Software Engineering
Papers
Fresh from arXiv, typeset properly. Highlighted words trade on the market.
All of arXiv
Friday 4 September
Automotive infotainment validation still relies on manual testing, slow, costly, and incompatible with agile releases and OTA updates. Scripted automation only partly helps: it couples test logic to implementation, yielding brittle, high-maintenance suites. Existing LLM-driven frameworks mostly target web/mobile apps, using single- or dual-agent setups that overload one or two models with perception, planning, action selection, and validation at once, prone to hallucinations and unproductive exploration loops given infotainment complexity. We present ARIA (Autonomous Real-time Infotainment Assessment), a multi-agent LLM framework that autonomously runs end-to-end tests on Android infotainment systems via visual interaction, using a closed-loop pipeline of four specialized agents per step plus a report stage. From single-sentence scenarios (path, action, expected outcome), ARIA runs the interactions and produces reports, reproducible scripts, and visual evidence per step. Evaluated on a manufacturer's physical Android infotainment system across 30 scenarios, ARIA completed 28 (93.3%) with a verdict (2 errored), 20 of which (71.4%) matched ground truth. It caught all 5 known defects, no fault passed as working; its 8 false positives stem from navigation/image limits and unsupported gestures, showing multi-agent LLMs can run infotainment tests industrially while exposing the cost of a low false-positive tolerance. A single-agent baseline confirms the multi-agent design's value: on the first pass, before stronger-model revisitation narrows the gap, it shows a far higher false-positive rate (72.0% vs. 52.6%), conflating navigational difficulty with system failure. We report first-pass/post-revisitation results, token/call/cost per scenario, and show via repeated runs that stability tracks complexity, with fault detection perfectly consistent, pointing to CI integration of visual testing.
Hardware Architecture
TreeFI: Value-Aware Statistical Fault Injection for Deep Neural Networks
Reliability evaluation of deep neural networks under hardware faults commonly relies on fault injection, but exhaustive campaigns are intractable for modern models and datasets. Statistical fault injection reduces this cost, yet existing approaches still require large injection budgets because they do not explicitly exploit a key property of floating-point faults: the effect of a bit flip depends strongly on the value being corrupted. We propose TreeFI, a value-aware statistical fault-injection methodology for FP32 single-bit faults in DNN activations and weights. TreeFI partitions each layer's value distribution into intervals with similar expected bit-flip behavior, learned using regression trees, and allocates injections across these intervals according to their relevance for failure-rate estimation. This stratified allocation preserves the target confidence and error margin while avoiding unnecessary injections in low-impact regions of the fault space. We validate TreeFI on CNN and Transformer models using CIFAR-10 and ImageNet. On ResNet8, where exhaustive activation fault injection is feasible, TreeFI provides more accurate estimates than state-of-the-art statistical FI baselines under the same campaign setting. Across the evaluated models, TreeFI reduces the required injection budget by up to 72.1x, with average reductions of 44.9x for activation faults and 11.2x for the executed weight campaigns.
Read the full abstract
Computer Vision
TourPhysics: Bringing Physics to World Models for Exploration and Manipulation from a Single Image
Interactive visual world models must distinguish observation from physical intervention. Camera motion reveals new surfaces, whereas intervention changes object motion, contact, and deformation. Current video world models are largely driven by appearance priors and often lose physical or spatial consistency over long horizons. We present TourPhysics, an online framework initialized from a single image and a declarative physical configuration. TourPhysics extends PhysOmni, our ACM Multimedia 2026 work, from finite physics-grounded video synthesis to persistent exploration and manipulation. TourPhysics combines deterministic simulation with video generation while assigning separate roles to simulator state, geometric evidence, generator controls, and appearance memory. For each action, the simulator computes a finite physical and camera trajectory before the corresponding observation is generated. Accepted observations publish the terminal state and update the appearance memory and subsequent generator controls, while the committed state and simulator geometry remain fixed throughout synthesis and retry. We further separate the simulator geometry used for projection and visibility from the relative depth used to condition the generator. A reference-anchored memory retrieves accepted static appearance through geometric cross-view correspondence and incorporates it through a bounded residual that reverts to the native path when no valid correspondence exists. On simulator-defined camera tours and object manipulations, TourPhysics follows prescribed camera and object trajectories more closely than the evaluated baselines, preserves the input scene, and reduces appearance drift during long-horizon revisits.
Read the full abstract
Machine Learning
Fast Gauss Sums via Flash Attention
Gaussian kernel sums are the computational core of maximum mean discrepancies (MMDs), kernel gradient flows, Stein variational gradient descent (SVGD), and many other kernel methods. At the same time, softmax attention has received an extraordinary amount of hardware-aware code engineering, culminating in flash attention. We show that Gauss kernel sums with arbitrary, signed weights can be evaluated via flash attention: two small input augmentations turn the normalized softmax reduction into the unnormalized Gauss sum, without writing a single line of custom GPU code. For feature dimension D>8 in fp16, this approach beats compiled PyTorch code as well as PyKeOps kernels (often significantly) in speed, memory-overhead and accuracy. Indeed, its memory scaling remains linear.
Read the full abstract
Software Engineering
Better Understanding, Better Fixes? A Study of Hallucination in LLM-based Automated Program Repair
Large language models (LLMs) have significantly advanced automated program repair (APR), yet existing evaluations remain largely result-centric and provide limited insight into hallucination during repair. In APR, hallucination may arise not only in final patches but also in the intermediate artifacts that guide patch generation. To address this gap, we perform a multi-layered analysis of hallucination throughout the APR process. Specifically, we characterize hallucination as the production of patches or intermediate artifacts that are not faithfully grounded in the available repair evidence. We examine repair hallucination in final patches and understanding hallucination in intermediate artifacts through three tasks, namely triggering testcase identification, line coverage prediction, and additional testcase generation. We then evaluate three representative LLMs on 832 Defects4J bugs through automatic evaluation and manual analysis. Our results show that both repair and understanding hallucinations remain prevalent. Across models and settings, only 21.0%-55.9% of generated patches pass the developer-written test suite. Moreover, although more accurate intermediate artifacts are generally associated with successful repairs, this relationship does not always hold. Manual analysis of 812 sampled repairs identifies repair hallucinations in 72.7% of cases, including patches that pass all available tests; incorrect causal localization and incorrect repair strategies account for 45.9% and 18.5% of these hallucinations, respectively. Meanwhile, models frequently misidentify triggering testcases, mispredict line coverage involving branching control flow, and generate additional testcases with missing bug-triggering conditions or incorrect expected behavior.
Read the full abstract
General Mathematics
Construction of Finite Hilbert–Pólya Matrices from Weil's Explicit Formula
Starting from the Riemann–$\Xi$ specialization of Weil's explicit formula, we construct finite real-symmetric Prime–Weil matrices $(\mathbf S)$ from pole, archimedean, and finite prime-power data. This construction is a finite-dimensional arithmetic model within the Hilbert–Pólya program, which seeks a self-adjoint spectral realization of the nontrivial zeta-zero parameters. Their off-diagonal entries form a Loewner-type divided-difference matrix with a rank-two displacement identity. We formulate the spectral quotient as a Hermitian definite generalized eigenproblem on the fixed zero-mean contrast space. This realization is invariant under positive affine rescaling, avoids ground-vector normalization and an ill-conditioned oblique projector, and preserves the finite quotient spectrum. For a dimension-matched zero-side matrix built from $N$ distinct positive ordinates $\gamma_k$, rational interpolation gives the exact contrast-pencil spectrum $\{\pm\gamma_1,\ldots,\pm\gamma_N\}$ and positive-parity square spectrum $\{\gamma_1^2,\ldots,\gamma_N^2\}$. Since the ordinates are inputs, this is a reconstruction theorem. Assuming RH, the same interpolation vector proves $\lambda_{\min}(\mathbf S)\to0$. At $N=L=13$, Lemke's ground-state quotient and the contrast pencil agree numerically and reproduce the first three zeta ordinates to the reported precision. The remaining problem is a relative prime-to-zero perturbation theorem with uniform control of the compressed metric. No proof of RH is claimed.
Read the full abstract
Quantum Physics
Microkelvin resolution thermometry at the nanometre scale
Accurate temperature readings of transient events at the nanometer scale are challenging due to the low sensitivity of available sensors. Nanodiamonds containing nitrogen-vacancy (NV) centers have been used for nanoscale thermometry in complex environments, including inside living cells. However, their performance has been limited by short coherence times and low photon counts. In this work, we use isotopically-purified dual-NV nanodiamonds and a bespoke quantum sensing chip to showcase an order of magnitude improvement in temperature measurement sensitivity compared with previous reports. We demonstrate robust temperature measurements with an error of 682 $\mu$K, experimental sensitivities below 50 mK/$\surd \text{Hz}$ and a shot-noise limited sensitivity of 9.6 mK/$\surd \text{Hz}$. To confirm the utility of these high-performance nanothermometers, we quantify the temperature change induced by the thermometry measurement itself, specifically the optical excitation laser used to probe the NV spin state. In addition, we observe directly at the nanometre scale the transient heating caused by the exothermic mixing of dimethyl sulfoxide in water. Sub-millikelvin resolution and millikelvin sensitivity thermometry unlock the possibility of monitoring minute thermal fluctuations in living systems and assessing catalyst performance at the nanometre scale.
Read the full abstract
Computer Vision
Methane Detection On Board Satellites from Unorthorectified Imagery
As a potent greenhouse gas, methane is a major driver of climate change. Its effective mitigation relies on timely detection. Conventional detection methods rely on orthorectification to correct geometric distortions and matched filters to enhance plume signals, which are steps designed for ground processing and poorly suited to onboard execution. We introduce UnorthoDOS, a dataset and approach for training machine learning models directly on unorthorectified hyperspectral imagery, bypassing both orthorectification and matched-filter products. Our U-Net models trained on unorthorectified data approach the performance of models trained on orthorectified data (IoU 16.91% vs. 18.47% on all plumes), while both substantially outperform the mag1c matched-filter baseline (IoU 4.76%). We further demonstrate the feasibility of onboard deployment: FP16 compression halves model size with under 0.3% output deviation. The trained ML models and two ML-ready datasets – orthorectified and unorthorectified hyperspectral imagery from the EMIT sensor – are publicly available at https://huggingface.co/datasets/SpaceML/UnorthoDOS, with code at https://github.com/spaceml-org/plume-hunter.
Read the full abstract
Quantum Physics
Ancilla mediated steady-state engineering in open quantum systems
Engineering the properties of a reservoir and its coupling to a quantum system is a powerful tool for simulating quantum thermodynamic processes and for generating otherwise inaccessible steady states. Yet tailoring both the reservoir and its coupling within a single platform remains challenging. Here we introduce a platform, in which an ancilla qubit mediates the coupling of a target system to a reservoir, providing independent control over the interaction form, coupling strength, and effective reservoir temperature. Our implementation uses the electron spin of a single nitrogen-vacancy center in diamond as the ancilla and a proximal $^{13}$C nuclear spin as the target. By alternating engineered unitary interactions with dissipative ancilla resets, we realize dynamics naturally described by a collision model, enabling straight-forward tracking of the work, heat, coherence, and entropy generated at every collision. We experimentally demonstrate conventional thermalization and also realize anti-thermalization: the stabilization of the target system in a temperature opposite to that of its reservoir. Finally, harnessing this steady-state engineering, we utilize the nuclear spin as a quantum battery, achieving a steady-state ergotropy exceeding $70\%$ of the theoretical maximum.
Read the full abstract
Programming Languages
CPL: A Compact C-like Systems Language with Explicit Low-Level Control
This paper presents Cordell Programming Language (CPL), a compact C-like systems language that retains C's direct access to memory, layout, and machine interfaces while experimenting with a smaller grammar and selected conveniences from newer languages. Also this paper studies whether C-like are more convenient to use for compiler experiments than modern approaches and paradigms. While the language and compiler provide primitive values, pointers, arrays, containers with methods, unions, generic functions, overloads, entry-point and section control, system calls, and inline assembly, they do not provide high-level constructs such as classes, built-in methods, a standard library, or memory protection. The article describes the language design, compiler pipeline, target backends, static-analysis architecture, and OS-facing use cases, then evaluates the prototype backend with reproducible x86_64 and i386 microbenchmarks against C compiler baselines. The obtained results suggest that the compiler can produce code comparable to that produced by production compilers such as GCC and Clang, as well as by small compilers such as TinyC and SmallerC.
Read the full abstract
Computer Vision
InterSing: Explicit Interaction Dynamics for 3D Duet Singing Animation and Beyond
We present InterSing, a framework for generating realistic 3D head animations for duet singing performances. Unlike solo singing, duet performance requires each singer to balance individual expressiveness with intermittent interaction at musically salient moments, such as phrase boundaries, synchronized rhythms, and call-and-response passages. Because these interactions are sparse and rhythm-dependent, existing audio-driven animation methods and conversational interaction models do not adequately capture their structure. Our key insight is that duet coordination can be represented as a time-varying signal that reflects how strongly performers engage with one another throughout a song. Based on this observation, we introduce interaction logits, an interpretable latent representation that models the degree of cross-performer engagement at each time step. We learn these logits using weak supervision and use them to condition an interaction-aware diffusion model jointly driven by audio features and interaction dynamics. This formulation enables unified multi-mode generation, spanning independent motion, coordinated behavior, and smooth transitions between them. Experiments show that InterSing generates realistic and expressive singing head animations with stronger coordination and musical alignment than existing methods, while preserving each performer's characteristic motion style. We further demonstrate that the same formulation generalizes to multi-singer performances and provides intuitive control over when and how performers engage.
Read the full abstract
Computer Vision
Sound-based Multi-Person 3D Pose Estimation
Can we recover the 3D poses of multiple people using only sound? This paper presents the first attempt to estimate multi-person 3D poses solely from acoustic signals. Estimating the poses of multiple individuals using acoustic signals is inherently challenging due to the superposition of motion-dependent signal variations. Unlike single-person scenarios, the presence of multiple subjects leads to overlapping acoustic signatures, making it difficult to attribute specific signal changes to an individual's pose. Furthermore, the complexity is compounded by inter-person reflections, which introduce intricate propagation delays that obscure the temporal motion-acoustic relationship. To address these issues, we propose SoundMHPE (Sound-based Multi-person Human Pose Estimator), a novel encoder-decoder framework consisting of two key components. First, the Acoustic Multi-scale Encoder captures diverse temporal and fine-grained frequency features to isolate subtle acoustic signatures from complex, overlapping signals. Second, the Temporal Pose Decoder employs an attention mechanism to disentangle multi-person information across successive frames. By jointly accounting for temporal dynamics and inter-person dependencies, this component precisely reconstructs frame-wise individual poses. To validate our approach, we constructed the 6-hour Acoustic Multi-person Pose (AMP) dataset consisting of 432K synchronized frames of multi-person pose and acoustic data, and demonstrated that our SoundMHPE outperforms baseline models. Project page: https://oumi03.github.io/sound-mhpe/
Read the full abstract
Machine Learning
Adaptation Interfaces for In-Context Tabular Foundation Models in Time-to-Event Prediction
Tabular foundation models (TabFMs) achieve strong performance on structured data, particularly for standard classification and regression problems. Yet, extending them to censored time-to-event prediction is challenging because it requires properly handling censoring and event-time dynamics. Building on our prior work, we further link TabFMs with CoxPH and DeepHit and revise the context-resampled training procedure. We evaluate temporal zero-shot reformulation, classification-based fine-tuning, and survival-head adaptation using frozen TabFM backbones on 74 single-risk data sets, and we additionally study 4 competing-risk data sets. Zero-shot inference is effective on smaller single-risk data sets, whereas supervised adaptation becomes increasingly advantageous as data sets scale. Cox provides the most reliably strong interface, especially for Integrated Brier Score (IBS) on larger data sets. DeepHit is relatively stronger for the time-dependent Concordance Index than for IBS, while cause-specific MTLR ranks highest among the TabFM survival heads in the four-data-set competing-risk analysis. Classification fine-tuning becomes more competitive with zero-shot inference as data sets grow but remains weaker for probabilistic prediction. Overall, our results indicate that effective TabFM transfer depends on the data regime and on the statistical structure represented by the chosen adaptation interface. The implementation scripts used for this work are available at https://github.com/kaylode/survival-fm.
Read the full abstract
General Relativity and Quantum Cosmology
Probing magnetic fields of compact objects with continuous gravitational waves
Spinning, deformed compact objects such as neutron stars are canonical sources of continuous gravitational waves. These objects may be born with magnetic fields that can strongly influence their spin evolution and, consequently, their gravitational wave detectability. Employing a Bayesian framework, we use for the first time, the non detection of continuous gravitational waves in the LIGO Virgo KAGRA (LVK) third observing run (O3), using reported amplitude upper limits from various LVK and Einstein@Home searches, to place population level constraints on the birth magnetic field distribution of Galactic compact objects. To this end, we simulate compact object populations whose spin evolution is governed by gravitational wave emission and magnetic dipole radiation. We explore multiple ellipticity models and magnetic field decay timescales, and find that the lower limit on the birth magnetic field distribution hyperparameter $B_0$ is constrained to lie in the range $10^{9.6}\mathrm{G} \lesssim B_0\,\lesssim 10^{13.3}\mathrm{G}$. Furthermore, we reinterpret the number constraints on the total population of Galactic compact objects, reported previously by \citet{Prabhu_2024}, as upper limits on the number of compact objects with a population averaged magnetic field, presented as a function of ellipticity and gravitational wave frequency for all searches considered here.
Read the full abstract
Cryptography and Security
The Security Feature Location Problem
Software security must be realized through security features such as authentication and encryption, but which features does a system implement, and where? We present security feature location: the task of relating code locations to security features, enabling developers to understand security implementations and assess whether intended security properties are enforced.
Read the full abstract
Computation and Language
RefactorPlatform: An Open-Source Harness for Controlled Evaluation of Repository-Scale Refactoring Agents
Repository-scale refactoring requires coding agents to propagate a single change across many interdependent files without altering program behavior, yet to our knowledge no existing harness isolates the design choices that determine agent success on this task. We present RefactorPlatform, an open-source evaluation harness that holds the environment fixed and varies each design axis explicitly: model backbone (via OpenRouter and GitHub Copilot CLI), execution regime (baseline, retrieval-augmented, and multi-agent), and prompt specificity. Each run executes in an isolated workspace with live terminal streaming, per-task logging of tokens, diffs, and transcripts, AST-based verification, and exportable telemetry for audit and reproduction. Demonstrating the platform on 100 multi-file RefactorBench tasks across four model families, we illustrate the analyses it supports: AST-aware chunking outperforms naive token-window chunking by 25-30 percentage points (pp) across prompt modes, whereas naive retrieval falls below the retrieval-free baseline; a lean retrieval-augmented single agent (86%) beats the sub-agent configuration we evaluated (66%) on matched tasks with no task passing under delegation that fails under retrieval; and retrieval's accuracy gains absorb its token overhead, leaving cost per successful refactoring unchanged. RefactorPlatform is open-sourced to make refactoring-agent evaluation reproducible and auditable.
Read the full abstract
Earth and Planetary Astrophysics
Disk dispersal freezes overstable resonant librations
Context. Convergent migration in a gaseous protoplanetary disk can capture a planet pair into mean-motion resonance. Eccentricity damping can subsequently make the resonant libration overstable and drive the pair out of resonance. Most studies of this process, however, assume a static disk. Aims. We examine how the decay of disk torques during dispersal changes this outcome and whether it can freeze an overstable libration before the pair escapes. Methods. We describe disk dispersal by allowing the migration and eccentricity-damping timescales to increase exponentially on a local timescale $\tau_{\rm d}$. Integrating the time-dependent growth rate predicts $\tau_{\rm d,crit}\propto\tau_{e,0}$. We test this scaling with direct $N$-body integrations and relate $\tau_{\rm d}$ to the time taken by a photoevaporative cavity edge to cross the local torque-producing region. Results. The simulations recover a linear boundary, $\tau_{\rm d,crit}\simeq S\tau_{e,0}$, with $S\simeq3$ over the explored parameter range. In a fiducial minimum-mass solar nebula, faster propagation of the cavity edge shortens the local dispersal time. The ratio $\tau_{\rm d}/\tau_{\rm d,crit}$ also decreases with orbital radius, so both effects favour resonant survival. Conclusions. When local disk dispersal is sufficiently rapid, the libration amplitude can freeze and the planet pair can remain in resonance instead of escaping through overstability. Late disk evolution can therefore alter the outcome of resonant overstability.
Read the full abstract
Quantum Physics
Neural networks learn to reconstruct multipartite entanglement from quantum marginals
Different sets of local correlations are not equivalent: some fragments of reduced data uniquely determine a global quantum state, while others leave it ambiguous. The quantum marginal problem asks whether a collection of reduced density matrices uniquely determines a compatible global quantum state. Although generic quantum states are uniquely specified by suitable sets of marginals, different collections of marginals are not equally informative: some uniquely determine the global state, whereas others leave it ambiguous. Identifying when uniqueness holds, and reconstructing the global state from partial information, remains computationally demanding and experimentally challenging. We show that information about the multipartite entanglement class and reconstructability in four-qubit systems is compactly encoded in small sets of two- and three-qubit marginals. Using semidefinite programming, we chart the reconstructability landscape across 49 inequivalent SLOCC entanglement classes and show that uniqueness strongly depends on both entanglement structure and marginal order. Neural networks trained only on reduced density matrices learn this structure directly. They accurately classify marginal reconstructability and, when uniqueness holds, reconstruct the full four-qubit density matrix with high fidelity from two- and three-qubit marginals. We benchmark the approach on a four-qubit nuclear magnetic resonance quantum processor and demonstrate that reconstructions from experimentally measured marginals remain faithful despite phase damping and control imperfections. Our results show that neural networks can learn when local correlations uniquely specify a global quantum state, and reveal how global quantum structure is encoded in reduced data.
Read the full abstract
Computation and Language
Cache-Aware Joint Router Adaptation for Memory-Efficient MoE Inference
Mixture-of-Experts (MoE) models activate few experts per token, yet their full expert sets can exceed GPU memory and require repeated weight transfers during decoding. We formulate expert-cache management as a model-side algorithmic problem and propose cache-aware post-training that jointly adapts the MoE backbone and lightweight auxiliary routers while preserving the native inference-time Top-K rule. The update-only Temporal Router learns same-layer retention across tokens without proactive loading. The full Spatio-Temporal Router adds a Spatio Router that uses the causal predecessor's hidden state to refine the temporal cache before target-layer access. We evaluate both modes on Qwen3 and GPT-OSS across GSM8K, MATH, and CommonsenseQA. Temporal Router consistently improves hit rate and reduces expert-weight traffic over matched LM-only baselines. On Qwen3, the full mode improves adjusted hit rate by 1.15–18.03 points and reduces traffic by 4.6–53.3% relative to the strongest evaluated prefetching baseline; GPT-OSS results are competitive but task-dependent. Auxiliary-only training preserves baseline accuracy but yields modest coverage gains; joint post-training achieves substantially higher coverage. Sensitivity analyses distinguish the effects of cache capacity, refinement budget, and cache-loss weight on coverage, traffic, and quality.
Read the full abstract
Artificial Intelligence
From Language Models to World-Acting Systems: Progress and Limits of Agentic AI across Digital, Social, Virtual, and Physical Environments
Large language models become consequential agents when surrounding systems let outputs change external state. Models now call tools, operate interfaces, delegate work, retain state, inhabit generated worlds, and control robots or laboratory equipment. Such advances are often narrated as one march toward autonomy, conflating model competence, system integration, persistence, and safe authority. This critical review synthesizes primary research and official technical specifications available by 31 August 2026. We organize the evidence along delegated authority, temporal persistence, and environmental coupling, while separating model, harness, and environment. Within the evidence examined, action-interface expansion is documented more convincingly than robust completion, recovery, authorization, or independent verification. Model Context Protocol and Agent2Agent improve interoperability but do not establish trustworthy delegation; multi-agent organization adds specialization alongside cost and correlated failure. Persistent simulations and world models support training and planning but do not themselves demonstrate agency; robotics and self-driving laboratories establish bounded feasibility rather than unattended open-world reliability. We propose justified delegation as an analytical and normative heuristic, not an observed law or certified score: expand action scope only where evidence supports provenance, bounded authority, failure detection, safe recovery, and calibrated human control. This framing yields a research agenda for coupled model-harness evaluation, capability-based permissions, durable state, cross-agent accountability, and staged physical validation.
Read the full abstract
Robotics
Reasoning Without Inference Cost: Latent Semantic Scaffolding for Robot VLA Policies
Vision-language-action (VLA) models are trained by imitation and capture what action to take but not why; adding causal reasoning improves manipulation, but current methods pay for it at inference time - generating reasoning tokens or rolling out predicted future states at every step, a cost that compounds over long horizons. We ask whether this benefit can instead be captured during training and discarded before deployment. We introduce Latent Semantic Scaffolding (LSS), an auxiliary loss applied during human-demonstration pretraining that aligns a VLA's action-token representations to text embeddings of physical-reasoning rationales through a small projection head. The head is dropped at inference, leaving the unmodified base policy with zero added cost. Our central finding concerns alignment granularity: aligning each action token to the rationale of its own manipulation phase (Dense LSS) rather than to a single pooled episode-level embedding (Pooled LSS) yields representations that transfer markedly better to held-out tasks. Dense LSS attains both the best in-distribution success and the best transfer to tasks unseen during alignment, whereas pooled alignment over-specializes to the training task. A representational probe shows Dense LSS induces roughly twice the per-phase separability in the backbone, supporting that phase-local alignment is the operative mechanism.
Read the full abstract
Astrophysics of Galaxies
Quantifying collision-driven mass loss in supermassive star formation: the role of stellar structure and accretion
Observations of high-redshift galaxies with JWST have renewed interest in scenarios where supermassive stars form via runaway stellar collisions in dense clusters, yet the impact of collision-driven mass loss on their growth remains uncertain. In this work, we perform a post-processing analysis of 3D hydrodynamical simulations of the formation of a supermassive star, applying an analytic mass-loss prescription to stellar collisions while exploring different assumptions for the internal stellar structure. We consider a polytropic main-sequence model, a semi-analytic accreting protostar model, and structures derived from stellar evolution calculations. We find that the cumulative mass-loss fraction depends sensitively on the adopted stellar structure, ranging from $\lesssim 10-25\%$ for more compact configurations to $\gtrsim 30-40\%$ for more extended protostellar models. The importance of mass loss further depends on the dynamical state of the system, including the ratio of stellar velocity dispersion to the stellar surface escape velocity. We find significant uncertainty depending on the prescription used. As a result, collision-driven mass loss could significantly limit the growth of the central object, at least in some cases. Overall, our results indicate that uncertainties in the internal structure of rapidly accreting protostars represent a major source of systematic uncertainty and must be better constrained to robustly assess the viability of runaway-collision pathways for forming massive black hole seeds.
Read the full abstract
Medical Physics
Attention-guided super-resolution of 4D flow MRI in carotid arteries
Four-dimensional (4D) flow magnetic resonance imaging (MRI) is a powerful non-invasive technique for visualizing and quantifying complex blood flow patterns in vivo. Despite its clinical promise, broader adoption is limited by low spatial resolution and sensitivity to noise, which restrict accurate assessment of critical hemodynamic biomarkers such as wall shear stress, pressure gradients, and turbulent kinetic energy. To overcome these challenges, we propose a deep learning-based super-resolution framework that integrates multi-scale feature extraction and attention mechanisms to enhance the quality of 4D flow MRI data. The model was trained on a dataset of 120 patients with 240 stenosed carotid arteries. High-resolution ground truth data were generated using patient-specific computational fluid dynamics (CFD) simulations based on segmented vascular geometries and physiologically realistic boundary conditions, and the resulting velocity fields served as targets for supervised learning. The proposed architecture uses convolutional block attention modules (CBAM) to guide the network toward clinically relevant spatial features and to suppress noise in low-resolution inputs. Quantitative results show that the attention-guided model substantially reduces the root mean square error (RMSE) compared with a baseline model without attention, and qualitative velocity contour analysis confirms improved reconstruction of intricate flow patterns. These findings highlight the capacity of the model to restore high-fidelity flow fields under noisy conditions and support the use of deep learning to extend the clinical utility of 4D flow MRI for non-invasive hemodynamic assessment.
Read the full abstract
Social and Information Networks
Finding Many Overlapping Dense Subgraphs Using Triadic Cohorts
Graphs are a standard representation for data in the social sciences, cybersecurity, computer infrastructure, bioinformatics, and more. Typical real-world graphs are sparse, meaning the average degree is small (in the tens, while the number of vertices is more than millions). When graph data is collected from a source, a major task is to perform data exploration. Thus, any region of “density" is of interest, since it indicates special structure. An important goal is to cover a significant portion of the graph using dense subgraphs. Existing algorithms that find many dense subgraphs do not have overlapping output, and hence provide limited coverage. Other methods that produce overlapping clusters do not generate dense subgraphs. The main goal of this paper is to develop provable and practical methods that can provide overlapping dense subgraphs that can cover large portions of real-world networks. Our contribution is an algorithm CohortRecovery that achieves this goal. We first develop a mathematical framework of triadic cohorts that captures the notion of “detectable dense subgraphs” that potentially overlap. We prove that CohortRecovery can output a set of dense subgraphs, such that each triadic cohort is almost completely contained in some dense subgraph. We give a practical implementation of CohortRecovery and demonstrate it on a variety of datasets. It typically runs in under ten minutes on a commodity machine even on graphs with tens of millions of edges. For numerous datasets, CohortRecovery is able to cover more than 25% of the vertices in non-trivial subgraphs of density more than 0.8, and is significantly better than a wide variety of scalable graph clustering/community detection algorithms. Moreover, we demonstrate that output of CohortRecovery captures ground truth clusters obtained by manual curation.
Read the full abstract
Mathematical Physics
Finite-Corner Reconstruction and Separation for Cones of Normal Maps
We develop a finite-corner framework for convex cones of normal maps between \(\mathcal B(\mathcal H)\) and \(\mathcal B(\mathcal K)\). The basic structural assumption is stability under finite cut–pad operations. We prove that every point-ultraweakly closed cut–pad stable cone is completely determined by its finite-dimensional corner cones, and conversely that every coherent family of closed finite-dimensional corner cones admits a unique global realization of this type. For a cut–pad stable cone that is not necessarily point-ultraweakly closed, reconstruction from the norm closures of its corner cones yields exactly its point-ultraweak closure. Combining this reconstruction principle with finite-dimensional separation and the Choi representation, we show that non-membership in a global cone is always detected on a single finite-dimensional corner by a finite-dimensional witness. We illustrate the framework for completely positive and decomposable maps; in the decomposable case, the Hermitian parts of the finite-corner witnesses are PPT.
Read the full abstract
Fluid Dynamics
Homoclinic-shedding in the steady forced water-wave problem
Recently, an infinite set of steady solitary-wave solution branches was discovered for inviscid and irrotational flow over a localised topographic depression. These solutions were restricted to critical flow, when the Froude number is unity. Several questions were raised, including i) what happens to the solution structure as the flow speed is continued into the supercritical regime? and ii) how does the decay rate of the topography affect the solution structure? In this article, we answer these questions by examining exact steady solutions of the fKdV model and numerical solutions of the fully nonlinear Euler system over i) trench, ii) Gaussian and iii) generalised Agnesi topographies. We uncover a highly non-trivial bifurcation structure consisting of spirals and closed loops, characterised by the phenomenon of `homoclinic shedding', where pairs of solitary waves are emitted up- and downstream as the amplitude is varied.
Read the full abstract
Earth and Planetary Astrophysics
Habitable-zone Earths at the detection frontier: Measured completeness and false-alarm rate of a transit pipeline for the PLATO M-dwarf sample
The PLATO mission will observe its first long-pointing field, LOPS2 (Long-duration Observation Phase, South field), for at least two years, bringing the habitable zones (HZ) of its M-dwarf sample within reach of a transit search. The completeness and false-alarm rate of a pipeline validated on TESS M dwarfs are measured on simulated PLATO photometry and converted into a candidate-level forecast for the LOPS2 M dwarfs. Simulated PLATO light curves (717 d baseline; three stellar cells of the faint half of the sample) were processed end to end in two arms, without and with a pre-launch model of the instrument systematics: 930 in the first (465 realisations, 525 injected planets, 465 noise-only twins) and 790 in the second. Injected planets measure completeness against the expected signal-to-noise ratio (S/N), separately for the long-period regime of the HZ; noise-only twins measure the false-alarm rate against the floor on the red-noise-aware statistic S/N_pink. Completeness reaches 50% at S/N = 7.7 +/- 0.15 over all periods and 8.6 +/- 0.3 in the long-period regime. The false-alarm rate falls from 79 +/- 2% to 0.6% as the S/N_pink floor rises from 6 to 8, and 6 +/- 1% at the adopted floor 7.5 (completeness 54 +/- 2%). In the field, the 50 known transiting objects of interest (19 confirmed planets, 31 candidates) would all be recovered; a transiting HZ planet of 1.0 (1.5) Earth radii would be recovered as a candidate with ~40% (65%) probability, giving about 4 HZ Earths and 7 super-Earths over 16 929 field M dwarfs. The systematics arm shifts the 50% point by +0.54 (+0.27/-0.22) in S/N, a penalty localised on the transition, leaving the floor robust and negligible on the long-period injections that feed the HZ. Habitable-zone Earths around the PLATO M dwarfs sit at the measured detection frontier, where completeness and reliability, both measurable before launch, decide the yield.
Read the full abstract
Computer Vision
SimFuse3D: Source-Guided Target Simulation and Confidence-Guided Multi-Stage Localization Reweighting for Cross-Platform 3D Object Detection
Changes in sensor height and viewpoint alter object-level point distributions, making cross-platform LiDAR unsupervised domain adaptation (UDA) difficult. Self-training uses labeled source scans and unlabeled target scans, yet a retained prediction may provide a useful target location while enclosing sparse foreground returns, background clutter, or points inconsistent with the predicted box. We refer to this mismatch as box-point inconsistency. We introduce SimFuse3D, which preserves the target placement and repairs the associated pseudo object using measured geometry from labeled source scans. Object Memory retrieves a similar labeled source instance. Target Simulation places the retrieved source geometry at the target location, aligns its points with the target viewing geometry, and filters the aligned crop to approximate the target observation. Confidence-Guided Multi-Stage Localization Reweighting (CMLR) maps each target pseudo-object confidence score to a bounded weight shared by RPN localization and R-CNN box regression. All components operate only during adaptation, leaving the detector architecture and inference graph unchanged. Across six cross-platform transfers, SimFuse3D consistently outperforms Pi3DET-Net and achieves the best performance among the compared adaptation methods on nearly all metrics. On nuScenes-to-KITTI, it ranks first among the compared adaptation methods with both evaluated detectors.
Read the full abstract
High Energy Physics (Phenomenology)
Di-hadron Fragmentation Functions beyond LO
In this talk, we discuss the extraction of Dihadron Fragmentation Functions (DiFFs) beyond leading order. In particular, we review the latest extraction by the MAP Collaboration of the Unpolarized Dihadron Fragmentation Functions (DiFFs) from a fit of the 2017 BELLE data for the inclusive $\pi^+\pi^-$ pair production within the same jet in $e^+e^-$ annihilations. This new extraction improves the perturbative QCD accuracy by including NNLO calculations, relies on Monte Carlo generators only for flavor decomposition, and complements the traditional physics-informed parametrization with a Neural Network extraction. We also discuss the setup for the extraction of the transversely polarized DiFFs, where we revisit the NLO calculation for $e^+e^-$ annihilation into a transversely polarized quark-antiquark pair, obtaining a result different from the one reported in the literature.
Read the full abstract
Numerical Analysis
Energy-Consistent Splitting and Decomposition Approaches for Coupled port-Hamiltonian ODEs
Operator splitting provides an attractive approach for the numerical integration of (coupled) port-Hamiltonian systems, as it allows the underlying system structure to be exploited at the level of the individual subproblems. However, the choice of the decomposition is not unique and may strongly affect both the computational efficiency and the preservation of the energy behavior of the original system. In this work, we investigate this interplay systematically and introduce energy consistency as a criterion for assessing splitting methods for port-Hamiltonian ordinary differential equations. We derive sufficient conditions under which a splitting based on a given decomposition inherits the energy behavior of the continuous system and use these conditions to analyze several decomposition strategies for coupled port-Hamiltonian systems. In particular, we compare decompositions that preserve the structure with approaches that exploit lower-dimensional subsystem dynamics or separated time scales. The analysis is complemented by numerical experiments using Strang splitting and its multiple-time-stepping extension. A scalable electro-thermal benchmark with fast electrical and slow thermal dynamics is employed to assess accuracy, energy behavior, and computational efficiency. The results demonstrate that preserving the port-Hamiltonian structure of the subflows is essential for energy-consistent splitting, whereas decompositions that exploit subsystem structure or time-scale separation can provide substantial computational advantages. In particular, the time-scale decomposition yields significant efficiency gains for systems with pronounced multirate characteristics, while structure-destroying decompositions may lead to undesirable energy behavior.