Peer Revenue Fully funded

Papers

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

All of arXiv

Saturday 26 September

Analysis of PDEs

Fitness is asymptotically irrelevant under uniform competition in phenotype-structured populations

Artur César Fassoni

We prove that, in phenotype-structured populations under uniform competition, the long-term phenotypic distribution does not depend on the proliferation rates: it is the stationary distribution of the switching dynamics alone. The density u(x,t) of cells with phenotype x evolves by advection and diffusion in phenotype space (switching) and by proliferation at a phenotype-dependent rate r(x,t), modulated by a factor g(U) common to all phenotypes, where U is the total population. This extends to continuous phenotypes a recent result for compartmental models. On a bounded interval with no-flux boundary conditions, the total population converges monotonically to the carrying capacity U*, the density converges in L1 to U*psi, where psi is the stationary density of the switching dynamics, and the nonlinear model is asymptotically equivalent to the linear one; by an explicit estimate, the imprint of fitness decays at the spectral gap of the switching dynamics. On the whole real line we prove the same convergence under two natural conditions, confinement and no escape to infinity, covering Ornstein-Uhlenbeck dynamics, multi-well landscapes, and heavy-tailed densities. If g'(U*) < 0 and psi satisfies a Poincare inequality, convergence is exponential in a weighted L2 norm, while no uniform L1 rate holds for Ornstein-Uhlenbeck dynamics. The proof follows the compartmental one: under uniform competition the reaction term has one sign, so its size is the growth rate of the total population, integrable in time, and the switching dynamics contracts the zero-mass perturbations it produces. Since it uses only the Markov semigroup generated by switching, the proof extends to general mechanisms; we apply it to nonlocal switching by jumps and to diffusions with non-gradient drift in bounded domains of R^d, whose stationary density is not of Boltzmann form and carries a circulating flux.
Read the full abstract

Artificial Intelligence

EMIR$^2$: Evolution-Aware Memory with Intent-Guided Multi-Round Retrieval

Jinlan Liu, Hongliang Sun, Yong Wang et al. (+4 more)

Long-term memory enables large language model (LLM) agents to leverage historical interactions for future tasks. However, existing memory systems struggle to utilize continuously evolving historical information, as they often rely on static memory representations and single-round retrieval strategies, failing to track factual changes or integrate distributed evidence across long-term interactions. To address these challenges, we propose EMIR$^{2}$, an Evolution-Aware Memory framework with Intent-Guided Multi-Round Retrieval, enabling LLM agents to maintain evolving historical knowledge and adaptively retrieve relevant evidence. Specifically, EMIR$^{2}$ constructs a State-Evolving Memory Graph (SEMG) that represents long-term memory as evolving knowledge states supported by temporal event trajectories and evidential associations. By maintaining semantic states through evidence-based updates, SEMG preserves historical evolution and enables evidence tracing under complex and conflicting scenarios. Building upon this, we introduce an intent-guided multi-round retrieval mechanism that iteratively identifies missing evidence and expands retrieval based on accumulated information. Experiments on LoCoMo and MemConflict demonstrate that EMIR$^{2}$ improves long-term memory utilization, dynamic and static conflict handling, and complex retrieval performance, achieving relative improvements of more than 12% in certain categories. These results highlight the effectiveness of jointly modeling memory evolution and adaptive evidence acquisition for long-term agent interactions.
Read the full abstract

All authors Jinlan Liu, Hongliang Sun, Yong Wang, Bolin Zhang, Dinabo Sui, Dianhui Chu, Zhiying Tu

Analysis of PDEs

On solutions for damped fractional Navier-Stokes equations in a critical Lei-Lin-Gevrey space

Bruno S. Donato, Wilberclay G. Melo and Thyago S.R. Santos

We study the fractional Navier-Stokes equations with cubic damping in the specific Lei-Lin-Gevrey space $\mathcal{X}^0_{a,\sigma}(\mathbb{R}^3)$ (with $a\geq0$ and $\sigma\geq1$). We establish local and small-data global well-posedness, quantitative blow-up criteria, Gevrey loss-of-radius estimates, and perturbative stability in the critical endpoint case $\gamma=\frac12$ of the usual Navier-Stokes system.
Read the full abstract

Astrophysics of Galaxies

Cool Embeddings: Predicting Galaxy Cluster Cooling Times in IllustrisTNG and TNG-Cluster with AstroCLIP

Rawan Karam, Carter Lee Rhea, Julie Hlavacek-Larrondo et al. (+4 more)

Foundation models trained on optical astronomical data have not previously been applied to X-ray cluster thermodynamics. We show that embeddings from AstroCLIP, a contrastive vision-language model trained on optical galaxy images and spectra, retain physically meaningful structure when applied to mock Chandra X-ray images of simulated galaxy clusters - a significant shift in observational modality and physical scale. We target the central cooling time ($t_{\mathrm{cool},0}$) of the intracluster medium, a key diagnostic of cluster thermodynamics traditionally requiring X-ray spectroscopy. We present a machine learning framework that predicts $t_{\mathrm{cool},0}$ directly from X-ray images, applied to 711 projections from TNG300 and 993 from TNG-Cluster from the IllustrisTNG simulations. AstroCLIP embeddings form a structured latent space in which clusters with similar thermodynamic properties are naturally grouped, with no equivalent separation by halo mass. Building on this, we introduce a dot-product $k$-nearest neighbour attention model that retrieves similar clusters from the training set and adaptively reweights their cooling times. This model achieves root-mean-square errors of 0.24 (TNG300) and 0.23 (TNG-Cluster) in $\log(t_{\mathrm{cool},0})$, and demonstrates strong out-of-distribution generalisation, outperforming a fine-tuned convolutional neural network trained on raw images (RMSE 0.50 versus 0.73). These results demonstrate that pretrained foundation model embeddings transfer across astrophysical domains and encode thermodynamic information sufficient for inference of cluster cooling properties - offering a scalable path toward cool-core characterisation in all-sky surveys such as eROSITA.
Read the full abstract

All authors Rawan Karam, Carter Lee Rhea, Julie Hlavacek-Larrondo, Laurence Perreault-Levasseur, Marine Prunier, Adam Alexandre, Hyunseop Choi

Computation and Language

HERO-MoE: Historical Expert Routing with Scale-Preserving Fusion

Junxiang Qiu, Zhengsu Chen, Xinting Hu et al. (+6 more)

Mixture-of-Experts (MoE) architectures have become a standard way to scale model capacity while keeping computation sparse, yet routing remains a key determinant of MoE quality and training behavior. Prior empirical studies suggest that MoE routing reflects input semantics and upstream computation across depth, but standard routers do not explicitly use the routing distributions produced by preceding layers. We propose HERO-MoE, Historical Expert ROuting with Scale-Preserving Fusion, a routing framework that injects historical routing priors into MoE routers by reusing detached, dense routing distributions collected from preceding MoE layers. The key idea is simple: HERO-MoE preserves the original token-conditioned routing branch and adds a residual historical routing contribution before the standard softmax and top-$k$ dispatch. To stabilize this historical signal, HERO-MoE introduces a scale-preserving fusion mechanism that matches the magnitude of historical routing memory to the current hidden representation and accounts for the number of visible historical layers, without introducing an auxiliary routing loss or a fusion-specific tuning parameter. By reusing routing distributions already computed by preceding MoE layers, HERO-MoE improves training-loss reduction with modest end-to-end overhead. The resulting router remains compatible with standard sparse dispatch, including top-$k$ and group-limited routing, and can be inserted into existing MoE backbones with minimal architectural changes. Experiments on an approximately 8B-parameter MoE model with 0.5B active parameters, trained from scratch on 100B tokens, show that HERO-MoE reduces the final loss from 1.6393 to 1.6184, while peak memory and FLOPs increase by only 0.44% and 0.64%, respectively.
Read the full abstract

All authors Junxiang Qiu, Zhengsu Chen, Xinting Hu, Shuo Wang, Hengheng Zhang, Shaofeng Zhang, Changcheng Li, Boyu Shi, Qi Tian

Quantum Physics

Stable Long-Range Charging and Boundary-Mediated Thresholds in a Waveguide Quantum Battery

Li-Hong Yu, Zhuo-Yu Zhang, Yu Wang et al. (+1 more)

The waveguide-integrated architecture holds promise for scalable on-chip integration of a charger, a transmission channel, and a quantum battery in a modular, independently optimizable design. We investigate a minimal charging configuration consisting of two emitters coupled to a semi-infinite waveguide and identify the optimal charging mechanism, wherein non-Markovian dynamics driven by out-of-band bound states give rise to single-frequency oscillations of the stored energy. We show that near-unity charging efficiency is achievable at weak to moderate coupling strengths via detuning control, provided that the battery-to-boundary separation $n$ is less than three times the charger-to-boundary separation $m$ ($n<3m$)—a constraint imposed by the mirror-induced feedback—while the average charging power decays exponentially with the spatial separation $n-m$. These findings offer useful insights for monolithic solid-state quantum energy storage and transport.
Read the full abstract

All authors Li-Hong Yu, Zhuo-Yu Zhang, Yu Wang, Xin Li

Machine Learning

DimPO: Dimensionality Reduction for Attention using Preference Optimization

Vojtěch Lanz, Yufei Cui and Prasanna Parthasarathi

A linear projection can reduce the dimension of query and key vectors without updating the pretrained model, but it remains unclear which training objective best preserves model behavior. We ask whether preferences over keys and attention mass on the highest-weighted keys provide a better signal than matching the full attention distribution with KL divergence, especially in long-context settings. We introduce DimPO, which combines listwise preference optimization with a lightweight top-k cross-entropy term for head-fidelity. DimPO is trained offline from the attention patterns of a frozen language model, with one map per layer, shared by the query and the keys and trained separately from the other layers. Across LLaMA3.2-3B, LLaMA3.1-8B, Qwen2.5-7B, and Qwen3-4B Instruct models, pairwise preference objectives outperform the triplet baseline and retain 98% of the original score on short-context tasks when projecting to half the dimension on the last 40% of the layers. With more projected layers or on long-context RULER, they degrade rapidly. In contrast, KL and DimPO, which use every key during training, retain about 95% of the original RULER 4k score on the 8B model when projecting up to 50% of the layers. KL-based projections remain closer to the original attention distribution and attention output, yet DimPO achieves better downstream performance. Beyond 50% of projected layers, DimPO increasingly outperforms KL on tasks including SQuAD, common-word extraction, frequent-word extraction, and variable tracking. These results suggest that under dimensionality reduction, preserving the ordering and concentration of task-relevant attention can matter more than reproducing the full attention distribution.
Read the full abstract

Quantitative Methods

Structured Population Models for Follicular Development

Edilbert Christhuraj, Claudio Iuliano and Alexander Lange

Ovarian follicle development has been modeled either by compartmental models, where developmental stages are represented by discrete compartments, or by physiologically structured population models (PSPMs), formulated as transport PDEs describing follicles along a continuous maturation variable. Compartmental models reproduce full-cycle hormonal and follicular dynamics but cannot describe cellular features. PSPMs capture multiscale behavior, yet lack a direct link to compartmental descriptions, making hormonal feedback difficult to incorporate. Here, we introduce a class of ODE systems derived as finite-dimensional reductions of PSPMs via a perturbative moment closure. These systems provide backbone approximations of the structured dynamics, retaining key nonlinear mechanisms governing recruitment, selection for dominance, and atresia while projecting maturation onto macroscopic variables. Zero- and first-order moments, derived from phenomenological assumptions regarding follicular maturation, allow us to realistically model follicle sizes and blood estradiol concentrations over one follicular wave. This application is supported by bovine data and likely generalizable to other mammalian species.
Read the full abstract

Computation and Language

Groupwise Agentic Grading and Advantage Redistribution for Code Agent RL

Jinhao Dong, Liang Zhao, Zihao Yue et al. (+7 more)

Reinforcement learning (RL) for code agents often uses executable tests to provide binary rewards. With these rewards, Group Relative Policy Optimization (GRPO) assigns identical advantages to test-passing trajectories within each rollout group, overlooking differences in implementation quality and adherence to task requirements. This leaves the policy without a learning signal that favors clean, targeted implementations over those containing unnecessary or out-of-scope changes. We introduce GAGAR, a framework for quality-aware credit redistribution in code agent RL. Built on dynamic sampling that retains groups containing both passing and failing trajectories, GAGAR places all trajectories from each group in a shared workspace, where an SFT-trained agentic grader jointly inspects them and ranks the test-passing candidates. Based on this ranking, we downweight lower-ranked trajectories and proportionally rescale the advantages of all test-passing trajectories to restore their original sum. This sum-preserving redistribution retains the relative weights established by quality-based downweighting while shifting credit toward higher-quality implementations. We evaluate GAGAR at industrial scale using pre-RL SFT checkpoints of MiMo-V2.6-Flash (310B total parameters) and MiMo-V2.6-Pro (1.02T total parameters). Controlled code-only Flash experiments show improved code agent performance, reduced trajectory-length growth, and more stable training. We further apply GAGAR in large-scale mixed-task RL with both Flash and Pro. Our results support combining test-based verification with groupwise agentic grading to improve the quality and stability of code agent RL.
Read the full abstract

All authors Jinhao Dong, Liang Zhao, Zihao Yue, Wenhan Ma, Linghao Zhang, Lei Li, Shicheng Li, Yifan Song, Bowen Ye, Fuli Luo

Robotics

Assisting for Open-Ended Tasks: Goal-Oriented Shared Autonomy as a Particle Filter

Mengxue Fu, Ethan Xu, Sam Iyer-Singh et al. (+3 more)

A common approach for shared autonomy blends human inputs with autonomous assistance based on the human's likely goal. However, most existing approaches assume that a static set of possible goals is known a priori, which limits the use of such methods in unstructured assistive settings. We instead investigate how to enable shared autonomy with open-ended and dynamically changing goals. We formulate goal-oriented shared autonomy as a particle filter in which particles represent candidate human goals. Unlike conventional approaches with a fixed goal set, our transition model dynamically proposes new candidate goals as the interaction evolves, and human actions update the belief over these goals in real time. We instantiate this framework with foundation models (e.g., vision grounding and large language models) that propose context-relevant semantic goals, generate goal-conditioned assistance from low-level skill primitives, and refine those skills from human corrections. We assess our approach through a user study where 12 participants perform a variety of tabletop manipulation tasks with our method and state-of-the-art shared autonomy baselines. The results show that our particle filter-based approach reduces the amount of time users spend teleoperating the system and improves user satisfaction. User study videos: https://youtu.be/Ii26XuRqm9c
Read the full abstract

All authors Mengxue Fu, Ethan Xu, Sam Iyer-Singh, Yinlong Dai, Michael Hagenow, Dylan P. Losey

Other Condensed Matter

Phonon angular momentum transfer torque

Verena Brehm, Daniel A. Bustamante Lopez, Shu Zhang et al. (+1 more)

Angular momentum in solids is carried by (quasi)particles such as magnons, plasmons, and phonons, and coupling between these reservoirs enables diverse hybrid phenomena. Here, we introduce the transfer of angular momentum from lattice vibrations to magnetic order through the phonon angular momentum transfer torque (PAMTT), generated by a polarized phonon bath. We identify thermal routes for generating phonon angular momentum: a phonon Edelstein accumulation in noncentrosymmetric crystals and temperature-gradient-induced PAM currents with longitudinal Seebeck-like and transverse Hall components. Using realistic order-of-magnitude estimates, we find that PAMTT could produce measurable ferromagnetic-resonance frequency shifts and, for favorable interfacial coupling, drive magnetization reversal on nanosecond timescales.
Read the full abstract

All authors Verena Brehm, Daniel A. Bustamante Lopez, Shu Zhang, Dominik Juraschek

Artificial Intelligence

CUE-Mem: Benchmarking Long-Term User Memory via Implicit Cues in Multimodal Conversations

Yulin Hu, Yanyan Zhao, Zimo Long et al. (+6 more)

Long-term memory is essential for multimodal agents that interact with users across sustained conversations. However, user memories are not always explicitly stated: they may also be implied by recurring background objects in images, ambient sounds in audio, or other peripheral multimodal cues. Existing benchmarks largely focus on text-only memory or explicit multimodal evidence, leaving implicit multimodal cues underexplored. We introduce CUE-Mem, a text-image-audio benchmark for evaluating long-term user memory from implicit cues. CUE-Mem contains 2,674 questions across explicit and implicit evidence settings and covers four tasks: Entity Recall, Long Pattern, Personalized Recommendation, and Answer Refusal. Across textualized memory systems, implicit performance remains far below oracle evidence, locating the main bottleneck in preserving and retrieving subtle cues rather than question answerability. Increasing caption detail recovers more of this evidence, but brings uneven gains and rapidly growing token costs, motivating native multimodal access. Yet native access does not uniformly resolve the bottleneck: evidence use depends strongly on the backbone, while multimodal indexing introduces substantial retrieval noise. CUE-Mem provides a testbed for memory systems that selectively retain, retrieve, and use subtle multimodal evidence.
Read the full abstract

All authors Yulin Hu, Yanyan Zhao, Zimo Long, Xing Fu, Mengtong Ji, Weixiang Zhao, Yutai Hou, Qianchao Wang, Dandan Tu

Machine Learning

Trapped by Their Own Rollouts: Understanding Aggregation–Rollout Feedback in Federated On-Policy Distillation

Jinqian Chen, Jihua Zhu and Chang Liu

On-policy distillation (OPD) is a promising approach to language-model adaptation, aligning teacher supervision with the student's own generated trajectories. When adaptation prompts are distributed across clients, can this process benefit from federated collaboration? We study federated OPD and find that substantial collaboration gains can be obscured by learning-rate sensitivity: FedAvg can perform no better than independent local training at a small learning rate, yet recover a clear advantage at a larger rate. We explain this phenomenon through the student's dual role as learner and generator of future training data. An aggregation-induced optimization lag can delay access to useful teacher supervision, which in turn slows subsequent learning. Our theory establishes this aggregation–rollout feedback in a solvable model with a common optimum and stable updates, and identifies two coupled roles of learning rate: learning from current supervision and reaching future supervision. Guided by this analysis, we propose FedTOPS (Federated Teacher-guided On-Policy Scaling), which reuses teacher feedback on current trajectories to adapt the FedAvg update magnitude under clientwise predictive-change constraints. Across six mathematical reasoning benchmarks, FedTOPS improves macro Avg@8 over FedAvg by 4.56–14.57 percentage points across the evaluated student models and local learning rates.
Read the full abstract

Cosmology and Nongalactic Astrophysics

Bimodal mass primordial black holes in three-field inflation

M. M. Nasrolahzadeh and N. Khosravi

In this paper, we study the evolution of curvature and entropy perturbations in three-field inflation with non-canonical kinetic terms. Using the kinematic basis in field space, we characterize the background trajectory through its turning rate and non-planar torsion. We also emphasize the inherited structure of perturbations from the kinematic system, which detaches the distinct roles of turning rate and torsion in the interactions among perturbations. We present and study numerically a three-phase model with an intermediate decelerated expansion stage in which two separated transient epochs amplify two distinct families of modes, generating a double-peaked curvature power spectrum. This double-peaked power spectrum leads to the formation of primordial black holes with two characteristic masses, $M_1\sim2\times10^{-12}M_\odot$ and $M_2\sim52M_\odot$ explaining, simultaneously, both dark matter and the LIGO-Virgo-KAGRA binary black hole mass range.
Read the full abstract

General Relativity and Quantum Cosmology

Exact Solutions for Chiral Gravitational Waves from Spin-2 Mixing

Mohammad Ali Gorji and Yuhang Zhu

A spectator spin-2 field can mix linearly with the metric tensor perturbations. We study the coupled tensor system in an inflationary background, including a parity violating effect parametrised by $\theta$. Without expanding in the linear mixing between the two fields, we solve the system exactly for arbitrary mass while treating the mixing strength nonperturbatively. The mode functions of both helicities are constructed by acting with an operator-valued Gauss hypergeometric function ${}_2F_1$ on Whittaker modes, and the late-time power spectrum of each graviton helicity is expressed in closed form in terms of the generalised hypergeometric function ${}_3F_2$. We find that one helicity is exponentially enhanced, showing that strong mixing with a spectator spin-2 field can generate a large and highly chiral primordial gravitational wave spectrum. The degree of circular polarisation is bounded by $\tanh(\pi\theta)$, independently of the spin-2 mass and the mixing strength. The same solution also gives the late-time power spectrum of the spectator field and its cross spectrum with the graviton. In the weak mixing limit, our exact result reproduces the perturbative Schwinger-Keldysh result. These solutions provide a nonperturbative framework for studying primordial gravitational waves sourced by additional tensor modes.
Read the full abstract

Machine Learning

CAESAR: Clustering via Autonomous Embedding-Space Agglomerative Reorganization

Ilan Bacry, Rémi Devaux and Antoine Jardin

Clustering algorithms that operate on nearest-neighbor graphs, such as FINCH (First Integer Neighbor Clustering Hierarchy), depend heavily on the quality of the embedding space they are given. However, pretrained vision and language model embeddings are not optimized for this purpose. We propose CAESAR, a method that reorganizes a pretrained embedding space: a reorganization network is trained to pull mutual nearest neighbors together and push non-neighbors apart, yielding a reorganized embedding space substantially better suited to clustering. CAESAR offers a second major advantage: it never requires the number of clusters $K$. This matters because in realistic unsupervised settings, $K$ is typically unknown and discovering it is often part of the problem, yet most strong clustering methods take it as an input. We therefore design the entire CAESAR pipeline to infer $K$ rather than assume it is known. Empirically, reorganizing the embeddings consistently improves clustering over the raw space on both text and image datasets. Since the few deep clustering methods that also infer $K$ do not release their code, we complement controlled comparisons with methods that infer $K$ on the same embedding space by comparisons with strong deep clustering methods that are given the true $K$, giving them a substantial oracle advantage. Even so, CAESAR outperforms all of them on text, achieves the best results on the most challenging image benchmark and remains competitive on the others. Reorganizing pretrained embeddings thus emerges as a simple and powerful route to clustering realistic data, where classes overlap and the number of clusters is unknown.
Read the full abstract

Computer Vision

OmniSmartHome: A Multimodal Reasoning Benchmark for Smart-Home Agents

Jihoo Jung, Suho Yoo, Jeongsoo Choi et al. (+5 more)

Smart-home assistants are expected to handle diverse, realistic requests that arise in daily life. In such interactions, users often rely on the surrounding multimodal context-pointing at objects or referring to what they see or hear, leaving their requests underspecified in language alone. Existing smart-home benchmarks, however, express user requests solely through language, leaving context-dependent real-world requests underexplored. To bridge this gap, we introduce OmniSmartHome, a multimodal smart-home benchmark where each spoken request is paired with the surrounding visual and spatial-audio context, providing complementary cues to disambiguate underspecified requests. OmniSmartHome comprises 1,360 synthetic and 272 real-world episodes. We evaluate 16 omnimodal large language models (Omni-LLMs) and reveal that, while they perform strongly when speech alone sufficiently conveys the user's intent, performance drops substantially when resolving it requires reasoning over multimodal contextual cues. As a simple agent baseline, we provide PROME (PROcedural Memory for multimodal Evidence gathering), which equips agents with specialized audio-visual perception tools and procedural memory for orchestrating their use. PROME generally improves performance across six Omni-LLMs. Demos and examples are available at https://omni-smart-home.github.io
Read the full abstract

All authors Jihoo Jung, Suho Yoo, Jeongsoo Choi, Hyebin Cho, Tae Wook Haam, Hyeonggon Ryu, Sumin Park, Joon Son Chung

Fluid Dynamics

Spall Failure of Coral Skeleton beneath Gas-Laden Canopies: An Idealized Blast-Fishing Model

Sandy Hardian Susanto Herho, Agus Wahyu Jatmiko, Rizki Dimas Permana et al. (+7 more)

Improvised explosives used for fishing on shallow Indonesian reefs shatter coral skeleton, yet the damage they cause has been described largely through empirical radii and ecological surveys. We formulate an idealized model of how free gas held within a coral canopy modifies the shock loading that such a charge delivers to skeletal plates. The canopy is treated as a relaxed bubbly mixture whose shock impedance follows from the conservation of mass and momentum, and the reef is represented as a layered column of water, canopy, skeletal plate, and canopy struck at normal incidence. Three closed-form results emerge. Above a crossover pressure set by the void fraction and the stiffness of seawater, the canopy becomes nearly transparent to the shock. The impulse transmitted through any lossless layered stack is independent of the canopy, so gas redistributes the pulse in time without changing its total push. A plate carries tension after reflection from its lower face only when the canopy impedance falls below a threshold fixed by the plate thickness and the pulse duration, which defines a critical thickness. For a one-kilogram charge directly overhead, a gas-rich canopy more than triples the standoff at which a twelve-centimeter plate spalls while shortening the standoff at which it is crushed. A prescribed daily cycle of photosynthetic gas makes the same charge markedly more damaging at noon than at night. Bubble dynamics show that the canopy does not reach equilibrium within the pulse, so the results are best read as upper bounds.
Read the full abstract

All authors Sandy Hardian Susanto Herho, Agus Wahyu Jatmiko, Rizki Dimas Permana, Iwan Pramesti Anwar, Alfita Puspa Handayani, Faruq Khadami, Karina Aprilia Sujatmiko, Rusmawan Suwarman, Deny Juanda Puradimaja, Dasapta Erwin Irawan

Computer Vision

Attribution Gaps in Zero-Training LLM+OVOD Pipelines: A Fine-Grained Analysis of the CAAP–SNAP Discrepancy

Yu-Feng Yen

LAOD and similar zero-training LLM+open-vocabulary-detector (OVOD) pipelines score two things separately: class-agnostic localization accuracy (CAAP) and semantic naming accuracy (SNAP). The two consistently diverge, and nobody has asked why. This paper asks why, on the full 5,000-image COCO-Val split (27,273 detections) rather than the small subset the original work evaluated on. Object visual complexity turns out not to be the driver – small and occluded objects are, if anything, localized better than large ones. Vocabulary novelty is: once the LLM's wording falls outside the detector's native category set, localization accuracy falls from 80.9% to 31.6%. That drop is not spread evenly across unfamiliar phrasing, though. Almost all of it comes from cases where the novel wording actually names a different object than the one COCO annotated (true synonyms still score 89.3%; semantically unrelated "noise" labels score 12.0%). A closer look at a further failure subset tells a similar story: 78-88% of what looks like complete localization failure is really the model correctly finding a real object that COCO's non-exhaustive 80-category scheme simply never labeled, not hallucination. Swap the detector backbone (YOLO-World for Grounding DINO) or the LLM (Gemma-3 for Qwen2.5-VL) and both the effect and its rough size hold up, so this looks like a general property of the pipeline family rather than a quirk of one model pairing. The upshot is that a large share of the apparent CAAP–SNAP gap traces back to closed-category annotation limits rather than a real grounding failure, which matters for how we detect hallucination, analyze failure modes, and design evaluation for grounded multimodal systems meant to work in the open world.
Read the full abstract

Machine Learning

Compositional Objectives: Learning Structure in Structure

Pranavchandra Vivekananda, Sumukh Bettadapura and Ajan Subramanian

Intelligence is defined in many ways. One of these definitions defines intelligence as the pursuit of learnable novelty. However, learnable novelty can be meaningless without the ability to compose the learned structures to take action and achieve goals. Learnable novelty builds on epiplexity, which is a way to measure learnable structure in data through a bounded observer. In this paper, we investigate a closed-form spectral approximation to compute epiplexity. We use a fixed-trace constraint and find that the epiplexity objective prefers a more uniform distribution of spectral mass rather than concentrating it in a small number of directions. However, a representation may spread information across many directions without organizing that information into features useful for a particular task. To address this gap, we propose a compositional objective whose observer measures the relationships between the parts and interactions of an image. We compare it with the original spectral objective given only the masked parts. In our ImageNet training runs, the spectral objective with masked parts produces an almost maximally spread representation while achieving the strongest frozen-feature classification performance on most evaluations, more than doubling the linear-probe accuracy of the whole-image baseline. Across multiple image benchmarks, changing what the observer sees matters more than adding relation and interaction tokens. At the same time, our prediction-oriented compositional objective produces substantially better held-out observer prediction but relatively weaker classification, revealing that spectral diversity, predictability, and downstream utility are distinct properties. These results suggest that the usefulness of spectral spreading depends not only on how much structure is preserved, but on which relationships the observer makes available to the objective.
Read the full abstract

Cryptography and Security

Reading Is Not Leaking: Local, Auditable Measurement and Reduction of Inference Exposure from Public Footprints

Mahmudul Faisal Al Ameen

Anyone with a public footprint leaks facts that were never stated, and language models make the inference cheap. We present a framework for measuring and reducing this inference exposure that runs on the owner's own CPU with no language model at analysis time, instantiated on organisations and on individuals. It starts from a measurement result: scoring an inference system against the target's private truth conflates how well the system reads the record with how much the record leaks. On a 128-question instrument over sixteen synthetic firms, almost half of the questions are never answered correctly by any of six readers, four of them language models, and a majority-class guess accounts for most of every reader's score. We therefore separate reading accuracy from leakage rate and introduce an injection protocol that creates cells with known support. Our analyser combines rules, statistical solvers and a 106M-parameter encoder trained from scratch that marks verbatim evidence and never generates text; every answer carries a graded certificate whose recorded proof replays. Its certified answers are correct in 93% of resolved cases, against 49-73% for the language models' quote-backed answers, whose citations are produced alongside the answer rather than deriving it; with plain-prose articles in the record, 70% of its evidence-bearing answers rest on evidence that establishes them, against 18-56% for the models. A constrained defence that rewrites each fact's carrier as a true but coarser statement hides every single-carrier fact from four language-model adversaries at 40% lower edit cost than deletion. On sixteen synthetic people the guessing term is larger still, and a decoy planner with no language model halves the correct answers of the estimator it targets without transferring to a second.
Read the full abstract

Artificial Intelligence

ProTTT: Learning to Learn Semantic User Memory with Test-Time Training

Sejun Park, Hyoungjo Bhang, Hyein Jeong et al. (+1 more)

Personalization requires language models to capture user-specific knowledge from a growing user history. Existing context-based approaches incur increasing inference costs as user history accumulates and rely on separate retrieval or summarization stages, while parametric-based approaches often require reconstructing user representations when new user data is added. We introduce ProTTT, a profile-supervised meta-learning framework for learning semantic user memory. The memory construction starts from a shared initialization and is updated for each user through test-time training on user history, allowing it to evolve continuously as the history grows. However, since test-time training alone does not explicitly encourage the memory to capture semantic user knowledge necessary for personalization, we learn this shared initialization using textual user profiles as supervision, so that test-time training on user history captures semantic knowledge more effectively. ProTTT consistently outperforms both full history ICL and all parametric baselines across diverse benchmarks, while substantially reducing inference cost by compressing user history into a lightweight parameterized memory. Our analysis also shows that profile supervision is a reliable objective for learning semantic user knowledge and that the resulting memory can track and retain evolving user preferences, while remaining robust across different history sizes. Overall, we demonstrate the effectiveness of test-time training for personalization and establish ProTTT as a baseline for continuously evolving user memory.
Read the full abstract

All authors Sejun Park, Hyoungjo Bhang, Hyein Jeong, Yohan Jo

High Energy Physics (Phenomenology)

Dissecting the moat regime at low energies II: Correlations

Fabian Rennecke and Shi Yin

We study the effects of the moat regime on meson and quark correlation functions and the quark-antiquark potential in low-energy QCD, building on the two-flavor quark meson model in a random-phase approximation. We investigate the relation between the moat regime and Friedel oscillations in the quark-antiquark potential at large density and zero temperature. As it turns out, oscillations from the moat regime are a distinct phenomenon, arising from poles on unphysical Riemann sheets in this region. Furthermore, we find that in addition to pions and the sigma meson, also other mesons, including vector mesons and the eta, are sensitive to the moat regime. This leads in particular to an enhancement of the meson spectral function in the spacelike region. The quarks, on the other hand, appear to be essentially insensitive. This can have far-reaching consequences for the phase diagram, as for example the chiral phase boundary, including the location of the critical endpoint, can be affected in the moat regime even in absence of inhomogeneous instabilities.
Read the full abstract

Artificial Intelligence

Artificial intelligences and human scientists exhibit complementary strengths in theory building

Ke Li, Spyros I. Zoumpoulis, Phanish Puranam et al. (+80 more)

We investigate the effectiveness of artificial intelligences (AI)-specifically large language models (LLMs)-relative to human scientists at high-level cognitive tasks in social science such as theory formulation, predictions of novel empirical results, and theory revision in response to new evidence. The research domain was academic discourse regarding gender and race inequality. Our findings, comparing 25 LLMs with 13 senior researchers and 60 doctoral scholars, reveal that the AIs outperformed most humans individually on most of the present tasks, while human theories were more diverse and exhibited greater gains in predictive accuracy from aggregation. AI-generated theories were more extensively elaborated, involving additional theoretical paths and latent variables, and were rated as higher quality than human theories by independent raters blinded to source. However, this theoretical complexity was in part ornamental, in that it was not associated with more accurate predictions about empirical patterns in data; in contrast, human scientists achieved greater predictive efficiency with simpler theories. The AIs were significantly more likely than human scientists to revise their theories to incorporate new evidence; human scientists updated their beliefs in a selective way that is sensitive to prior prediction errors. We speculate that the superior processing capacity of artificial intelligences makes them especially well-suited to tasks requiring grappling with complexity, but that the greater diversity of human ideas is essential to wise crowds and collective creativity.
Read the full abstract

All authors Ke Li, Spyros I. Zoumpoulis, Phanish Puranam, Philip Parker, Matthew Eshbaugh-Soha, Izzy Gainsburg, Michael Gilead, Igor Grossmann, Britt Hadar, Yoel Inbar, Almog Simchon, Robb Willer, Rui Ai, Ruicheng Ao, Gavin J. Bala, Matthew Bidwell, Shuang Cai, Kai Chang, Skyler Y. Chen, Cory J. Clark, Irmak Dai, Abhinandan Dalal, Connor Douglas, Alexis Du, Zhehang Du, Leyun Feng, Isabel Fernandez-Mateo, Linnea Gandhi, Cyrille Grumbach, Anmol Gupta, Vansh Gupta, Maria Hademer, Jay H. Hardy III, Chen Kai Huang, Jacob Xiangyu Jin, Ufuk Keskin, Na Hyun Kim, Mert Kobaş, Byounghoon Koh, Gabrielle Lamont-Dobbin, Gregory Lanzalotto, Sun Young Lee, Dingzhe Leng, Chenjun Li, Weiyuan Li, Zeyuan Li, Zhongyuan Liang, Ning Liu, Peihong Liu, Yuhan Liu, Jiuyao Lu, Wanteng Ma, Nicolas Martinet, Natnael Mulat, Christina A. Nguyen, Khai Nguyen, Quang Minh Nguyen, Naja Pape, Chanwoo Park, Stefanos Poulidis, Jeffrey Sanchez-Burks, Michael Schaerer, Isabelle Solal, Yanbo Song, Junghyo Sun, Qingyao Sun, Rui Sun, Roderick Swaab, Kevin Tan, Dequn Teng, Michelle A. Vaccaro, Robin Vigerbaeck, Xiaomeng Wang, Randol H. Yao, Duygu Yilmaz, Shun Yiu, Ecem Yucesoy, Allen Zang, Ruijia Zhang, Xilan Zhang, Yichi Zhang, Zhanhao Zhang, Eric Luis Uhlmann

Analysis of PDEs

Formalising Linear Elliptic PDE Theory in Lean 4

Alejandro José Soto Franco and Kobe Marshall-Stevens

We formalise in Lean 4, on top of Mathlib, the solvability of the Dirichlet problem for second-order linear elliptic operators in divergence form. The machine-verified results, with no sorry in the development, include the Poincaré inequality, the existence of weak solutions by the Lax-Milgram theorem, Rellich-Kondrachov compactness, the Fredholm alternative, the spectral theorem, interior regularity estimates, and the Sobolev embedding theorem. From these results we obtain a formalisation of classical solvability for sufficiently regular coefficients and data. Our Lean library includes a self-contained theory of Sobolev spaces developed independently of existing formalisations. Throughout the paper we associate each prose statement with the named machine-checked Lean declaration that discharges it.
Read the full abstract

Computation and Language

How to Reduce Whisper Hallucination

Husein Zolkepli

Whisper is still what runs in production: one permissively licensed checkpoint, 99 languages, no per-language tuning. But it writes sentences nobody said. On 42 clips of pure room tone, whisper-large-v3 emits words on 61.9% of them and emits something on 100%. The usual response is to distil a student from Whisper pseudo-labels, filtering the hallucinations out of the corpus first, which deletes the evidence while leaving the behaviour in place: the student is fitted to the subset where the teacher was right, and inherits a failure mode absent from its own training data. The second response, adding non-speech audio so the model learns to stay quiet, already ships and works, but teaches suppression without discrimination. The checkpoint best on every non-speech arm here is also the one that recovers the fewest genuinely spoken phrases and deletes 58% of repeated speech. The teacher has to be fixed, with both halves of the signal. Our benchmark scores both at once: 11,852 clips over eight arms, 6,267 synthetic positives, 8,296 clips of real audio that made a production model fail, and FLEURS in 58 languages. We collect 40,891 hallucination phrases in 100 languages, choose a text-to-speech system by measurement, and synthesise those phrases as positives, so the model meets the same text both as something to suppress and as something to transcribe. Across 33 matched pairs of fine-tunes differing only by those positives, adding them lowers word emission on real voice-free audio in 31 pairs and raises phrase recovery in 32; among the 30 pairs that hold English accuracy it is 30 out of 30. The best checkpoint takes hallucination on silence from 61.9% to 2.4% and words over real voice-free audio from 99.9% to 47.8%, while raising phrase recovery from 69.8% to 82.7%. The benchmark, lexicon and synthetic corpus are released at https://huggingface.co/datasets/Scicom-intl/Whisper-Hallucination
Read the full abstract

Computer Vision

CFCH: Coarse-Fine Collaborative Hierarchical Learning for Anterior Segment Disease Analysis

Peng Wang, Haohan Zou, Yanlin Wu et al. (+3 more)

Accurate classification of anterior segment diseases is crucial for ophthalmic screening and diagnosis. However, slit-lamp image analysis remains challenging due to substantial variability in imaging conditions and the intrinsic anatomical-disease hierarchy of ocular pathologies. Existing methods typically formulate this task as a flat multi-class classification problem, ignoring the structured dependency between anatomical regions (e.g., cornea, conjunctiva, and lens) and disease manifestations.To address these limitations, we propose CFCH, a Coarse-Fine Collaborative Hierarchical learning framework that explicitly models anatomical context and disease semantics through a dual-branch architecture. To enable effective cross-granularity collaboration, CFCH introduces semantic and cross-granularity attention consistency constraints, encouraging aligned yet complementary feature learning across branches. In addition, we construct AS-9K, a large-scale anterior segment dataset with 8975 images covering 12 common disease categories. To the best of our knowledge, AS-9K is the largest publicly available dataset for anterior segment image classification. Extensive experiments on two anterior segment datasets demonstrate that CFCH outperforms state-of-the-art methods. Qualitative visualizations further show more focused and lesion-relevant activation responses, validating the effectiveness of the proposed framework. Code will be available at https://github.com/ybupengwang/CFCH.
Read the full abstract

All authors Peng Wang, Haohan Zou, Yanlin Wu, Xueshuo Xie, Yan Wang, Tao Li

Optimization and Control

Smooth or Separable? Sparse Block Acceleration for Entropy-Regularized Linear Programs

Irina Podlipnova, Maxim Mashtaler, Artem Agafonov et al. (+2 more)

We study entropy-regularized linear programs with sparse affine constraints and compare two exact dual representations induced by whether a redundant normalization constraint is retained or eliminated. Eliminating it yields a partially separable sum-exp dual with sparse affine factors but unbounded curvature. Our main result is an accelerated randomized block method whose acceleration preserves sparse data-access cost. The key ingredient is a global entropy-specific Hessian–gap bound that remains valid for signed constraint matrices, admits computable block-overlap refinements, and provides deterministic curvature certificates before sampling. This geometry supports square-root importance sampling and an exact lazy implementation in which sparse arithmetic cost is weighted by local curvature rather than by a worst-block factor. We further introduce a gap-dependent curvature dimension that connects the generalized-smooth regime to the classical effective-rank intuition for coordinate methods. Retaining the normalization constraint yields a log-sum-exp dual with globally bounded higher derivatives, enabling gradient-regularized and cubic-regularized Newton methods. Experiments compare the resulting methods in terms of coordinate epochs, arithmetic operations, and wall-clock time, separating the effects of acceleration, touched-data sparsity, and problem-specific structure.
Read the full abstract

All authors Irina Podlipnova, Maxim Mashtaler, Artem Agafonov, Yuriy Dorn, Alexander Gasnikov

Dynamical Systems

Recurrent Histopolating Fractal Functions

Pavjeet Singh, Vinay Kumar and S.K. Katiyar

Recurrent fractal interpolation extends the usual fractal interpolation framework by introducing a recurrence structure. Fractal histopolation, on the other hand, is concerned with the area-matching properties of fractal functions. In this paper, we combine these two approaches and introduce a recurrent fractal histopolants.
Read the full abstract

Computation and Language

Language as an Independent Information Layer: A Conceptual Model of Communication, Cognition and Decision-Making

Anastasiia Alifanova and Elena Benderskaya

Based on an analysis of the role of language in thought and communication, this article proposes a new concept for designing corporate knowledge bases. The concept integrates the probabilistic vector space of a corporate vocabulary, reflect-ing industry specifics, subject focus, terminology, and culture, with traditional ontological modeling. This combination enables the efficient extraction of knowledge from accumulated corporate documents while strictly accounting for specific business processes. Consequently, this concept bridges statistical and semantic (cause-and-effect) methodologies. Furthermore, analyzing the projec-tions of probabilistic spaces and causal relationships can help identify bottlenecks in business logic. As a dynamic system, language functions as a separate, inde-pendent layer within the overall information architecture. Introducing a dynamic component into the probabilistic space of word distribution allows it to be mod-eled as a multidimensional solution space for various problem formulations. In this context, input data defining the problem conditions serve as control parame-ters for dynamic transformations.
Read the full abstract