Probability
Bulk universality of random regular graphs
Yukun He, Jiaoyang Huang and Xiaoyu Wang
We consider the adjacency
matrix of a uniformly random simple $d$-regular graph on $N$ vertices. For every fixed $d\geq3$ and every fixed bulk energy, we prove that the rescaled
eigenvalue point process converges to the $\mathrm{Sine}_1$ process with intensity $1/\pi$. We also establish universality of consecutive gaps at deterministic bulk labels and of fixed-energy correlation measures. The eigenvector input consists of
polynomial estimates for mixed fourth moments on a uniformly sampled
star. These estimates prevent loss of variance in the qualitative joint Gaussian limits of Backhausz–Szegedy, yielding independent variance-one Gaussian waves for fixed tuples at nearby energies. Exact switching identities then give a stationary marked flow and the microscopic loop hierarchy, which identifies the point-process limit. The appendices prove the stronger correlation and gap statements using counting-moment bounds and conditional log-gas laws for tagged limits.
Read the full abstract ▾
Computer Vision
Supermarket Product Detection and Recognition: Utilizing Deep Learning with Rectified Imagery
Mayank Sah and Jimson Mathew
Product Identification has sprung up to become one of the most challenging problems in the automation of the retail industry. With the new industry 5.0 standards, automated inventory management, and catalog creation tasks are vitally important. Object identification models have emerged as a viable answer with their unprecedented identification and localization accuracy. However, the close-knit rack design of supermarkets generates the problem of angle variation in capturing images. The angle-variant densely packed images(a single image contains many objects) become overwhelming for these models alone. In this paper, we try to supplement object detection models with traditional Hough transform (HT) and homogeneous estimation concepts. We study the effect of rectified images using homography estimation and hough transform and their limitations on the problem of grocery identification. We make a case for creating a new
dataset to test the effects of such rectification and produce analytical results on different scenarios of angle variation and object densities per image. Extensive experiments on different object detection models suggest that image rectification of angled images improves the detection accuracy of grocery products in images. The results also highlight the limitation of rectification on the angle of image capture and the object density of the image.
Read the full abstract ▾
Audio and Speech Processing
Conversation Is a Two-Body Problem: Dyadic Evaluation of Full-Duplex Dialogue Models
Sungnyun Kim, Sungwoo Cho, Jihwan Oh et al. (+1 more)
Full-duplex spoken dialogue models listen and speak at the same time, enabling voice
agents to have natural, low-latency interactions that turn-based systems cannot offer. However, they are commonly evaluated against single-sided interlocutors: pre-recorded audio that cannot react, or an automated examiner that reacts in real time but only administers a fixed sequence of tests and is never graded. These single-sided frameworks evaluate only half of a two-body problem, where turn-taking, overlap, and interruption are joint products of two coupled speakers. We propose DyaFDB, a framework that evaluates full-duplex models in a dyadic setup: two models converse directly under assigned roles with cooperative or conflicting goals, and both sides are scored offline with an external judge. DyaFDB probes how the two models behave toward each other, such as how they take turns or carry an assigned role under different interests. We instantiate four tasks as 140 scenarios and record 7,560 conversations, covering six self- and cross-play pairings. Throughout the experiments, we observe that how a model behaves continually reshapes its partner. We thus demonstrate that each model must be both the examiner and examinee of the other, and no single fixed interlocutor can play both parts. We will release the scenarios, role
prompts, and recording protocols between two full-duplex models, without any pre-recorded audio.
Read the full abstract ▾
All authors Sungnyun Kim, Sungwoo Cho, Jihwan Oh, Se-Young Yun
Solar and Stellar Astrophysics
Exploring an X1.8 flare and associated quasi-periodic pulsations using SDO and Aditya-L1 observations, alongside NFFF extrapolation
Pawan Kumar, Sanjay Kumar, Sadashiv et al. (+2 more)
The paper seeks to explore the initiation of an X1.8 class
solar flare (SOL2024-10-09T0124) and the related quasi-periodic pulsations (QPPs), which are situated within NOAA active region 13848. To achieve this, we perform a multi-wavelength analysis of the event utilizing data from the
Solar Dynamic Observatory (SDO) and Aditya-L1. Importantly, the observations indicate the emergence of a sigmoidal brightening at the onset of the
flare, alongside the
evolution of a complex morphology of
flare brightenings and ribbons around the peak time of the
flare. Furthermore, QPPs are detected during the impulsive phase of the
flare, exhibiting a dominant period of approximately 4 minutes in the hard X-ray light curve. In order to comprehend the initiation of the
flare, we construct the coronal
magnetic field at a pre-
flare stage by employing the non-force-free-field (NFFF) extrapolation model. The extrapolated
magnetic field illustrates the existence of various significant
magnetic features, including a
magnetic flux rope, a three-dimensional (3D) null, sheared arcade, and quasi- separatrix layers (QSLs) within the flaring region. The recurrent reconnection occurring at these features is proposed to play a role in the initiation of the
flare as well as the associated QPPs.
Read the full abstract ▾
All authors Pawan Kumar, Sanjay Kumar, Sadashiv, Sachin Rajkapoor Kanaujiya, Yogesh Kumar Maurya
Robotics
Beyond Waypoint Regression: Query-Based Cost Learning over Reachable Ego Futures for End-to-End Driving
Ahmed Abouelazm, Rupert Polley, Qingyuan Zhang et al. (+4 more)
End-to-end planners based on waypoint regression achieve strong open-loop accuracy, but they primarily learn to mimic expert geometry and remain difficult to adapt to deployment-time safety constraints. We propose a query-based cost-learning framework that estimates bounded costs for dynamically reachable ego trajectory queries, rather than dense BEV
cells or a small regressed trajectory set. Compact joint scene
tokens capture coherent
multimodal agent futures, while contingency-aware cost aggregation and cost-guided intra-cluster MPPI mixing convert the learned cost
topology into feasible ego plans. On nuScenes, our method improves over prior cost-estimation planners such as ST-P3 and NMP, outperforms most regression baselines in collision rate, while remaining competitive in L2, and retaining an interpretable cost interface. On real-world driving logs, the proposed planner reduces collision rates compared with SparseDrive and Alpamayo without fine-tuning, while maintaining a diverse set of candidate trajectories.
Read the full abstract ▾
All authors Ahmed Abouelazm, Rupert Polley, Qingyuan Zhang, Yin Wu, Philip Schörner, Carl Esselborn, J. Marius Zöllner
Soft Condensed Matter
Efficient Searches for Low-Energy Structures in Clusters with Thousands of Particles: Application to the Thomson Problem
Paolo Amore and David J. Wales
We propose an
algorithm for identifying low-energy minima in systems containing thousands of particles, at moderate computational cost. We focus on spherical
crystals, in which particles form ordered structures on the surface of a sphere. Our method uses symmetric seed configurations, or known structures, to initialize searches across contiguous ranges of system sizes. Newly identified minima are then used to guide searches at neighbouring sizes, allowing information to propagate efficiently through configuration space. We have tested this approach on the Thomson problem, one of the few interacting-particle systems for which global optimization has been considered for systems of this size. We substantially improve on many previous solutions, particularly at larger system sizes, where systematic global optimization is computationally expensive. Since favourable packings for the Thomson problem are reflected in systems spanning atomistic to mesoscopic length scales, the newly characterized defect patterns may inform structure prediction for a broad range of problems with spherical
topology.
Read the full abstract ▾
Classical Analysis and ODEs
Counterexamples to Hardy–Littlewood maximal inequalities on solvable groups
Koji Fujiwara
We prove that the centered Hardy–Littlewood maximal operator is not of weak type $(p,p)$ for any $1\le p<\infty$ in three solvable settings: the lamplighter group $(\mathbb{Z}/2\mathbb{Z})\wr\mathbb{Z}$ with the switch–walk–switch word metric, the Baumslag–Solitar groups $BS(1,n)$, $n\ge2$, with their digit word metrics, and the three-dimensional Sol group with any left-invariant Riemannian metric and its Riemannian volume. To the best of our knowledge, we give the first examples of finitely generated groups, equipped with word metrics and counting measure, for which the centered Hardy–Littlewood maximal operator fails to be of weak type $(1,1)$.
Read the full abstract ▾
Robotics
iGPC: Generative Motion Priors for Object-Aware Humanoid Interaction
Anujith Muraleedharan, Abdul Ahad Butt, Nolan Fey et al. (+5 more)
Humanoid robots operating in unstructured environments must combine robust whole-body control with the ability to perceive and physically interact with surrounding objects. While large-scale human motion data provides powerful priors for natural and versatile
humanoid control, effectively transferring such priors to perception-driven object interaction remains challenging. To address this bottleneck, we propose a framework that extends the recently proposed Generative Pretrained Controller (GPC) from general human motion to full-body
humanoid-environment interaction. First, we adapt GPC into interaction experts conditioned on scene affordance cues and privileged state information. This adaptation begins with interaction and affordance pre-training on human motions paired with support scenes. These experts leverage the pretrained human motion prior while learning task-specific contact behaviors, including reaching toward objects, grasping environmental supports for stabilization, and pushing movable objects. Second, we introduce a perception-driven student that retains the pretrained GPC policy and distills interaction skills from the experts using onboard sensory observations. To bridge the gap between privileged expert observations and sensory inputs, we train the student's perception not only to reconstruct expert observations, but also to preserve the expert's motion decisions and resulting joint targets. Notably, our experiments across multiple whole-body interaction tasks demonstrate that large-scale generative human motion priors provide an effective foundation for learning deployable policies for
humanoid interactions in contact-rich real-world environments.
Read the full abstract ▾
All authors Anujith Muraleedharan, Abdul Ahad Butt, Nolan Fey, Yash Prabhu, Anamika J H, Sandor Felber, Maurice Rahme, Ivan Laptev
Robotics
AutodidactWAM: Cross-Modal Self-Distillation from Generated Video to Robot Actions
Sergei Kurchev, Iaroslav Kolomiets, Miguel Altamirano Cabrera et al. (+2 more)
World-action models (WAMs) such as Cosmos 3 jointly generate future
video and
robot actions from an observation and instruction. Adapting one such model with a lightweight LoRA fine-tune to a previously unseen
robot, a Unitree G1
humanoid with five-fingered BrainCo hands, exposes a
video-action asymmetry: the
video renders plausible task executions, while the co-generated action is systematically mis-targeted. We evaluate closed-loop real-
robot trials at three cumulative stages: pre-grasp, grasp, and pick-and-place. The native action succeeds only approximately 17%, 10%, and 7% of the time, respectively, and performs worse on held-out objects. We propose AutodidactWAM, a hand-pose
estimator trained without teleoperation, followed by inverse kinematics, that runs on the model's generated
video to recover action estimates. Paired with the native prediction, these recovered actions provide preferred targets for fine-tuning only the action-related layers, while the generated
video is teacher-forced. We compare supervised relabeling with a rectified-flow adaptation of
Diffusion-DPO. After one-time embodiment adaptation, self-
distillation requires no additional task-specific teleoperation. The recovered-action gate reaches approximately 75%, 47%, and 42% pre-grasp, grasp, and pick-and-place success, compared with 17%, 10%, and 7% for the native action. A hybrid objective combining preference supervision, supervised target fitting, and Cartesian trajectory anchoring (DPO+SFT+DTW) performs best: on Oreo, the training object, it reaches 90% pre-grasp and 20% full-task success; on a held-out object, it reaches 80% and 30%. Plain Flow-DPO reaches 0% success despite 1.000 validation preference accuracy, indicating that the combination of training objectives, rather than the contrastive objective alone, drives the observed gains.
Read the full abstract ▾
All authors Sergei Kurchev, Iaroslav Kolomiets, Miguel Altamirano Cabrera, Artem Lykov, Dzmitry Tsetserukou
Machine Learning
Attenuated in-context identification in time-series foundation models: diagnosis under counterfactual inputs and repair by synthetic forced-system fine-tuning
Hong-In Won
Covariate-aware time-series foundation models (TSFMs) promise training-free what-if answers for instrumented plants: the change in output that a different future input would cause. We test this on forced engineering systems with exact counterfactuals, comparing Chronos-2, TimesFM-2.5 and TabPFN-TS with classical system identification fitted to the same context. Through their default
covariate interfaces, TimesFM-2.5 and TabPFN-TS are memoryless: the predicted effect of an input change is a same-time function of that change ($R^2 = 1.000$ for TimesFM-2.5). Chronos-2 identifies dynamics in context but attenuates them. Its predicted effect is 0.33-0.80 of the true effect, its recovered impulse response has the wrong shape, and its error on a one-degree-of-freedom oscillator levels off at 0.57 with 8192 context samples, where ARX fitted to 256 samples reaches 0.02. Context dither at inference lowers the what-if error on all six synthetic classes without training. A 26-minute fine-tune on synthetic forced systems restores the response magnitude (sensitivity 0.83-0.96) and outperforms structure-agnostic identification on Wiener-Hammerstein and a held-out friction class. A specialised in-context identifier trained on the same data comes close, so the forced-system data carry most of the gain. On three of four measured plants classical identification remains clearly better, and the fine-tuned model loses part of its univariate
forecasting skill. Paired counterfactual inputs, together with shuffled future inputs on measured records, test two properties: whether the
covariate interface can represent dynamics and whether the pretraining prior covers the plant's time scale. Only the counterfactual pairs expose the attenuation.
Read the full abstract ▾
Physics and Society
Social Physics: A manifesto
János Kertész, Marc Barthelemy, Guido Caldarelli et al. (+19 more)
Social Physics seeks quantitative, empirically testable explanations of collective human behavior. Its name has a long and contested history, but its contemporary program is neither the claim that society is literally a physical system nor an attempt to replace the social sciences with physics. It is an interdisciplinary practice: observation and experimentation, model construction, mathematical and computational analysis, and repeated confrontation with data. The field has been transformed by analyzing digital traces, network science,
agent-based modeling, large-scale experiments, and the capacity to study coupled dynamics across economic, social, technological, biological, and physical layers. We review advances in financial analysis and modeling; economic complexity; the structure and
evolution of social networks; spreading, opinion, and information dynamics; inequality, cooperation, conflict, mobility, cities, culture, science, and collective behavior; and the emerging study of human-AI ecosystems. Across these domains, common features recur - heterogeneity, interaction, feedback, adaptation, non-
equilibrium dynamics, multiscale organization, and emergent collective outcomes. We also confront the limitations that have arisen in this field: convenience samples, platform dependence, weak causal identification, over-universal claims, insufficient engagement with meaning and institutions,
privacy risks, and the danger that predictive models become instruments of manipulation. We propose a future Social Physics organized around theory - data reciprocity; causal and generative explanation; measurement validity and cross-context generalization; responsible, participatory data governance; genuinely interdisciplinary collaboration; and reflexive study of interventions and
algorithms that alter the systems being observed.
Read the full abstract ▾
All authors János Kertész, Marc Barthelemy, Guido Caldarelli, Bikas K. Chakrabarti, Tiziana Di Matteo, Serge Galam, Marta Gonzales, Yurij Holovatch, Janusz A. Holyst, Sarika Jalan, Hawoong Jeong, Neil F. Johnson, Renaud Lambiotte, Linyuan Lü, Thomas Lux, Rosario N. Mantegna, Matjaz Perc, Jari Saramäki, Michael Small, Misako Takayasu, Taha Yasseri, Wei-Xing Zhou
Disordered Systems and Neural Networks
Soft Contrastive Learning for Unsupervised Discovery of Phases of Matter
Vít Beneš, Pavel Baláž, Dmytro Bohdanov et al. (+1 more)
A machine learning (ML) framework for phase-diagram construction in condensed matter systems with unknown phase structure is presented. The method combines physically motivated descriptors with contrastive representation learning and a two-stage data-generation workflow. First, a sparse set of configurations is generated by Monte Carlo simulated annealing in order to train a contrastive neural network that produces a low-dimensional
embedding. Consequently, representative phase configurations are used as initial conditions for gradient-based optimization on a dense parameter grid, enabling efficient generation of refined
datasets for high-resolution phase-diagram construction. In order to optimize the workflow, a variant of the method, in which derived descriptors are used as soft labels in the contrastive objective, is examined as well. The framework is demonstrated on a PbZrO3-inspired model potential formulated in terms of coupled order-parameter fields and used as a physically motivated testbed for exploring configuration diversity and phase behavior. The proposed approach provides a physically informed route to ML-assisted phase-diagram construction that integrates data analysis with adaptive
dataset refinement.
Read the full abstract ▾
All authors Vít Beneš, Pavel Baláž, Dmytro Bohdanov, Jiří Hlinka
Systems and Control
Context-Conditioned Hamilton-Jacobi Reachability for Adaptive Safety Filtering
Ali Fuat Sahin, Yunus Yazoglu, John Talbot et al. (+3 more)
Hamilton-Jacobi reachability constructs safety certificates for specified dynamics and safety constraints, tying each certificate to the deployment context for which it is synthesized. We ask whether a single certificate can instead represent a family of context-dependent safety problems and be queried across deployment conditions without re-synthesis. We learn a backward reachable tube for an eight-state vehicle model conditioned on local boundary geometry, friction coefficient, and
adversarial disturbance scale. Geometry enters through an ego-frame boundary observation that defines the local containment constraint, while friction and disturbance scale enter as explicit operating-condition variables. This allows the same value function to be queried across friction coefficients from 0.4 to 2.0 and on geometries absent from synthesis. On 11 held-out evaluation geometries, the certificate maintains containment across the full tested friction range, including simultaneous geometry and grip shifts, while remaining within 1.2 percentage points in intervention rate and 0.09 m/s in speed of certificates re-synthesized with knowledge of the test geometry. We then deploy the certificate as a sampled discrete-time control barrier function filter on a full-scale vehicle near the handling limit. Lateral containment holds in every hardware session under both
adversarial driving and autonomous racing, with 99th-percentile acceleration magnitude reaching 0.99 g. Across three certificates evaluated under a fixed autonomous racing controller, lap time varies by only 3.1%, demonstrating that a context-conditioned reachability certificate can transfer to deployment geometries absent from synthesis with modest performance cost.
Read the full abstract ▾
All authors Ali Fuat Sahin, Yunus Yazoglu, John Talbot, Zeyuan Feng, James Dallas, Somil Bansal
Optimization and Control
Optimal Control of a Class of Reaction-Diffusion Systems with Pointwise State Constraints
Eduardo Casas, Konstantinos Chrysafinos and Mariano Mateos
We study optimal control problems governed by a system of semilinear reaction-
diffusion equations. The assumptions on the governing system are formulated so that our results apply to several models, including the Brusselator, Schnakenberg, and Gray–Scott systems. We prove the existence and uniqueness of a solution to the governing system without imposing any non-negativity assumption on the data. This allows us to study unconstrained, control-constrained, and state-constrained optimal control problems. For all these problems, we derive first-order necessary and second-order sufficient optimality conditions. The control acts on the activator equation through the boundary. Numerical examples demonstrate that both the inhibitor and the activator can be efficiently controlled.
Read the full abstract ▾
High Energy Physics (Theory)
Conformal Anomalies and Generalised Energy Conservation in De Sitter Spacetime
Zong-Zhe Du
We show that for bulk discrete series fields whose dual operators have integer conformal dimensions, the Generalised Energy Conservation (GEC) condition for de Sitter (dS) scattering amplitudes defined in even-dimensional extended Poincaré spacetime corresponds precisely to the absence of boundary conformal anomalies, including beta-function contributions, at the level of tree-level contact interactions. In contrast,
holographic consistency in AdS/CFT enforces only the locality of the conformal anomaly. We find novel, unique four-point interactions involving three identical conformally coupled scalars and one discrete series scalar by performing both GEC and
holographic bootstraps. This demonstrates that the solutions to GEC form a measure-zero subset of the holographically consistent solutions. Our
holographic analysis agrees precisely with the expected renormalisation conditions for conformal field theory correlators. In addition, the constraint for a particular four-point configuration may be relevant for testing
holographic consistency of scale-separated non-supersymmetric AdS$_4$
vacua.
Read the full abstract ▾
Robotics
Energy-Aware Path Following: Comparative Analysis of Reinforcement Learning and NMPC for Electric Vehicles
Mohamed Sabaa and Mostafa Emam
Path-following control strategies typically follow the bi-objective optimization dilemma: minimizing deviations from a reference path while maintaining smooth speed profiles. The latter objective is especially relevant for Electric Vehicles (EVs), since their limited driving range can be extended by recovering energy through regenerative braking, a feature that has not yet been sufficiently studied in the literature. In this work, we perform a comparative analysis of four controllers under one common Frenet frame-based kinematic vehicle model, utilizing a validated energy model (VT-CPEM) with explicit regenerative braking. Herein, we implement the following controllers: Nonlinear Model Predictive Control (NMPC), Proximal Policy Optimization (PPO), gain-scheduled Ackermann state-feedback baseline (PID-SF), and a Stanley
geometric baseline. To satisfy real-time requirements, we implement the NMPC using JIT-compiled CasADi. Moreover, we train the PPO using traditional straight and S-curve tracks, after which we successfully transfer the unmodified policy to unseen tracks, including: an ISO 3888-1 lane-change, a chicane, randomly-generated parameterized-splines, and a $\pm3^\circ$ graded road. In addition, the policy transfers to a dynamic single-track vehicle model with linear tires, zero-shot with an acceptable initial performance, which was optimized after brief fine-tuning. Thereby, we demonstrate that our PPO is readily transferable to more comprehensive vehicle models. We conclude with a performance analysis of developed controllers and discuss ideas for future work.
Read the full abstract ▾
Differential Geometry
Positivity of the canonical bundle and Hermitian metrics with quasi-negative holomorphic sectional curvature
Xueyuan Wan
We prove that if a compact Kähler
manifold admits a Hermitian metric with quasi-negative holomorphic sectional
curvature for the Chern connection, then its canonical bundle is ample. The Hermitian metric realizing this
curvature need not be Kähler. This proves the Kähler case of the quasi-negative
conjecture of Yang and Zheng.
Read the full abstract ▾
Robotics
Reactive Exploration of Unknown Environments for Redundant Robots using Virtual Model Control
Alessio Canzolino, Omar Faris, Alessandro De Blasi et al. (+1 more)
The exploration of confined, occluded, and partially known spaces poses significant challenges in robotic manipulation. The overall pose of the robotic arm must be carefully controlled to respect tight
geometric constraints while avoiding newly discovered obstacles. We address this problem by proposing an active exploration approach for redundant robotic arms with an eye-in-hand camera configuration. Our approach navigates and acquires information in real-time based on a novel scoring method that directly selects a target voxel from the unexplored space using the
robot's current state and expected information gain. To move the
robot safely toward the target voxel, we utilize Virtual Model Control, which guarantees compliance and enables whole-body reactive obstacle avoidance without the need for path replanning. Simulated and real-
robot experiments in both confined and open environments demonstrate the effectiveness of our approach, achieving over $90\%$ mapping coverage across all tested environments in under $120$ seconds without colliding with obstacles.
Read the full abstract ▾
All authors Alessio Canzolino, Omar Faris, Alessandro De Blasi, Fulvio Forni
Computer Vision
Beyond Training from Scratch: Foundation Models for Data-Efficient and Generalizable Cardiac MRI Reconstruction
Anam Hashmi, Mayug Maniparambil, Julia Dietlmeier et al. (+2 more)
Cardiac
magnetic resonance imaging reconstruction aims to recover high-quality images from undersampled acquisitions, enabling faster scans while preserving diagnostic fidelity. Recent reconstruction methods are typically trained from scratch and often require large amounts of task-specific data, limiting their robustness under data scarcity and distribution shifts. In this work, we investigate whether pretrained
vision foundation models can serve as effective priors for accelerated cardiac MRI reconstruction. We propose a reconstruction framework that integrates frozen and parameter-efficiently adapted visual encoders, including CLIP, BiomedCLIP, and DINOv2, within a
transformer-based reconstruction architecture. Extensive experiments on the CMRxRecon2023 and CMRxRecon2024
benchmarks demonstrate that pretrained representations consistently outperform a
transformer trained from scratch across multiple acceleration factors. We further evaluate performance under limited supervision and cross-
dataset transfer, showing that foundation models provide superior data efficiency and generalization. While frozen representations are particularly effective in extreme low-data regimes, Low-Rank Adaptation (LoRA) yields additional gains when moderate amounts of training data are available. Among the evaluated backbones, DINOv2 achieves the strongest overall performance. These findings highlight the potential of
vision foundation models as robust and transferable priors for cardiac MRI reconstruction.
Read the full abstract ▾
All authors Anam Hashmi, Mayug Maniparambil, Julia Dietlmeier, Kathleen M. Curran, Noel E. O'Connor
Machine Learning
Enhancing Diffusion Language Models with Autoregressive Post-Training Weights
Yiming Qin, Ke Wang, Amel Abdelraheem et al. (+2 more)
Diffusion language models (dLLMs) have emerged as a promising alternative to autoregressive (AR)
language models, offering flexible
token-update orders and parallel decoding. Recent dLLMs are often initialized from pretrained AR models before
diffusion conversion in order to inherit their learned representations. After the conversion, however, they typically ignore the extensive post-training ecosystem of their AR ancestors. In this work, we show that these existing AR post-training weight updates can instead be effectively recycled to enhance
diffusion models. Despite the changes by AR-to-
diffusion conversion, directly adding an AR post-training weight update to a
diffusion base model remains effective, bringing its performance close to that achieved by direct
diffusion post-training. Notably, AR and
diffusion post-training updates are nearly orthogonal in weight space, yet induce substantially more aligned representation changes in the
diffusion model. Their distinct updates are also complementary: composing their weights can retain gains from both regimes and further improve the post-trained
diffusion model. Based on these findings, we propose A2D, a simple training-free framework for enhancing
diffusion models with existing AR post-training resources. A2D can transfer capabilities from AR post-trained models to
diffusion base models, and further improve already post-trained
diffusion models by composing AR and
diffusion post-training updates. Across various dLLMs, including Dream, DreamReasoner, DiffuCoder, Dream-Coder, Nemotron-Labs-
Diffusion, and DiffusionGemma, A2D reliably improves instruction following, mathematical
reasoning, and coding with both supervised fine-tuning and reinforcement learning updates, without additional training, or inference-time computation.
Read the full abstract ▾
All authors Yiming Qin, Ke Wang, Amel Abdelraheem, Adam Hazimeh, Pascal Frossard
Sound
Exploiting Acoustic and Content-Oriented Speaker Verification Attacks Against Multilingual Voice Anonymization
Ridwan Arefeen, Ze Li, Rong Tong et al. (+2 more)
Attacker ASV systems for voice anonymization have been studied primarily in English, leaving their behavior in
multilingual settings largely unexplored. Conventional ASV has shown that both acoustic and contextual information are important for
multilingual speaker verification. Inspired by this, we investigate whether the same holds for attacker ASV on anonymized
speech. We evaluate both acoustic- and content-oriented attackers on
multilingual anonymized
speech and construct a
multilingual voice-converted
dataset to improve cross-lingual generalization. Our results show that attacker effectiveness depends on the linguistic utility of the anonymized
speech. Overall, acoustic-oriented attackers achieve better performance. However, when linguistic information is well preserved, the performance gap between content- and acoustic-oriented attackers narrows compared with conditions involving stronger
speech distortion. The
multilingual voice-converted
dataset further improves performance and partially reduces the cross-lingual gap. These findings highlight the need for more comprehensive attacker modeling and evaluation protocols that consider both
privacy and utility, rather than relying on a attacker strategy\footnote{Full code and pretrained models and MultiVC
Dataset link are available at: https://github.com/monkeyDarefeen/DAST
Read the full abstract ▾
All authors Ridwan Arefeen, Ze Li, Rong Tong, Ming Li, Xiaoxiao Miao
Artificial Intelligence
ChartBmkAgent: Harness-Governed Multi-Agent Construction of Chart QA Benchmarks from Sparse Error-Taxonomy Specifications
Langxi Huang, Pingping Zhang, Lanyun Zhu et al. (+4 more)
Multimodal large
language models (MLLMs) advance rapidly, while conventional
benchmark development lags behind, delaying investigation of newly observed capability gaps. Such investigation requires an expressive task format and an on-demand construction process: information-rich charts make chart question answering (Chart QA) suitable for probing coupled perception and
reasoning. Automated Chart QA construction is intended to shorten the
benchmark-development cycle by turning identified gaps into targeted samples on demand. Current methods, however, commonly separate target guidance from scratch generation: target-guided systems often require prepared data, charts, or templates, while scratch-generation systems primarily ensure artifact validity, without explicitly controlling whether newly synthesized requirements and content remain aligned with an externally specified diagnostic target. We introduce ChartBmkAgent, which turns an identified capability gap into targeted diagnostic evidence by constructing complete Chart QA samples from sparse error-taxonomy specifications. Throughout construction, a central harness governs specialized
agents, requires stage-specific evidence of
alignment with the original error category, and records the basis for each acceptance decision. On 300 taxonomy-wide samples, MLLM accuracies ranged from 32.7% to 84.3% with distinct category profiles, showing that generated samples reveal capability differences. Across three source-model comparisons, targeted follow-ups scored 50.0% versus 82.2% on matched controls ($p=8.96\times10^{-6}$); all six cross-model comparisons had the same direction, demonstrating targeted validation and diagnostic-data generation. Multiple evaluator models assessed whether each sample tested its specified error category; 86.4% met this criterion, providing empirical evidence of target preservation.
Read the full abstract ▾
All authors Langxi Huang, Pingping Zhang, Lanyun Zhu, Chunyang Jiang, Jiawei Shao, Haocheng Yuan, Peilin Chen
Robotics
Navigation with RF Cues: Embodied Perception Action under Multipath Uncertainty
Wenlihan Lu, Tianshun Li, Liuqing Yang et al. (+1 more)
Smart factory inspection requires
robots to reach connected equipment without a prior map or known target coordinates.
Radio frequency (RF) signals from the target can provide directional cues to complement visual observations when occlusion or poor lighting limits target detection. However, multipath propagation can distort these cues, making it difficult to infer the target's true direction from instantaneous RF measurements. To enable
navigation research under these conditions, we first construct a Habitat Sionna RT
benchmark that uses detailed scene geometry and assigned material properties to generate aligned visual and RF observations in response to
robot actions. Building on this
benchmark, we propose an uncertainty aware
multimodal navigation framework that jointly estimates target direction and its uncertainty from a history of RF, visual, and pose observations. These estimates inform action selection alongside visual context. Experiments in unseen scenes show relative improvements of 18.2% in success rate (SR) and 11.5% in success weighted by path length (SPL) over the strongest evaluated baseline.
Read the full abstract ▾
All authors Wenlihan Lu, Tianshun Li, Liuqing Yang, Shijian Gao
Differential Geometry
Biconservative and Biharmonic Hypersurfaces in Product Spaces
V. Branding, S. Montaldo, C. Oniciuc et al. (+1 more)
In this paper we investigate biconservative and biharmonic hypersurfaces in product spaces $L^m(\varepsilon)\times {\mathbb R}$, where $L^m(\varepsilon)$ denotes the space form ${\mathbb S}^m$ or ${\mathbb H}^m$ of constant sectional
curvature $\varepsilon=1$ or $\varepsilon=-1$ respectively. We obtain their complete description in the case of isoparametric hypersurfaces and hypersurfaces of constant angle. Moreover, we prove some classification results for biconservative and biharmonic hypersurfaces for which the gradient of the mean
curvature function is a principal direction. In this context we obtain new examples of complete, non-cylindrical, non-totally umbilical biconservative hypersurfaces in ${\mathbb H}^m\times {\mathbb R}$. Then we prove a
rigidity theorem related to these examples. Finally, we prove a global non-existence
theorem for complete proper biharmonic hypersurfaces in ${\mathbb H}^m \times {\mathbb R}$ such that the square of the mean
curvature function is bounded and the
Ricci curvature is bounded from below.
Read the full abstract ▾
All authors V. Branding, S. Montaldo, C. Oniciuc, A. Ratto
Probability
The mixing time of Mean-Field Gibbs-Ornstein-Uhlenbeck processes
Johan Jonasson
We introduce mean-field Gibbs–Ornstein–Uhlenbeck (MFGOU) processes, a class of discrete-time Markov processes with increments of order $1/n$, whose conditional means approximate a smooth vector field. If the field has a unique attracting zero and satisfies a global Lyapunov condition, we prove \[ \tau_{\mathrm{mix}} = \frac{1}{\lambda_1}n\log n+O(n), \] where $\lambda_1$ is the smallest
eigenvalue of the drift Jacobian at the attractor. Mean-field block measures form an important subclass encompassing mean-field
spin systems and projected Gibbs samplers for Latent Dirichlet Allocation (LDA) and related mixture models. The leading mixing constant has a spectral expression involving the free-energy Hessian and the covariance of single-step dynamics. A block-refinement argument transfers projected-chain results to the full chain under a mild spectral condition, preserving this constant. The theory recovers known cutoff locations for Curie–Weiss Ising and Potts models and complete multipartite Ising models, gives an external-field formula for Curie–Weiss Ising, and covers regular-graph interactions. It yields explicit $n\log n$ asymptotics for multinomial mixtures and partially collapsed LDA. For full LDA, mixing is determined conditionally on the stationary-point structure of the free energy, distinguishing metastability from topic-label switching. The framework also applies to continuous-state Gibbs sampling, constant-step
stochastic gradient descent, and frequently refreshed
Hamiltonian Monte Carlo, rigorously for truncated Gaussian increments and heuristically without truncation. The proof combines variance and mean estimates with path coupling allowing slight expansion in expected distance. After rescaling, the coupling distance is a nonnegative supermartingale whose quadratic variation gives a positive probability of coalescence.
Read the full abstract ▾
Artificial Intelligence
DSV-Mem: Evaluating Multimodal Memory in Professional Workflows for MLLM Agents
Jike Zhong, Ritwick Chaudhry, Xuanbai Chen et al. (+4 more)
Conversational MLLM
agents are increasingly expected to assist in professional workflows, from AI research and engineering design to product management and business operations. Yet this capability remains underexplored: existing
benchmarks largely focus on informal, everyday interactions and personal-life scenarios featuring photographic natural images, isolated static artifacts, and recall-oriented questions. In contrast, professional scenarios often involve structured, information-heavy artifacts that undergo frequent revisions and authority updates, and compositional queries requiring reconciliation of many artifact versions while tracking state precisely. To address these challenges, we introduce DSV-Mem, a
benchmark for evaluating Dense Stateful Visual
Memory. DSV-Mem comprises expert-reviewed scenarios and 1,000 questions across five user-oriented categories (Current State, Past State, Derived State, Change History, and Conflict/Refusal). A Hartley-inspired criterion favors questions with broader visual-evidence inspection demands. We also introduce a generation harness that produces evaluation suites by decoupling state-transition synthesis from conversation filling. Evaluation over 27 configurations spanning frontier and open-weight models and
memory management methods reveals that the strongest baseline scores below 45% on DSV-Mem. Analysis surfaces findings: 1) multimodality and information density both contribute to difficulty, but state
evolution, particularly the number of governing updates, is the dominant tested factor. Raw conversation/haystack length, OCR, and arithmetic are not the primary bottlenecks; 2) models often fail to verify user premises against prior state updates before answering; 3) increased
reasoning effort and
memory management methods yield limited gains, whereas state-aware designs prove more effective. The
benchmark and code will be publicly released.
Read the full abstract ▾
All authors Jike Zhong, Ritwick Chaudhry, Xuanbai Chen, Tianchen Zhao, Linghan Xu, Yifan Xing, Nishant Sankaran
Artificial Intelligence
Beyond Corrected Memory: Execution Consistency in Multi-Agent Systems
Zhe Yu, Zixuan Wang, Peidong Wang et al. (+3 more)
Shared
memory coordinates
agents' actions, but correct records do not establish that those actions satisfy task requirements.
Memory governance and failure diagnosis regulate or inspect recorded information; they do not by themselves establish whether it is sufficient to judge task duties. We define execution consistency through duties governing state use, information handoffs, and final-state agreement, with explicit evidence conditions for judging fulfillment. Our core claim is that identical retained records can correspond to compliant and violating executions under the same task rule. Controlled removal of evidence such as receipt, action dependence, or response validity leaves 82.4% of opposite-label pairs indistinguishable; restoration separates 97.9% of the merged pairs. Natural-log annotations identify the defined violations in actual executions. However, existing logs do not always explicitly represent the execution relationships needed for these judgments. To assess the definition's practical value, we use CAVERT, a framework for consistency diagnosis and recovery, to extract supported relationships from logs and apply these criteria. It consistently outperforms contract-prompted LLM and rule-based baselines in diagnosis across all 12
benchmark-executor settings. Under the same gate and executor limits, it also outperforms rule-guided recovery in all four evaluated environments. These findings identify execution evidence that
agent-
memory and execution interfaces should preserve for reliable judgment.
Read the full abstract ▾
All authors Zhe Yu, Zixuan Wang, Peidong Wang, Hehai Lin, Ruochen Zhao, Chengwei Qin
Statistical Methodology
The Geometry of Existence and Uniqueness of Maximum Likelihood Estimation in Categorical Response Models
Lukas Sablica, Kurt Hornik and Thomas Rusch
Nonexistence of the maximum likelihood estimate (MLE) under separation is treated as a solved problem for binary logistic regression and as a scattered collection of model-specific results everywhere else. We show that it is one phenomenon with one criterion. In a latent polyhedral categorical response model, every observed outcome corresponds to a polyhedral event in latent variables whose faces shift linearly with the parameter. Random-utility choice, cumulative-link, ranking, multivariate binary and ordinal, sequential, adjacent-category logit, and fixed-score stereotype models belong to this family. The likelihood sees each observed factor only through the columns of its threshold map, the structure vectors. A finite MLE exists if and only if the pooled structure vector set has overlap. Sufficiency requires only continuity, necessity requires only strictly threshold-increasing probabilities, and neither likelihood concavity, exchangeability, nor full design rank is needed. Positive strictly log-concave latent densities and threshold-identifiable polyhedra then yield uniqueness on the estimable span. A single linear program determines whether overlap holds, and convex cone geometry measures the dimension of separation. An application using cumulative-link models for willingness to share health data identifies observations and model terms associated with nonexistence and illustrates how the diagnostics inform model revision and sensitivity analysis.
Read the full abstract ▾
General Relativity and Quantum Cosmology
Regular black holes do not violate the first law of thermodynamics
Bobir Toshmatov, Bobomurat Ahmedov, Nozima Isamadinova et al. (+1 more)
Regular black holes in general relativity coupled to nonlinear electrodynamics have been reported to exhibit an apparent mismatch in their thermodynamics: the temperature obtained by differentiating the mass along the family of regular metrics differs from the Hawking temperature by a factor $\Xi$, and the first law seems to need a modified
entropy or a corrected internal energy. We examine this question for an electrically charged family of regular black holes in $D\geq4$ dimensions whose metrics reduce to the Bardeen and Hayward geometries in four dimensions. The source is written as a
Hamiltonian density $H(P)$ with fixed couplings and no solution parameters. This theory has a two-parameter family of static solutions, of which the regular metrics form a one-parameter subfamily. We prove analytically that on the whole solution space the standard first law of thermodynamics holds with the Hawking temperature, the area
entropy and the horizon potential, in agreement with the general first law for nonlinear electrodynamics. The factor $\Xi$ appears when the mass is differentiated along the regular metric family at fixed length parameter, a variation that moves through the space of theories and also changes the conserved charge. We show that $\Xi^{-1}$ is the Jacobian of this projection, equal to the fraction of the mass inside the horizon, give its general form for an arbitrary regular mass function, and show that $\Xi^{-1}dM_{ADM}$ is not a closed one-form on the two-parameter metric family, so this correction cannot be the differential of a local state function there. We also point out that the Legendre map between the $F$ and $P$ frames degenerates at a radius that lies outside the horizon of near-extremal solutions in $D\geq5$.
Read the full abstract ▾
All authors Bobir Toshmatov, Bobomurat Ahmedov, Nozima Isamadinova, Chengxun Yuan
Cryptography and Security
Surviving the Router: Optimizing Skill Injections for Retrieval and Execution
Haneen Najjar, Luca Scionis, Haritz Puerto et al. (+1 more)
AI
agents increasingly rely on modular third-party "skills" that are dynamically selected by skill routers to execute complex tasks. While recent studies highlight the threat of
prompt injections embedded in these skills, existing evaluations often assume settings where the malicious skill is already selected for execution. We show that this assumption can substantially overestimate attack success. In realistic multi-skill environments, injected skills must first compete for
retrieval, reducing the effective attack success rate (ASR) of existing injections by 87-97%. To address this limitation, we introduce CORSA (Cluster Optimization for Router-Aware Skill Attacks), a router-aware attack that optimizes skill injections for both
retrieval and execution across clusters of related tasks. We evaluate skill injection attacks under router-managed multi-skill settings by extending the
benchmark introduced by SkillRouter with eight malicious payload categories. CORSA uses successive optimization stages to first improve
retrieval and then optimize end-to-end attack success, while we evaluate user utility and injection naturalism separately. Our experiments show that CORSA substantially improves both
retrieval and end-to-end attack success over existing skill injections while preserving user utility, and that the resulting attacks transfer across different router architectures and LLM backbones.
Read the full abstract ▾
All authors Haneen Najjar, Luca Scionis, Haritz Puerto, Sahar Abdelnabi