D\longrightarrow1\] be exact, with $N\neq1$ free pro-$p$ of arbitrary rank and $D$ Demuškin. Writing $W=N/\Phi(N)$, we prove \[H^3(G,\mathbb{F}_p)^\vee\simeq W^D,\qquad \textrm{cd}_p G=3\text{ if and only if }W^D\neq0.\] Thus dimension three is detected on the first Frattini layer of the free kernel. Finite-rank kernels always satisfy the criterion, so a dimension drop requires an infinite-rank kernel, while the criterion remains stable under passage to open subgroups. On the fixed-point-free locus $W^D=0$, the remaining dimension-one/two alternative is governed by $H^1(D,W^\vee)$ together with the Hochschild–Serre transgression. Dualizing the latter produces a relation-defect class in the topological coinvariants of $W$, which admits an explicit description in terms of a lifted defining relation of the Demuškin quotient. This yields a complete criterion for the lower-dimensional branch and separates the top-degree fixed-point obstruction from the extension data controlling dimensions one and two.
Read the full abstract

2610.00021 · 31 Aug · Question 10.1 from the previous version has been essentially resolved by an argument privately communicated by Henrique Souza; the resulting statement now appears as Theorem 5.5. The author gratefully acknowledges this contribution

Quantum Physics

Attosecond topological interference beyond Floquet-Volkov paths

Combining attosecond science with semiconductor physics provides a promising interdisciplinary platform for the attosecond control of semiconducting properties. In each field, the reconstruction of attosecond beating by interference of two-photon transitions (RABBIT) and the time-resolved and angle-resolved photoemission spectroscopy (tr-ARPES) have been central experimental techniques for probing microscopic dynamics of electrons. However, the integration of the two techniques, bridging the two fields, has been missing so far although they have many technical similarities. Here, we develop a theoretical framework that integrates the two techniques based on the second-order time-dependent perturbation theory. Within this framework, we investigate the RABBIT spectroscopy in semiconductors focusing on the recent hallmark of tr-ARPES, namely quantum path interference between Floquet and Volkov states. Moreover, we theoretically identify a novel quantum path incorporating the virtual excitation, which substantially interferes with the Floquet and Volkov states in both the RABBIT spectroscopy and tr-ARPES of semiconductors. Importantly, we find that such virtual excitation channel induces an attosecond topological interference with the Floquet-Volkov paths, encoding topological quantities such as the local Berry curvature into the time shift of RABBIT sidebands, which is directly related to the photoelectron emission delay. Our findings establish a theoretical framework connecting attosecond spectroscopy to tr-ARPES, paving the way for future applications of attosecond science in semiconductors.
Read the full abstract

Computation and Language

PPTBench: Can Coding Agents Reconstruct the Visual World through Structured, Editable Slides

Coding agents are increasingly moving beyond text-based software tasks to reconstruct visual targets through code. This capability, visual coding, requires agents to translate their understanding of visual targets into executable code. Slides provide a natural testbed for this capability, combining rich visual structure with objects that can be programmatically created and edited. To measure this capability, we introduce PPTBench, a benchmark for reconstructing scientific flow diagrams as editable PowerPoint slides. PPTBench covers 500 scientific flow-diagram tasks across 10 presentation domains, drawn from real research papers, and is evaluated with a four-stage agentic judge covering artifact validity, process and connector fidelity, rendering quality, and visual fidelity. Evaluation of ten models across 36 model–harness–effort configurations shows a substantial gap in reliable visual coding: the best configuration achieves 77.34, while the median across configurations is 24.38. Fine-grained analysis shows that agents can generally produce valid slide files, but struggle to produce high-quality reconstructions that faithfully recover the semantics and visual structure of the target. Further analysis shows that increasing reasoning effort primarily improves hard-gate passage rather than mean detail quality on each configuration's passing tasks, while configurations with more inspection tend to achieve higher overall scores. PPTBench establishes a measurable testbed for studying visual coding and advancing agents toward more reliable visual creation.
Read the full abstract

All authors Xiaoqiu Wang, Yizhe Chi, Wenyi Li, Deyao Hong, Zhihan Shan, Mingju Gao, Kaisen Yang, Youjie Zheng, Calvin Xiao, Qinhuai Na

Computer Vision

TopoFuse: Topology-Aware Tri-Planar Fusion for 3D Cryo-Electron Tomography Segmentation

Automated segmentation of cryo-electron tomograms routinely produces masks that are voxel-accurate but topologically broken: membranes fragment, organelles merge into one another, and enclosed cavities collapse. Existing topology-aware losses reduce these violations but cannot eliminate them, because topology is encouraged through gradient pressure rather than structurally enforced. We introduce TopoFuse, which reframes topology as a differentiable projection operator rather than a loss penalty. At each forward pass, the projection operator $\mathrm{Proj}_T$ (a PH-guided sparse edit) identifies the critical voxels responsible for topological violations via bottleneck matching and applies sparse edits to satisfy a specified topology target (diagram feature counts and lifetime budgets) for dimensions $d \in \{0,2\}$. If the projection converges, the output satisfies those constraints on the downsampled grid ($s=2$); when it does not, a repair certificate exposes this explicitly, enabling downstream filtering. A topology prior head predicts the correction target directly from input features, removing any dependence on ground-truth topology at inference. Across three cryo-ET benchmarks, TopoFuse reduces Betti number error by 54% over the strongest soft-loss baseline ($p < 0.001$), improves Dice by 4.6 points, and edits only 3.1% of voxels to achieve this.
Read the full abstract

All authors Rohit Kumar Salla, Neelesh Gupta, Xingjian Li, Min Xu

Metric Geometry

Hadwiger's Theorem on hyperbolic space

We present a proof classifying continuous isometry invariant valuations on polytopes in hyperbolic space $\mathbb{H}^n$ as linear combinations of the intrinsic volumes. The argument relies on an adaptation of the spherical case recently established by the author that circumvents an approximation step which does not generalize to the hyperbolic setting.
Read the full abstract

Machine Learning

Revalidation Beats Stateful Routing for Scientific Surrogates Under Distribution Shift

Surrogate models are often chosen during development and then left in place as new measurements arrive. That practice becomes risky when noise, input support, or physical parameters change. We asked whether such changes call for a stateful adaptive controller, or whether it is enough to validate the candidate models again on each new batch. To study this question, we built RegimeShift-Surrogates, a reproducible streaming benchmark spanning eight analytic and dynamical tasks, four stationary or shifting regimes, ten held-out seeds, and eight classical, multilayer-perceptron, and Kolmogorov-Arnold network surrogates. The confirmatory run contains 30,720 model fits and 3,200 scored deployment windows. Choosing the model with the lowest validation loss in the current window yields mean log regret 0.091 against a per-window oracle; the best fixed model chosen in hindsight yields 0.192. The paired difference is -0.101 (hierarchical bootstrap 95% CI [-0.165, -0.040]; Holm-adjusted p = 0.0469), with revalidation ahead in 26 of 32 task-scenario combinations. None of the stateful alternatives, including exponential smoothing, dual-timescale adaptation, Page-Hinkley resets, or margin gating, improves the pooled result, and delayed bias correction makes it worse. Oracle choices also differ substantially by task: k-nearest neighbors dominate the damped oscillator, vanilla KAN is often selected for two-dimensional surfaces, and MLPs lead on the Runge and Van der Pol tasks. In this benchmark, fresh validation evidence is useful; carrying old evidence forward is often not.
Read the full abstract

Systems and Control

RAPTOR: RAndom-projection Physics-informed Transient sOlveR

The complexity of time-domain simulation of modern power systems has increased significantly because converter-based resources introduce control dynamics that must be simulated alongside slower system-level and fast electromagnetic dynamics. The resulting wide range of timescales may force classical time-domain solvers to use small timesteps, complicate the solution of nonlinear equations at each timestep, and reduce solver reliability under strongly nonlinear and multi-timescale transient conditions. This paper introduces RAPTOR, a first-of-its-kind time-domain simulation framework for power system dynamic simulations. At its core, RAPTOR introduces a new integration technique that represents the unknown trajectory of hybrid differential-algebraic equations (DAEs) over a time interval using a physics-informed random-projection neural network (PIRPNN) built from fixed Gaussian radial basis functions (RBFs). A nonlinear solver, e.g., Newton-Raphson, determines how to combine the fixed RBFs so that the resulting trajectory satisfies the governing equations. By combining RBFs with broad and localized shapes, RAPTOR can represent complex multi-timescale behavior over comparatively long time intervals. This enables larger effective simulation advances, fewer overall nonlinear solver iterations, and faster integration over long time horizons. Numerical studies on stiff RMS models, IEEE 9-, 14-, 39-, 57-, and 118-bus RMS simulation benchmarks, and EMT test cases show that RAPTOR accurately finds solutions with substantially fewer sequential simulation advances while offering superior accuracy-runtime performance. With speedups of over 10x in certain cases, these results showcase RAPTOR's potential to outperform long-standing and widely used integration methods, such as the Radau method and the trapezoidal rule.
Read the full abstract

All authors Petros Ellinas, Benjamin Vilmann, Spyros Chatzivasileiadis, Johanna Vorwerk

Software Engineering

DSpec2Test: Specification-Driven Test Generation in Dafny

Verification-aware languages, such as Dafny, integrate logical constructs into code and enable automatic verification of program correctness. However, tests remain helpful in scenarios that verification alone does not address (e.g., to support test-driven development). Existing Dafny test generation tools are implementation-based, limiting their applicability in this context. We present DSpec2Test, a specification-driven test generation tool for Dafny that automatically derives tests from formal specifications, without considering implementation details. Our tool extends Dafny's generate-tests command with a new blackbox mode based on Disjunctive Normal Form (DNF) equivalence class partitioning and optional Boundary Value Analysis (BVA). DSpec2Test relies on the Z3 SMT solver to synthesize inputs and expected outputs that meet the specification-derived constraints. We evaluate DSpec2Test on programs from DafnyBench mutated using MutDafny and compare it against Dafny's existing implementation-driven Block mode. DSpec2Test achieves a 93.9% mutation kill rate on a dataset of 131 mutants, outperforming Block's 82.4%, and uniquely killing 17 mutants. These results suggest that specification-driven testing is an effective and complementary approach for testing Dafny programs.
Read the full abstract

All authors Sofia Vieira Pinto, Álvaro F.Silva, João Pascoal Faria, Alexandra Mendes

Classical Analysis and ODEs

Form domination and tent space estimates for operators on the half-space

We study operators acting on functions defined on the half-space, with methods inspired by sparse domination. Using the specific link between dyadic grids on the boundary and Whitney regions on the half-space, we obtain a very precise form domination with model operators of Hardy type. We introduce mixed-norm off-diagonal estimates that measures both tangential and transversal decay without requiring pointwise control in the transversal variable, substantially weakening the assumptions used in earlier tent space theories. This allows us to prove optimal weighted tent space extrapolation from local boundedness plus off-diagonal decay. The resulting framework yields new bounds for operators arising from non-autonomous elliptic and parabolic PDEs with non-smooth coefficients.
Read the full abstract

All authors Pascal Auscher, Hedong Hou, Emiel Lorist, Andreas Rosén

Machine Learning

Decoupling Knowledge and Privacy: Post-Task Self-Distillation Replay for LLM Continual Learning

Privacy-preserving continual learning (PPCL) must reduce the reproduction of sensitive content while retaining useful knowledge across sequential tasks. Formal privacy guarantees characterize randomized mechanisms, whereas operational output control concerns whether a trained model selectively reduces the likelihood of sensitive content in its outputs. In this work, we investigate the latter together with continual-learning utility under realistic task evolution. Retention and privacy correction operate at different granularities: task acquisition requires broad preservation of current- and old-task behavior, whereas privacy correction targets sparse annotated positions. Joint optimization leaves the current-task preservation target continually changing. We propose SPARK, a retention-correction decomposition that first freezes the learned post-task distribution and then applies selective correction around this stable reference. Self-Distillation Replay learns the current task while distilling behavior from previous tasks, and Post-Task Privacy Correction reduces annotated-PII likelihood while anchoring current- and old-task non-PII behavior to the resulting checkpoint. Extensive evaluations demonstrate that SPARK achieves effective selective PII suppression while preserving strong continual-learning utility and knowledge retention across diverse settings. Code and data will be released upon publication.
Read the full abstract

All authors Shengtao Wen, Yunying Yang, Xiang Chen, Lingbing Guo, Yu Tian, Sheng-Jun Huang

Machine Learning

Three Ways Classical Test Theory Can Mislead About LLM Judges

Evaluations that use a large language model (LLM) as a judge have begun to borrow reliability statistics from classical test theory and its extensions. We examine three such statistics that need one administration and no gold labels. None of them can isolate the judge, because one judge under one prompt supplies no variance component of its own. Claude Haiku 4.5 judged 210 constructed short answers against ten-element checklists. On the 180 with parsed verdicts, the Kuder-Richardson coefficient (KR-20) came out at 0.5223 on the judge's verdicts and 0.5231 on error-free gold verdicts. In simulation, bank design alone moves KR-20 from 0.01 to 0.68 at the judge's measured 4.72% error rate. The dependability index $\Phi(\lambda)$, a ratio of mean squared distances from the pass mark, sits 0.22 to 0.38 below the judge's accuracy against gold and returns 0.54 to 0.68 on error-free gold verdicts. Livingston-Lewis accuracy treats the rubric elements as a sample, and at a pass mark of five elements it credits error-free gold scores with 0.78, close to the judge's 0.81. A statement about the judge therefore needs gold labels or a varied scorer facet, and a reliability ratio needs the bank's spread beside it. One of the four closest judge-evaluation papers varies the prompt and still reads a reliability below 0.7 as a sign that a model cannot serve as a judge, although that reliability moves with the spread of the samples scored. We derive a decision table and four reporting lines from these two rules.
Read the full abstract

2609.29709 · 31 Aug · 16 pages (7 of main text), 4 figures. v2 adds the gold-computed null for all three statistics and a decision table, corrects the reading of the Livingston-Lewis difference, adopts Brennan's estimator for Phi(lambda) and revises the appendix. Code and data: https://github.com/louisyzhu/llm-judge-reliability

Logic

Completeness of the Gödel–Löb Provability Logic for Finite Derived Ordinal Topologies

We study the Gödel–Löb provability logic $\mathbf{GL}$ for finite derived topologies on the ordinals. Assuming GCH and the existence of a measurable cardinal $\kappa$ which carries a coherent sequence of normal measures $\vec{U}$ with $o^{\vec{U}}(\kappa)\geq\kappa$, we obtain a cofinality and GCH preserving generic extension in which $\text{Log}_d(\kappa,\tau_n)=\mathbf{GL}$ for every $n<\omega$, where $\tau_n$ is the $n$-th derived topology. The result answers a well-know open question. Extending Beklemishev's theorem, we show that in the same model, for every $k<\omega$, $ \text{Log}_d(\kappa;\tau_{2k},\tau_{2k+1})=\mathbf{GLP}_2, $ where the two modalities of $\mathbf{GLP}_2$ are interpreted respectively by $d_{2k}$ and $d_{2k+1}$.
Read the full abstract

Distributed Computing

Where Does the Energy Go? Profiling LLM Agent Inference on Blackwell GPUs

LLM agents that iteratively reason, plan, and invoke tools create workload profiles fundamentally different from single-pass inference, yet how their energy consumption is distributed across hardware components and workload phases remains poorly understood. Characterizing these workloads therefore requires simultaneous visibility into both component-level power and phase-level execution. We conduct a full-stack energy profiling study combining NVML GPU counters, Intel RAPL CPU/DRAM counters, and Intelligent Platform Management Interface (IPMI) system-level sensors on 2x NVIDIA RTX PRO 6000 Blackwell GPUs, and profile three representative workloads with Qwen3.8-27B. For mathematical reasoning, we compare thinking-enabled and thinking-disabled modes. Our measurements reveal that GPU-only telemetry misses 41-45% of system energy across all three workloads, with non-GPU components accounting for the remainder. In our setup, the sequential agent workload consumes 63x more system energy per output token than saturated serving, reflecting the absence of batching, context growth across turns, and tool-induced idle periods. Extended thinking generates 21-75% more tokens per problem, while per-token energy differs by less than 1% between modes within each dataset, indicating that the resulting increase in energy is driven by output volume rather than a change in per-token efficiency. Continuous batching improves system-level energy efficiency by 3.2x from 1 to 16 requests per second, as GPU power plateaus while throughput continues to increase with batching depth. For the memory-bandwidth-bound reasoning workload, throughput remains unchanged under per-GPU power caps of 400-600 W but drops sharply at 300 W. These findings suggest that context-management techniques such as summarization and selective retrieval may help reduce energy consumption in agent deployments.
Read the full abstract

All authors Qi Luo, Kunlin Li, Ziwen Wang, Yun Chen

Machine Learning

Safety-oriented pedestrian trajectory prediction at urban intersections using time-to-collision and crossing-zone context

Accurate pedestrian trajectory prediction is important for proactive road-safety applications, particularly at urban intersections where pedestrian motion is shaped by both vehicle interactions and crossing context. This study presents a safety-oriented trajectory-prediction framework that combines pedestrian motion history with Time-to-Collision (TTC) information and crossing-zone indicators. Using naturalistic trajectories from one urban intersection in the inD (Intersection Drone) dataset, several neural architectures were evaluated with 1.6 s observation and 2.4 s prediction horizons. A pooled Long Short-Term Memory (LSTM) separately encodes TTC histories and crossing-zone context before integrating them with pedestrian positions. In addition to conventional Average Displacement Error (ADE) and Final Displacement Error (FDE), prediction performance was assessed using the frequency and magnitude of errors exceeding a study-defined 1 m tolerance. A weighted loss was also introduced to place greater training emphasis on large coordinate-wise errors. Applying this loss to the position-only LSTM reduced ADE from 0.210 to 0.190 m and FDE from 0.550 to 0.503 m, while reducing ADE and FDE exceedance counts by 34.8% and 19.8%, respectively. The final pooled configuration incorporating TTC and crossing-zone information achieved an ADE of 0.184 m and FDE of 0.491 m, with further reductions of 33.5% and 6.3% in ADE and FDE exceedance counts relative to the safety-oriented position-only LSTM. The results indicate that safety-oriented training and structured integration of interaction and contextual information can reduce large trajectory-prediction errors, although broader validation across pedestrians, sites, and datasets is required.
Read the full abstract

Software Engineering

On the Factors in Quantum Software Quality

As quantum computing evolves, the demand for high-quality quantum software is increasing. Classical software quality models, such as McCall's model, provide a useful foundation, but they do not explicitly capture distinctive properties of quantum software, including probabilistic measurement outcomes, backend-dependent behavior, and hybrid quantum-classical workflows. This paper proposes a base quality model for quantum software using McCall's factor-criterion-metric structure as the organizing backbone. It broadens the set of quality factors by drawing on representative classical software quality models and standards and introduces quantum-specific extensions only where needed. We define the resulting factor set and quality criteria, provide a many-to-many mapping of core factor-to-criterion relationships, discuss representative candidate metrics together with the execution, reference, and interpretation conditions needed to use them, and illustrate how the resulting model can be applied across the quantum software lifecycle. The model is intended as a common starting point for structuring, assessing, and communicating quality concerns in quantum software while supporting context-specific adaptation. Further expert feedback, case studies, and empirical studies can be used to evaluate and refine the model and its application across different quantum software contexts.
Read the full abstract

Cryptography and Security

Detect First, Explain Later: Training-Free Temporal-Memory Digital Twin Anomaly Detection with Post-Hoc LLM Interpretation for ICS

Industrial Control Systems (ICS) are increasingly exposed to cyber-physical attacks that manifest as subtle and temporally evolving deviations in process behavior. Detecting such anomalies requires reasoning over persistence, cross-signal dependencies, and process-level constraints. Digital Twins (DTs) encode system knowledge through physical and logical relationships between signals, but existing DT-based approaches rely on instantaneous rule violations and lack mechanisms to aggregate weak evidence over time. This paper proposes a training-free anomaly detection method that combines deterministic DT constraints with explicit temporal memory. The DT monitors process signals and produces anomaly scores based on constraint violations, while a lightweight memory mechanism captures persistence and contextual relationships across time. The approach is evaluated on the HAI and BATADAL datasets. Ablation results show that the memory-less detector fails completely, demonstrating that temporal aggregation is essential for DT-based detection. On HAI, the memory-aware DT achieves stable detection with only 4 false alarm events, and on BATADAL, it remains effective without retraining, with 21 false alarms under domain shift. In comparison, Isolation Forest (IF) produces substantially more false alarms (328 on HAI and 127 on BATADAL), while Autoencoder (AE) exhibits dataset-dependent behavior, achieving high precision on BATADAL but low recall and inconsistent performance overall. A gated LLM is used for post-hoc interpretation, providing structured explanations without affecting detection performance. Our findings highlight the importance of temporal memory in constraint-based detection and support the use of decoupled reasoning for interpretability in ICS monitoring.
Read the full abstract

Digital Libraries

Comparability in the public olfactory record: volatile measurement, human perception, and the join between them

Research data infrastructure measures metadata completeness, the share of fields populated. This is the wrong quantity: a populated field is not a joinable field. We measure comparability instead: whether two records can be used together, and for what. Working from 180,877 records across eight public sources, we adapt measurement invariance from psychometrics into a four-rung ladder, each rung established by its own test and licensing one specific comparison. Over all 167,331 pairs of gas chromatography analyses in a public metabolomics repository, 100% reach the bottom rung, 58.85% the second, 0.072% the third, and exactly two pairs reach the rung that licenses comparing values directly, an upper bound under a generous test. The deficit is recoverable: the missing values survive in prose. We formalise crosswalking as five operations ordered by reliability, each with a characteristic failure mode. The temperature programme that determines retention is structured in none of a second repository's 355 gas chromatography studies yet written in prose in 78.6%; we recover 206 ordered programmes comprising 529 steps, audited over three rounds. The perceptual record fails differently: its datasets do not share a language. Among eight that declare they measure odour character in humans, 42.9% of pairs share no descriptor, and two studies rating the same molecules with the same word agree at mean correlation 0.31. Joining volatility measurement to human percept across the entire public record yields 132 molecules. Across seismology, meteorology, metrology and chemistry, a field is populated when the primary consumer cannot complete the primary task without it: the same optional, unvalidated field sits at 100% in one discipline and near zero in another. Resources should publish a conformance ledger, reporting what is missing as a tracked quantity; we release one, with 48,356 triples.
Read the full abstract

Multiagent Systems

Multi-Agent Debate for Explainable Trading: Reasoning, Consensus, and Performance in Simulated Markets

Large language models (LLMs) are increasingly used for financial decision-making, yet it remains unclear whether improvements in reasoning quality translate into better economic outcomes. We investigate this question using a multi-agent debate framework for portfolio allocation in historical market simulations, where specialized agents propose, critique, and revise investment decisions. Reasoning quality is evaluated across four dimensions: logical validity, evidential support, alternative consideration, and causal alignment, and compared with downstream financial performance. Across 210 controlled runs, aggregate reasoning quality shows no meaningful relationship with Sharpe ratio (r = 0.07, p = 0.29) or total return (r = 0.03, p = 0.70). Structured prompting increases measured reasoning quality from about 0.72 to 0.84 (+17.7%, Cohen's d about 2.0), but these gains do not consistently translate into higher returns. We identify sycophantic convergence as a central failure mode, where agents abandon independent positions during critique-revision cycles and converge toward similar allocations. A Jensen-Shannon divergence intervention that preserves disagreement improves Sharpe by +0.14 (p = 0.028) and Sortino by +0.25 (p = 0.026), while interventions enforcing stronger causal reasoning do not improve financial performance. Our results suggest that multi-agent debate is most valuable when it preserves independent informational signals rather than simply improving measured reasoning quality.
Read the full abstract

All authors Juli Huang, Alanood Alrassan, Deveen Harischandra, Theodore Wu, Veljko Skarich, Matthew Hayes

Risk Management

Storage Options and Endogenous Commodity Prices in Continuous Time

We study the price formation of a storable commodity when the decision to sell or keep the commodity is treated as an embedded storage option. The price process is not imposed exogenously. Instead, a candidate price function determines the demand dynamics, while the optimal stopping value generated by those dynamics produces a new price function. We call a fixed point of this feedback mechanism an endogenous storage-option equilibrium (ESOE), using a separate acronym to distinguish the continuous-time construction studied here from the stationary rational expectations equilibrium terminology of the classical competitive-storage literature. A global existence result is established on a compact admissible class under a transparent drift–discount condition. The argument avoids the interval-by-interval extension of local fixed points and does not claim uniqueness from Schauder's theorem. We then characterize the equilibrium as a nonlinear obstacle problem and develop a monotone finite-difference/policy-iteration method with an asymptotically consistent far-field boundary condition. The numerical analysis separates fixed-point convergence from spatial discretization error and is supplemented by Monte Carlo validation, scaling diagnostics, comparative statics, and the distribution of optimal selling times. The results clarify how demand elasticity, discounting, depreciation, and volatility jointly determine the storage premium and the selling threshold.
Read the full abstract

Probability

Critical-curve regularity for finite-lifespan frog models via local-to-global comparisons

We study the phase boundary of the finite-lifespan frog model. For rate-one continuous-time simple random walk on an infinite, connected, locally finite transitive graph of superlinear growth, we prove that the critical-density curve is continuous and strictly decreasing, with $-\log\lambda_c$ locally bi-Lipschitz. The inverse critical-lifespan curve has the corresponding regularity wherever it is finite. Thus the critical-curve conjecture of Angel, de la Riva, Hermon, and Shi reduces to finiteness of the critical lifespan at every positive density. We identify the exact short-lifespan limit $t\lambda_c(t)\to\Delta\log(1/(1-p_c(G)))$, where $\Delta$ is the degree and $p_c(G)$ is the bond-percolation threshold, and extend critical-curve regularity and sharpness to a class of long-range frog models. The main tool is a local-to-global principle for activation processes generated by independent finite rooted ranges: local one-hit comparison implies comparison of global reachability and survival. The comparison allows disconnected ranges and source-dependent laws, and therefore also applies to activation models beyond random walks on a fixed graph.
Read the full abstract

Differential Geometry

A two-step ideal sub-Riemannian structure with no measure contraction properties

We construct a compact, equiregular, ideal sub-Riemannian manifold of step $2$ such that, for every smooth positive measure, the $\mathrm{MCP}(K,N)$ fails for all $K\in\mathbb{R}$ and $N \in (1,\infty)$. This shows that the real-analyticity assumption in the measure contraction theorem of Badreddine and Rifford in arXiv:1712.09900v2 cannot be replaced by smoothness. Moreover, our structure is ideal, i.e. it admits no non-trivial abnormal minimizing geodesics. Although failures of the measure contraction property for ideal structures were recently obtained in the higher-step setting, our construction shows that the phenomenon can already occur on compact, equiregular, ideal structures of step $2$. The proof exploits a differential consequence of the measure contraction property, namely a uniform upper bound for the sub-Laplacian of the squared distance near its base point, and constructs a structure for which this quantity is unbounded.
Read the full abstract

Cryptography and Security

Understanding and Exploiting Initialization Anchoring Weakness in Feedback-Based Agent Planning

Feedback-based planning improves agent reliability by incorporating tool observations and corrective feedback. However, its protection may not be distributed uniformly across planning stages. We conduct a round-wise analysis of four representative feedback mechanisms and uncover an initialization anchoring weakness: the first feedback round corrects 46% of adversarial directions, whereas the rates fall to 13% and 7% among directions surviving into the next two rounds. Our analysis attributes this weakness to three interacting factors: a contextually plausible shift in the initial plan, insufficient counterevidence, and the persistence of accepted directions in the accumulated trajectory. Based on these findings, we propose InitAnchor, a black-box framework for exploiting this weakness through attacker-controlled external materials. It operationalizes the three factors as directional-shift, contextual-plausibility, and counterevidence-resilience signals under either limited target access or no target access. Across 112 tasks from 16 domains, six agent architectures, and five backbone LLMs, InitAnchor achieves average ASRs of 76.1% and 72.0% under the two settings while reducing first-round mitigation rates to 21.0% and 25.0%, respectively. It also remains effective against six defenses and across six real-world agent systems. These findings show that feedback-based agents can retain early biases even when later correction is available.
Read the full abstract

All authors Chuanchao Zang, Jianing Wang, Wenyu Chen, Xiangtao Meng, Li Wang, Xinyu Gao, Peng Zhan, Zheng Li, Shanqing Guo

Machine Learning

An Agnostic Sample Compression Scheme for Squared Loss of Near-Linear Size in the Fat-Shattering Dimension

We study the squared-loss sample compression question posed by Attias, Hanneke, Kontorovich, and Sadigurschi in their ICML 2024 paper. For every nonempty class $\mathcal{F}\subseteq[0,1]^{\mathcal{X}}$ and accuracy $0<\alpha\le1$ with $d=\operatorname{fat}(\mathcal{F},c'\alpha)<\infty$, where $c'>0$ is universal, we prove the existence of a deterministic agnostic scheme of total size $O(d\log^3(2/\alpha))$. Its code contains original labeled observations and auxiliary bits, and its reconstruction has empirical loss at most $\alpha$ above the infimum over $\mathcal{F}$ on every finite sample. This establishes the polylogarithmic formulation in Section 5 of arXiv:1810.01864v2. The construction approximates a near-optimal internal target by a fixed-step boosting ensemble and takes its equal-weight median. Average loss permits an accuracy-dependent number of rounds, so the entire ensemble can be retained without dual-dimension sparsification. Quantized target constraints and round-membership bits determine the reconstructed function exactly. We also prove that, when retained observations must come from the sample, a fixed class of fat-shattering dimension one requires total size $\Omega(\log(1/\alpha))$. With observations and auxiliary bits charged together, this rules out the proceedings version's log-free formulation under its original-example restriction. The framework extends to average power loss at scale $\alpha/p$.
Read the full abstract