Materials Science
Papers
Fresh from arXiv, typeset properly. Highlighted words trade on the market.
All of arXiv
Friday 28 August
CeCrB$_4$ belongs to the family of MTB$_4$ ternary metal borides, some of whose magnetic phases have recently been identified as quantum spin dimers. Employing density functional theory within the GGA+$U$ framework, this work investigates the key ground-state properties of this system, including its ferromagnetic ground state, the mixed-valence nature of cerium, and the charge transfer from the cationic cerium–chromium layers to the anionic boron layers. In particular, we focus on elucidating the bonding mechanism within the structurally embedded Cr-Cr dimers. Specifically, we evaluate a previously proposed interpretation that the localized Cr $3d_{z^2}$ states are molecular-like bonding and antibonding states. The calculated dependence of the $3d_{z^2}$ electronic structure on the intra-dimer distance correlates with the level-splitting characteristics known for classical homonuclear diatomic molecules.
Machine Learning
Efficient Online Continual Foundation Model Fine-Tuning for Predictive Process Monitoring
Predictive Process Monitoring (PPM) models are increasingly deployed in dynamic environments where concept drift causes the underlying process distribution to shift over time. While recent work has moved toward online continual learning, existing methods train compact, task-specific networks entirely from scratch, leaving a persistent cold-start problem. Foundation Models (FMs) offer a compelling solution to this problem, but their continual fine-tuning in the process mining domain remains unexplored. We propose COMPASS (Continual Online foundation Model-based PPM with Adaptive SubSpaces), the first framework for online continual fine-tuning of FMs for PPM. COMPASS adapts loss-plateau drift detection to autonomously identify task boundaries in event streams and maintains a unified knowledge subspace including both pre-trained and task-specific directions. We evaluate our approach on nine event streams covering synthetic and real-world concept drift scenarios, across task-free and task-aware settings with multiple backbones and with consistent hyperparameter tuning across all methods. Our approach outperforms three SOTA non-FM competitors and two update strategy baselines, with particularly strong gains on streams exhibiting recurrent drift and complex, long-running cases, while incurring acceptable computational overhead compared to the non-FM competitors.
Read the full abstract
Machine Learning
D-TAIA: Domain-Aware LLM Adaptation for Multi-Task Predictive Process Monitoring
Predictive Process Monitoring (PPM) enables organizations to forecast future process behavior, such as the next activity and remaining time of ongoing cases. In practice, three conditions cause existing methods to degrade, namely data scarcity, high process entropy and distributional shift. While Foundation Models (FMs), especially Large Language Models (LLMs), offer a new paradigm through broad sequential reasoning, adapting them to multi-task PPM under these conditions remains an open challenge. Existing FM-based approaches either lack mechanisms for handling distributional shift or rely on direct regression heads that can be structurally misaligned with continuous time prediction tasks. This paper introduces D-TAIA (Domain-aware Training and Attention-based Inference Architecture), a framework for a joint next activity and remaining time prediction task via parameter-efficient fine-tuning of an FM backbone. Our approach combines domain-aware triplet loss (DATL) pre-training with FAISS-based nearest neighbor retrieval for remaining time prediction, and adopts the TAIA inference strategy to preserve pre-trained sequential reasoning during fine-tuning. Evaluated across four real-world event logs, D-TAIA consistently shows SOTA or competitive performance compared to a fine-tuned LLM and a recurrent neural network baseline. Ablation studies confirm that techniques from NLP and computer vision can be transferred effectively to PPM with only a 10M-parameter backbone, though component contributions vary by dataset entropy.
Read the full abstract
Quantum Physics
Direct Adaptive Certification of High-Dimensional Entanglement with Bell Tests
Entangled photons play a crucial role in quantum applications, and determining and characterising their entanglement is vital to using them effectively. High-dimensional entangled states offer richer possibilities, but their additional measurement degrees of freedom make them increasingly demanding to characterise. However, adaptive Bell-test methods based on complex simultaneous perturbation stochastic approximation (CSPSA) have so far focused mainly on qubits. Here we numerically investigate a Bell-inequality-violation-based method for detecting entanglement in unknown quantum states. We extend CSPSA to high-dimensional Bell testing by using the Collins-Gisin-Linden-Massar-Popescu (CGLMP) inequality for bipartite qudits. The resulting protocol can detect Bell-nonlocal correlations in unknown entangled states, whether pure or mixed, without first reconstructing their density matrix. Using 100 optimisation iterations in each of 100 independent finite-shot runs per number of dimensions d, we demonstrate certified CGLMP violations throughout d=2-8. For isotropic mixed states tested at a visibility of just 0.05 above the standard-Fourier violation threshold, we likewise observe confidence-certified CGLMP violations throughout the range of d studied. We compare this direct stochastic approach with quantum state tomography, the standard method for characterising an unknown state. In the matched benchmark, CSPSA uses fewer measurement configurations per attempt from d=6, whereas tomography requires fewer detected pairs per certified result through d=8. We also derive the phase dependence of the CGLMP parameter and clarify the features of its landscape that govern the adaptive search. Because the measurement-setting cost of each CSPSA iteration is independent of dimension, the method offers a particularly attractive route to the certification of high-dimensional entanglement.
Read the full abstract
Differential Geometry
A note on the positivity of $Q$-curvature via the continuity method
Suppose $(M, g)$ is a smooth, closed, $5$-dimensional Riemannian manifold with positive Yamabe invariant $Y(M, [g]) > 0$ and positive Yamabe-type $Q$-curvature invariant $Y_{4}^{\ast}(M, [g]) > 0$. Using the continuity method, we prove the existence of a metric in the conformal class $[g]$ with positive scalar curvature and positive $Q$-curvature, assuming an additional condition on an “initial metric" in the class $[g]$.
Read the full abstract
Artificial Intelligence
REINS: Refusal-Enhanced Inhibitory Steering with Sparse Autoencoder Features
Steering with Sparse Autoencoders (SAEs) offers a lightweight inference-time path for adapting the behavior of large language models without retraining. By exposing sparse and interpretable features, SAE steering provides a promising interface for safety control that guides harmful continuations toward refusal. However, we observe that complex wrappers can still undermine existing SAE steering methods on harmful prompts. To evaluate this failure mode systematically, we construct Generalized Undercover Instruction Safety Evaluation (GUISE), a dataset of harmful prompts with complex wrappers. Existing single direction SAE steering methods do not reliably produce refusals on harmful prompts, suggesting that refusal enhancement alone can be too weak when the harmful continuation path remains active. This motivates us to propose Refusal-Enhanced INhibitory Steering (REINS), which suppresses harmful continuation features and enhances safe refusal features in the same SAE feature space. Experiments on GUISE and other datasets show that prior methods either intervene too weakly or achieve only apparent safety through collapse, while REINS substantially reduces harmful responses, markedly improves safe refusals and largely preserves general capabilities.
Read the full abstract
High Energy Physics (Phenomenology)
HEPToolBench 1.2: Testing How Reliably Language Models Can Drive Particle Physics Software
Scientists increasingly want to drive research software by natural-language request, but fluent output helps only if it becomes a correct machine-readable artifact. We introduce HEPToolBench, a benchmark of 28 collider-simulation tasks scored by deterministic, task-specific scorers, plus a three-task structured-debugging extension. We evaluate 42 deployments, from small locally served open-weight models to hosted frontier systems. The central experiment compares direct generation of native HEP-tool syntax with a schema-mediated interface where the model returns a typed representation that deterministic software serializes. Across five matched requests, the mean score rises from 0.418 to 0.902 and task passes from 21/210 to 159/210, with 41 of 42 deployments improving; eleven deployments, seven locally served open-weight models, go from 0/5 passes under native syntax to 5/5 under the structured interface. Values use the corrected v1.2.1 contract, in which the two native scorers that had enforced operational conventions absent from their own prompts are rescored prompt-faithfully cohort-wide. One asymmetry remains: native scorers score the recorded response verbatim and reject Markdown-fenced replies, whereas structured scorers recover JSON from such wrappers; no full-cohort rescoring under a single symmetric extraction rule has been done. An audit of archived responses from 17 locally served deployments shows this does not explain their gains: passes remain six to ten times more frequent under the structured interface when both sides use the same extraction rule. A task pass alone does not guarantee runtime or scientific viability. Within this scope, moving syntax generation into deterministic software can substantially improve reliability for both small local and frontier models. Prompts, scorers, responses, and regeneration scripts are released for independent evaluation and extension.
Read the full abstract
Mesoscale and Nanoscale Physics
Quantum Geometric Origin of Nonlinear Current Induced Orbital Magnetization
Electric generation of magnetization is a focus of condensed matter research, and has recently been advanced into the nonlinear regime. However, due to the nonlocal nature of orbital magnetism, how to properly formulate nonlinear current-induced orbital magnetization remains a fundamental challenge. Here, we develop the proper theory for this effect. This is based on the microscopic derivation of field-corrected orbital magnetic moment of a Bloch electron, a critical missing piece in the present theory. We show that the quantum geometric origin of this phenomenon lies in both the anomalous orbital polarizability and the Berry-connection polarizability, which often provide competing contributions. Combining our theory with first-principles calculations, we predict significant, experimentally accessible nonlinear orbital magnetization generated in strained bilayer graphene, monolayer 1T' $\mathrm{MoS_2}$ and $\mathrm{MoTe_2}$. Remarkably, nonlinear orbital magnetization can dominate over its spin counterpart in materials with topological band features, irrespective of the spin-orbit coupling strength.
Read the full abstract
Software Engineering
Adaptive Strategy Generation for Boundary Value Exploration Beyond Numeric Inputs
Software behavior often changes abruptly at boundaries between input regions, and these transitions are known to be fault-prone. Boundary Value Exploration (BVE) automates boundary discovery by searching for pairs of similar inputs that nevertheless trigger different program behaviors. Existing automated BVE techniques rely on mutation operators hand-engineered for each input type, or even for each function under test, which has confined their use to numeric inputs. We present ABEX, an agentic LLM-based framework that replaces operator engineering with adaptive strategy generation: specialized LLM agents propose, select, and execute boundary-exploration strategies, guided by execution feedback and a quality-diversity (QD) archive. Because strategies are expressed in natural language, they can encode both type-level and function-specific knowledge, and effective strategies can even be stored and reused. We evaluate ABEX in a black-box setting on 20 functions with numeric, string, array, and mixed inputs. On numeric functions, ABEX outperforms a state-of-the-art QD method on 10 of 11 functions, with average QD-scores 11.7x higher. On non-numeric functions, addressed here for the first time in automated black-box BVE, ABEX discovers domain-aligned boundary behaviors for all subjects. Mutation testing shows the discovered boundaries are fault-revealing: with equally sized test suites, ABEX reaches an average mutation score of 86.2% versus 61.9% for the QD baseline, and kills nine times as many hard-to-detect stubborn mutants. An ablation study identifies adaptive strategy generation as the primary driver of these gains.
Read the full abstract
Artificial Intelligence
Stay Within Your Bounds: Distance-Guided Decoding for Guaranteed Context-Free Grammar Compliance
Grammar-constrained decoding helps large language models produce syntactically valid structured outputs, such as code, JSON, and SQL. For context-free grammars, many practical decoders enforce local prefix feasibility: each token must keep the current prefix extendable to some valid completion. Yet, under tokenizer-grammar mismatch and finite token budgets, feasible prefixes may still fail to reach acceptance. We propose a lookahead-guided decoding framework for context-free grammars based on pushdown automata. Offline, we compute bounded pushdown summaries with reachability labels and upper-bound distances to acceptance. Online, these estimates guide horizon-aware pruning and beam search. The resulting decoder is syntactically sound: every output is accepted by the target grammar. Experiments on JSON, SQL, and Linear Temporal Logic (LTL) show both consistent syntactic validity and improved completion quality over existing baselines.
Read the full abstract
Artificial Intelligence
Generative AI Alignment with Hinduism's Theological Plurality and Sacred Representation
Generative AI systems are increasingly used to answer personal questions and mediate everyday practices, including religion. However, existing discussions around AI alignment and ethics have largely centered secular, Western, and Abrahamic assumptions about religion, offering limited attention to other faith-based traditions. In this paper, we examine how Hindu users engage with generative AI systems in relation to their religious knowledge, belief, and practice. Drawing on 15 semi-structured interviews with Bangladeshi Hindu participants, we analyze how users interpret AI-generated religious representations, scriptural explanations, devotional interactions, and synthetic religious media. We found that AI can be both accessible and ethically troubling. While AI supported scriptural inquiry, devotional visualization, and religious storytelling, our study also identified concerns about theological flattening, cultural misrepresentation, devotional manipulation, and the simulation of sacred presence and authority. We conclude by arguing that religious alignment in generative AI requires interpretive alignment: systems that disclose their limits, preserve plurality, and avoid simulating sacred authority and sycophantic personalization.
Read the full abstract
Functional Analysis
The Matricial Univariate Rational Truncated Moment Problem
We solve the matricial univariate rational truncated moment problem on the real line and on a half-line. We give explicit necessary and sufficient conditions for the existence of a positive matrix-valued representing measure and show that every solvable problem admits a finitely atomic representing measure whose total atomic multiplicity equals the rank of the associated moment matrix. The rational problem is reduced to an ordinary matricial moment problem with the additional requirement that the representing measure avoid the real poles of the rational data. The main new ingredient is a simultaneous prescribed-node result: the smallest attainable multiplicities at finitely many prescribed points can be realized by a single minimal representing measure. Our proofs are constructive and use flat extensions, block-column relations, and localizing conditions.
Read the full abstract
Instrumentation and Methods for Astrophysics
The PLACID active coronagraphic imager instrument: commissioning status
The world's first adaptive stellar coronagraph, the Programmable Liquid-crystal Active Coronagraphic Imager for the 4-m DAG telescope (PLACID) uses a spatial light modulator operating from H- to Ks-band to dynamically adjust the focal-plane phase mask entirely in software. Positioned between the TROIA XAO system and the DIRAC infrared detector, PLACID was installed on the Nasmyth platform of the Turkish 4-m DAG telescope in 2025, followed by optical alignment and successful preliminary acceptance tests with the calibration light source in early 2026. On-sky commissioning is expected, as soon as the TROIA XAO system will be operational. When on-sky, it will enable high-contrast imaging of exoplanets, brown dwarfs, disks, and binary systems, being able to easily deploy any desired phase pattern in the focal plane. Upcoming features include self-calibration of non-common path aberrations, coronagraphic nulling of binary-stars, and coherent differential imaging. We present the preliminary acceptance procedures and internal alignment results, discovery-space estimates, new binary star features and observation tools, to be ready for first science by late 2026/early 2027.
Read the full abstract
Instrumentation and Methods for Astrophysics
Design and implementation of an SLM-driven coherent differential imaging scheme
Direct imaging of exoplanets from the ground remains fundamentally limited by fast-evolving post-adaptive optics (AO) residual wavefront errors, and by quasi-static speckles arising from non-common path aberrations (NCPAs) downstream of the AO wavefront sensor. To first order, both sources of contrast degradation remain coherent with the stellar light, and recent studies have shown that coherent differential imaging (CDI) techniques can address their quasi-static and slowly evolving components. Recent laboratory demonstrations using focal plane phase diversity with an active LCoS spatial light modulator (SLM) have further established that local phase modulation can distinguish coherent speckles from incoherent astrophysical signals. Here we present a framework for the design and implementation of an SLM-based CDI integrator. Through simulations, we study the effects of varying the number of phase steps, modulation ring width, planet separation, and planet-to-star contrast. The numerical results are further supported by an initial laboratory validation. In the future, with the emergence of ultra-low-noise fast near-infrared detectors such as avalanche photodiode and microwave kinetic inductance detector (MKID) arrays, time-domain focal-plane CDI detection schemes could potentially also be able to suppress rapidly varying post-AO residual wavefront errors.
Read the full abstract
Computational Engineering
Towards Stellarator Geometry Optimisation for Nuclear Fusion
We present a local geometry refinement method and a 2D latent representation that took us to the top of the ConStellaration leaderboard on the geometric task in May 2026.
Read the full abstract
Dynamical Systems
Invariant Sphere Theorem and Ring-Coupled Systems
In this work, we show how heteroclinic networks can arise in a simple class of network dynamical systems through the application of the Invariant Sphere Theorem. Ring-coupled systems are ODE networks in $\mathbb{R}^n$ in which each variable $x_i$ interacts only with its predecessor $x_{i-1}$. We derive conditions on the coefficients of a cubic polynomial that guarantee the existence of a globally attracting invariant sphere via the Invariant Sphere Theorem. Moreover, setting one of these coefficients to zero, we identify conditions on the remaining coefficients that guarantee the existence of a heteroclinic network on the invariant sphere. Focusing on the case $n=3$, we investigate perturbations of the vanishing coefficient in a neighbourhood of the heteroclinic network. Under such perturbations, the heteroclinic network is destroyed and periodic orbits emerge. When the original heteroclinic network is asymptotically stable, the resulting periodic orbits shadow the network. In one parameter regime, a unique attracting periodic orbit appears and shadows the entire heteroclinic network. By contrast, when the heteroclinic network is not stable, repelling periodic orbits arise that shadow only part of the heteroclinic structure, namely half of the heteroclinic connections. Symmetry plays a fundamental role throughout the analysis. Exploiting the symmetries of the system, we reduce the heteroclinic network to two homoclinic orbits. Furthermore, the local dynamics near the heteroclinic network can be studied on a quotient space consisting of an annulus attached to a Möbius band, with each homoclinic orbit lying on one of these components. This reduction provides a geometric framework for understanding the bifurcations and the emergence of the periodic dynamics.
Read the full abstract
Information Theory
Shortest self-orthogonal and LCD embeddings of linear codes over Fq+uFq
This paper determines the exact lengths of shortest self-orthogonal and LCD embeddings of linear codes over $\mathbb{F}_q+u\mathbb{F}_q$. By decomposing Gram matrices over $\mathbb{F}_q+u\mathbb{F}_q$ into pairs of symmetric matrices over the finite field $\mathbb{F}_q$, the embedding problems are reduced to the congruence classification of symmetric and alternate matrices over finite fields. Complete formulas for the shortest self-orthogonal embedding length are obtained, with two distinct cases arising in both even and odd characteristic. We also show that every self-orthogonal code over $\mathbb{F}_q+u\mathbb{F}_q$ with nonzero free rank can be viewed as a shortest self-orthogonal embedding of another code. We use Witt theory to construct all shortest self-orthogonal embeddings. A complete characterization of shortest LCD embeddings is also established in terms of invertible and arbitrary matrices of prescribed sizes appended to a generator matrix. Examples of self-orthogonal and LCD embeddings with the largest minimum distance for the code considered are also presented, some of whose Gray images are optimal codes over $\mathbb{F}_q$.
Read the full abstract
Statistical Methodology
Exact two-sided p-values in natural exponential families: coincidence, non-uniqueness, and sample-size stability
We study the non-uniqueness of exact two-sided $p$-values in continuous one-parameter natural exponential families (NEFs). For directed one-sided problems, the tail $p$-value agrees with the $p$-values using UMP, UMPU, and likelihood-ratio (LR) tests. For a two-sided simple null, we distinguish four constructions: equal-tail, density-ordered, UMPU, and LR $p$-values. At a fixed null parameter, UMPU and equal-tail $p$-values coincide if and only if the null law is symmetric about its mean; under a regular two-branch density-level condition, the same fixed-null symmetry characterization holds for UMPU versus density ordering and equal-tail versus density ordering. Requiring any of these coincidences throughout the NEF characterizes the Gaussian family. We combine these results with the theorem of Bar-Lev, Bshouty and Letac that UMPU and LR $p$-values coincide throughout a continuous NEF precisely for the normal, gamma and inverse-Gaussian families. We also investigate the two LR pairings not covered by those results. If equal-tail and LR $p$-values coincide throughout a NEF satisfying our standing regularity assumptions, then $(V^{2/3})^{\prime \prime \prime }=0$ on the mean domain. The same coincidence also forces an explicit density-at-the-mean identity. For an i.i.d.\ sample with canonical sufficient statistic $T_n=\sum_{i=1}^nX_i$, persistence of equal-tail-LR coincidence throughout the family along an unbounded sequence of sample sizes forces Gaussianity. A corresponding density-LR statement is given conditionally on an explicitly stated differentiated local Edgeworth expansion. Finally, inverse-Gaussian and hyperbolic-secant examples quantify numerical $p$-value differences, disagreement of rejection decisions, and differences in power.
Read the full abstract
Mathematical Physics
Augmented Star Products and their Applications
In this paper, we introduce a generalized framework for star products that preserve associativity while not necessarily obeying the canonical commutation relations. Within this framework, we formulate the augmented star product and investigate the associated augmented star exponentials. We further demonstrate several applications of these constructions to special functions and to problems arising in quantum physics.
Read the full abstract
Computer Vision
RASA: Disentangled Spatial-Motional Priors for Cross-Identity Character Animation
Cross-identity character animation aims to drive a target identity from a reference image to follow the motion of a source character from a driving video. The core challenge lies in the inherent entanglement of two capabilities: cross-identity spatial mapping (aligning position, scale, and skeletal proportions) and motion control (refining joint articulation, volumetric consistency, and view coherence). We introduce Reference-Aware Structural Alignment (RASA), a framework that disentangles spatial mapping from motion control by injecting structured priors into a Diffusion Transformer (DiT). Our approach has two stages. First, a Spatial Prior Calibrator (SPC) fuses reference identity with driving pose to generate a spatially grounded initial noise latent, ensuring correct positioning, scaling, and alignment with the driving skeleton. Second, an Inherent Motional Guider (IMG) encodes shape-agnostic SMPL articulation parameters into a semantic motion vector beyond appearance-biased 2D keypoints. Injected into intermediate DiT layers, this vector complements the base pose condition for anatomically consistent articulation and view-aware volumetric refinement. We curate CIM-Bench, a high-quality benchmark with rigorous curation, for evaluation. Extensive experiments show RASA significantly outperforms state-of-the-art methods in motion fidelity and visual quality. Our work establishes a new paradigm showing disentangled spatial and motional priors are key to robust character animation. Project page: https://hidream.ai.github.io/RASA/
Read the full abstract
Computer Vision
Focus Where It Counts: A Salience-Driven Vision-Language Model for Low Vision Assistance
Vision-language models (VLMs) are rapidly progressing and offer promising capabilities for assistive technologies supporting persons with blindness or low vision. However, existing VLMs are primarily designed for general-purpose captioning and do not explicitly model human perceptual priorities, thereby limiting their ability to emphasize the most relevant information in a scene. To address this gap, we propose a salience-driven captioning framework that prioritizes scene elements according to their importance for human-centered assistance. We curate three salience-aware datasets, namely, Salience COCO, Salience Flickr, and Salience VizWiz, with object-level salience annotations designed to reflect the visual information most relevant to low vision users across different environments. Building on these datasets, we introduce Salience-LLaVA, a salience-aware VLM that incorporates salience cues to generate captions in which important elements are mentioned in the order of importance. Our work makes four main contributions. We build salience-aware datasets verified by low vision participants, propose Salience-LLaVA to describe objects in the order of importance, introduce SCMI to evaluate ordering accuracy, and deploy the system on assistive glasses to demonstrate real-world practicality. Code and datasets are available at: https://github.com/topo-focus/Topofocus
Read the full abstract
Signal Processing
Beam scheduling policy for communications and PNT services from LEO satellites
Future low Earth orbit (LEO) constellations are expected to provide positioning, navigation, and timing (PNT) as a native service alongside broadband, removing the GNSS dependency inherited by 5G non-terrestrial networks. This is known in the literature as fused PNT. This paper asks how a multibeam satellites in a LEO constellation should share their beams and power between a communication service (COM) and PNT in the scenario when communications keep full priority. Each beam on a satellite can provide a communication service, a PNT service or both (this under certain limitations). Then, the PNT service is evaluated by the share of area the reach $95\%$-of-time availability for a fixed maximum power available on each satellite. This paper present this PNT availability question as a beam-power budget, as the PNT beams do not need to transmit at the same power as the communication beams thanks to the processing gain on the receiver. Two satellite beam scheduler policies are compared: one where on a cell the beam serving COM demand is exclusive for COM, therefore the PNT service in this cell is provided by the other satellites in line of sight, it requires no signal modification and attains $70.5\%$ / $88.2\%$ availability under homogeneous/population-weighted traffic profiles; and a second policy, the COM beam can provide also PNT signals, this in-beam ranging embeds the ranging signal in the COM waveform, reaching the ceiling at the cost of waveform redesign. The availability gap between the two policies is concentrated on the thin-overlap equatorial belt.
Read the full abstract
Computer Vision
WALDO: One-Shot Exemplar-Conditioned Object Detection in Cluttered Scenes with World-Model Features
Finding a particular object instance in a cluttered scene from a single reference image and saying when it is not there is typically given to large vision-language models. We ask whether the representations of a world model add something that the strongest self-supervised image features do not have. We introduce WALDO, a one-shot detector that finds a specific object instance in a cluttered real scene from a single reference image. Instance detection has usually needed per-object training. The alternative is a category-level open-vocabulary detector or vision-language model, which can find a mug but not this mug. We show that a lightweight conditioned head on frozen self-supervised features is enough to localize unseen instances without any per-object adaptation. We also find that image features and world-model features are complementary: DINOv3 and the video world model V-JEPA 2.1 succeed on different queries. There is one catch. Used the way video models are usually applied to still images, V-JEPA's dense features underperform. Reading them through the model's image pathway improves their localization. Combining the two backbones with a validation-tuned late fusion modestly improves DINOv3's localization of unseen instances without hurting its ability to tell whether the object is in the scene. A strong vision-language model and an image-guided open-vocabulary detector almost never localize the specific instance, and take far longer per query. Our fusion with world-model features strengthens localization from image features alone.
Read the full abstract
Populations and Evolution
Linking individual bioenergetics to ecosystem dynamics with integral projection models
Linking individual-level processes, such as survival and reproduction, to ecosystem dynamics is challenging due to interactions among populations and species. These interactions depend on organism traits, such as age, body size, and mass, and are influenced by fluctuations in population structure. Integral projection models (IPMs) have advanced our understanding of dynamics arising from population structure in continuous traits. Yet, these models have only been applied to relatively simple systems, featuring one or two species and spanning a fraction of the trophic levels observed in natural food webs. We extend these models in four ways. First, we provide a general multi-species IPM where population structure can be included in primary producers and primary and secondary consumers. Second, we link individual-level processes to ecosystem functioning by mapping fluxes of biomass between species to fluxes within individuals, using bioenergetic principles for allocation to maintenance, reproduction, or growth. Third, we introduce a nutrient recycling loop through which decomposers break down organic matter into nutrients that enable plant growth. Finally, we provide an efficient fitting algorithm to calibrate the model with time series of population sizes. We showcase our approach by parameterising and fitting the model to counts of elk, bison, wolves, and cougars in northern Yellowstone National Park to study the impact of predator extirpation and recovery. Our model predicts that bison decrease following predator removal and recover following predator recovery, suggesting a positive contribution of wolves and cougars to the increase of bison in northern Yellowstone. It also reproduces indirect effects of predators on woody deciduous vegetation regeneration. Our approach is suitable for a broad range of ecosystems, with applications for population management and ecological forecasting.
Read the full abstract
Robotics
Probabilistic Multi-Robot Gas Source Localization with Uncalibrated Sensors: A Distributed Estimation Approach
Estimating environmental states with multi-robot systems becomes particularly challenging when robots are equipped with uncalibrated and therefore heterogeneous sensors, whose nonlinear and inconsistent responses prevent reliable information fusion. In this paper, we propose a distributed probabilistic framework for source localization tasks that enables calibration-free estimation in the presence of sensor heterogeneity. The key idea is that each robot independently estimates a local belief using a rank-based feature that captures the relative evolution of observations and is invariant to sensor scaling and nonlinearities. These local beliefs are then fused through a product of experts formulation to obtain a consistent global estimate across the team. To further improve the efficiency of team coordination, we introduce an informative region allocation and path planning strategy that reduces redundant exploration while balancing exploration and exploitation. We validate the proposed framework using high-fidelity simulations with realistic gas sensor models. Results demonstrate that our method significantly outperforms a benchmark method based on standard measurement aggregation, achieving reliable source localization accuracy despite strong sensor heterogeneity. More broadly, this work demonstrates how calibration-free sensing representations can be effectively extended to distributed robotic systems, paving the way for their application to other estimation tasks involving heterogeneous sensors.
Read the full abstract
Robotics
PAMoR: Parameterized Affective Motion Generation in Real Time for Humanoid Robots
People read a humanoid robot's motion in social settings not only for the action performed but for the affect conveyed. Motion carrying that affect has so far been generated for human avatars, where style is taken from a reference clip or an emotion word, neither of which can be quantitatively parameterized. We present PAMoR, which turns affect into a measured control parameter: a valence-arousal (V-A) coordinate computed natively on robot kinematics. It is obtained in closed form from postural expansion and movement energy, and these measurements serve directly as generation conditions, with no human annotation. An action prior and two affect priors, trained in a shared latent space, are composed at each denoising step: the action prior fixes what is performed, the affect priors modulate how. Whole-body motion rolls out autoregressively on a 29-DoF Unitree G1 in real time, with action and affect both editable. Generated motion tracks the commanded V-A over its full range while text-to-motion fidelity still matches text-only baselines. In a perceptual study, raters identify the commanded emotion on 0.38 of trials, above both baselines and approaching the 0.44 reported for acted human bodies.
Read the full abstract
Multimedia
MuSP-Bench: Advanced Multimodal Benchmarking of Music Understanding across Score and Performance
Musicians commonly communicate musical intent through scores and performances. To investigate whether models can reason across both modalities, we introduce MuSP-Bench, a human-authored benchmark of 490 questions assessing understanding across Musical Scores and Performances, both individually and in combination. The benchmark spans interpretive and long-horizon reasoning across classical piano and orchestral works, represented through ABC notation, score images, MIDI, and audio. Our evaluation of frontier multimodal large language models reveals substantial difficulties in understanding musical scores and even greater challenges in reasoning about audio.
Read the full abstract
Group Theory
Maximal subgroups of homeomorphism groups
We show that the homeomorphism groups of the following spaces have precisely $2^{2^{\aleph_0}}$ maximal subgroups: the rational numbers $\mathbb{Q}$, the Baire space $\mathbb{N}^{\mathbb{N}}$, the space $\mathbb{N}\times 2^{\mathbb{N}}$ where $2^{\mathbb{N}}$ is the Cantor set, the ordinal $\omega^2$ under its order topology, and the Sorgenfrey line $\mathbb{S}$. More generally, we find sufficient conditions on a group $G$ acting on a topological space which imply that $G$ has at least $2^{2^{\aleph_0}}$ maximal subgroups. Moreover, if the groups $\operatorname{Homeo}(\mathbb{Q})$ and $\operatorname{Homeo}(\mathbb{N}^\mathbb{N})$ are equipped with the pointwise topology, then it is shown that $\operatorname{Homeo}(\mathbb{N}^\mathbb{N})$ has precisely $2^{\aleph_0}$ open maximal subgroups, and $\operatorname{Homeo}(\mathbb{Q})$ has precisely $\aleph_0$ open maximal subgroups and $2^{\aleph_0}$ closed maximal subgroups.
Read the full abstract
Analysis of PDEs
Detection of a moving point-like scatter by using moving receivers and emitter
We consider an inverse scattering problem for the scalar wave equation in which the point emitter, a point-like scatterer, and multiple receivers are all moving. The goal is to reconstruct the trajectory of a moving point-like scatterer from time-dependent measurements of the scattered field. To this end, we establish a rigorous point-interaction model for the moving scatterer in the time domain, proving well-posedness of the forward scattering problem, including existence, uniqueness and continuous dependence of the scattered field on the scatterer trajectory and scattering parameter. By exploiting the retarded-time structure, we introduce distance functions that connect emission, scattering, and observation processes, and show that they satisfy a coupled system of nonlinear ordinary differential equations determined by the measurement data. Based on this formulation, we propose a reconstruction algorithm that combines initial localization with the solution of the derived ODE system. Numerical experiments demonstrate that the method is accurate and robust with respect to noise, with reconstruction errors remaining stable under moderate perturbations.
Read the full abstract
Machine Learning
Generalized Context in Cross Attention for Transfer Learning of Disjoint Tabular Data
Unlike images and text, applying transfer learning to tabular data is challenging due to heterogeneity in feature types, structures, and semantics across disparate domains. Existing methods assume shared features across data tables to enable knowledge transfer between domains, which is unrealistic in practice. \mds{This paper introduces generalized context learning to remove the requirement of shared features across domains. The generalized context captured by transformer projection weights for $key$, $value$, and $query$ provides rule-based generalization rather than the domain-specific context conventionally learned from transformer activations. Projection weights for $key$ from the source domain interact with the weight for $query$ in the target domain to achieve Cross-domain Attention Transfer Learning (CATTLE) in a data-agnostic manner. Our experiments on ten pairs of disjoint source-target data sets show that CATTLE can learn generalized context from a single source data set and is rank-wise and statistically superior to nine state-of-the-art baselines, including machine learning, deep learning, and transfer learning methods using large-scale pre-trained models. CATTLE achieves the best average rank (2.9) and delivers a 3.7% average AUROC gain over the baseline methods.} The CATTLE source code is available at https://tinyurl.com/pr5s8ywn.