Quantum Physics
Efficiently Optimizing the Quantum Value of Bell Inequalities using Batched Gradient Descent
Xinyu Xu, Ping Zhu, Weikang Li et al. (+3 more)
Exploring the world of Bell inequalities requires numerical methods that scale effectively with the number of parties, inputs, and outputs. Such methods can be used to study large-scale Bell inequalities beyond the reach of analytic methods, to design experiments for violating more complicated Bell inequalities, and to enable applications of Bell inequality violation. Such applications include randomness generation and multi-
agent coordination with restricted communication. However, current general purpose optimizers such as the see-saw method struggle to handle Bell inequalities with only a dozen inputs and outputs. We introduce an optimizer for the
quantum value of a Bell inequality based on batched gradient descent (BGD). More precisely, our BGD optimizer is a differentiable search over feasible
quantum strategies in which states and projective measurements are generated from unconstrained variables. The Bell expression is evaluated by a direct tensor contraction without forming the prohibitively large Bell operator. This formulation makes parallel random restarts inexpensive and is thus naturally implemented by a GPU. We evaluate our optimizer against the see-saw method and find a significant speedup for a wide variety of Bell inequality families. Implemented on a GPU, our BGD optimizer can optimize Bell inequalities in our numerical experiments with more than a thousand inputs or outputs within minutes. For these large-scale Bell inequalities, we also efficiently compute the classical value by enumerating possible strategies using a GPU or expressing the optimization as a mixed-integer linear program. Our optimizers can serve as numerical evaluation tools for Bell experiments and multi-
agent coordination problems modeled by Bell inequalities with many inputs and outputs, a common feature of real-world scenarios such as high frequency trading and distributed systems.
Read the full abstract ▾
All authors Xinyu Xu, Ping Zhu, Weikang Li, Pierre Pocreau, Dalu Ding, Dawei Ding
Robotics
ActiveWAM: Evidence-Aware Active Vision for World-Action Models
Renjun Wu, Luzhou Ge and Xuesong Li
Active
vision manipulation requires a policy to control both its camera and its end-effectors, yet camera motion determines which evidence remains visible within finite observation windows. Acquiring a new view can displace task-critical cues, while retaining a view forgoes potentially useful observations. We formulate this as an evidence-aware retain–acquire problem and present ActiveWAM, a unified world–action model that learns observation and manipulation jointly. To this end, we propose training-time inversion which constrains a frozen
video prior by task-bearing source evidence and visible temporal changes, eliminating the need for test-time inversion or candidate ranking. At deployment, the policy generates bimanual and pan/tilt actions-including stay and reacquisition behaviors-from view-aware history, and updates context from newly measured RGB observations. Future-
video prediction serves as a co-training signal, while action generation requires neither future-
video decoding nor optimal viewpoint annotations. We introduce RoboTwin-AV, a 50-task
benchmark with executable pan/tilt control and automatically generated demonstrations. ActiveWAM improves TAVIS out-of-distribution success by up to 17.0 percentage points over the strongest baselines, achieves 20.0 additional points over Fast-WAM on RoboTwin-AV, and outperforms it by 26.7 points on real-world physical kitchen tasks.
Read the full abstract ▾
Machine Learning
Learning PDE Dynamics between Submanifolds Using Green's Observation Operators
Jan Tauberschmidt, Jephte Abijuru, Naukshatro Bose et al. (+5 more)
Many physical systems are driven and observed only on lower-dimensional submanifolds of a larger spatial domain, while their dynamics are governed by the ambient medium occupying that domain. Examples include
laser-heated parts imaged by an infrared camera, and ground-level emissions measured on a sensor plane. Full-domain solvers, however, compute the entire volume for every new source although only the observation submanifold is needed, and black-box surrogates do not exploit that the ambient medium remains fixed. We introduce the Green's Observation Operator (GObO), which maps the ambient medium once to the Green's kernel of a linear PDE restricted to the source and observation submanifolds. New sources then cost one lower-dimensional integral and no network evaluation. Exponential rates in the kernel yield an exact finite streaming state with horizon-independent
memory; we prove its stability and an approximation rate for the restricted heat kernel. On three-dimensional heat conduction and advection–
diffusion with collocated and distinct source and observation geometries, GObO trained on static sources predicts responses to moving sources zero-shot with 4–8$\times$ lower error than black-box surrogates, at 1.4\,ms per query after a single conditioning pass. The same kernel transfers across resolutions and admits corrections for mild nonlinearities, including radiative losses and temperature-dependent conductivity, without retraining, at the cost of lower in-distribution accuracy.
Read the full abstract ▾
All authors Jan Tauberschmidt, Jephte Abijuru, Naukshatro Bose, Samuel Okon, Sophie Fellenz, Marius Kloft, Jonas Latz, Sebastian Josef Vollmer
Computation and Language
Acmite: Mitigating Gender Bias in LLMs through Concept-Guided Mutual Information
Tian Lan, Xiaoqing Cheng, Han Zhang et al. (+1 more)
Large
language models (LLMs) can reproduce social stereotypes from their training data, motivating extensive research on model debiasing. However, existing methods often rely on explicit biased examples or predefined group-term substitutions, making them sensitive to wording and less effective at capturing stereotype concepts shared across diverse contexts. More importantly, they typically suppress biased outputs without explicitly modeling the statistical dependence between model outputs and the underlying stereotype concepts. We propose Acmite, a lightweight concept-guided framework for targeted and selective debiasing. Acmite represents stereotypes as structured semantic concepts and uses maximal marginal relevance (MMR) to select diverse concepts for debiasing. Inspired by mutual information minimization, it approximates this dependence with
token-level KL divergence while preserving task semantics. A lightweight LoRA adapter is trained with the base model frozen and activated at inference time only when the input is sufficiently similar to stereotype-related concepts; otherwise, the original model is used directly. We evaluate Acmite on BBQ, CrowS-Pairs, and StereoSet, and assess general capability preservation on ARC-Challenge, GSM8K, and PIQA. Experiments across three LLMs show that Acmite effectively mitigates gender bias across complementary evaluation formats while maintaining competitive performance on bias-unrelated tasks. Anonymous code and data are available at https://anonymous.4open.science/r/Acmite-18E2/.
Read the full abstract ▾
All authors Tian Lan, Xiaoqing Cheng, Han Zhang, Jiang Li
Audio and Speech Processing
Teaching LLMs to Hear Who Spoke What: Metadata-Supervised Pretraining for Encoder-Free Speech-LLMs
Mohan Shi, Ruchao Fan, Sunit Sivasankaran et al. (+2 more)
Encoder-based
speech large
language models (
Speech-LLMs) commonly employ pretrained
speech encoders that prioritize linguistic content but may discard fine-grained acoustic cues essential for speaker discrimination and paralinguistic understanding. Encoder-free
Speech-LLMs instead map Mel-spectrogram features directly into the LLM input space through lightweight
embedding layers, enabling the LLM to learn from low-level acoustic features. However, systematic pretraining strategies for encoder-free
Speech-LLMs remain underexplored, limiting their ability to compensate for the absence of large-scale pretrained
speech encoders. We propose metadata-supervised pretraining (MSP), which leverages
speech attributes such as speaker identity and
emotion to develop speaker-discriminative and paralinguistic capabilities. We further introduce speaker-aware utterance composition (SAUC) to strengthen speaker discrimination and apply random span masking to regularize pretraining. We primarily evaluate our approach on joint ASR and speaker diarization in multi-speaker conversations, complemented by experiments on paralinguistic
speech-understanding tasks. Under matched training-data conditions, our encoder-free model outperforms its randomly initialized encoder-based counterpart. With limited metadata-annotated data, it is competitive with models using
speech encoders pretrained on substantially larger corpora, outperforming them in several settings. These results demonstrate the potential of encoder-free architectures for building native
multimodal LLMs that acquire diverse
speech capabilities.
Read the full abstract ▾
All authors Mohan Shi, Ruchao Fan, Sunit Sivasankaran, Keqi Deng, Jinyu Li
Representation Theory
Harmonic Norms on Quiver Representations over Non-Archimedean Fields
Oren Ben-Bassat
Let $K$ be any complete non-Archimedean field, not necessarily spherically complete. We prove that every polystable representation of a finite quiver over $K$ admits a split harmonic norm in the sense of Haiden–Katzarkov–Kontsevich–Pandit (HKKP), answering a question from their preprint Towards Categorical Kähler Geometry. Moret-Bailly's closed-image
theorem gives a uniform separation from representations with destabilizing subobjects. A finite estimate on norm weights then yields a linear coercivity bound for stable representations and hence a minimum of the extended HKKP potential on the complete CAT(0) space of all norms. A direct first-variation computation identifies split minimizers with HKKP harmonic norms. Strict supporting hinges reduce split attainment to semistable reduction for quivers with arbitrary real radius labels. Small changes of the lifted splitting make each reversed extension nonsplit; a finite slope–rank descent proves termination. Consequently, every minimum in the completed norm space is attained by a split norm with the same arrow singular profiles, without restrictions on the value group or the quiver. Nonsplit minimizers illustrate split attainment in one and two radius modifications, and a two-parameter Berkovich analytic family separates polystability, semistability, and harmonicity of a chosen split norm.
Read the full abstract ▾
Quantum Physics
Predicting electric-field noise in ion traps using fluctuation electrodynamics
Markus Teller, Da An, Alberto M. Alonso et al. (+4 more)
Electric-field noise in microfabricated surface ion traps contributes to gate errors in trapped-ion
quantum computers, but this noise has been challenging to predict. Here, we present a method to predict the electric-field noise arising from fluctuations in the bulk of dielectric and metallic materials. The method is valid for arbitrary trap geometries, and the only relevant material property is energy loss. We apply it to an ion trap with an electrically floating electrode and find the contribution from dielectrics to be dominant. Our model predicts the observed noise for this trap within a factor of four. In addition, we apply our model to typical surface trap designs and analyze how the electrode structures shield the noise.
Read the full abstract ▾
All authors Markus Teller, Da An, Alberto M. Alonso, Philip C. Holz, Philipp Schindler, Hartmut Häffner, Tracy E. Northup
Machine Learning
Artifact Annotations Partially Substitute for Per-User Calibration: SAFE-EDA and a Normalization-Controlled Evaluation of Wrist-EDA Affect Recognition
Haochen Chai, Xinbi Luo, Zining Liu et al. (+1 more)
Wrist electrodermal activity (EDA) differs in amplitude from one person to the next, so affect-recognition models normalize their input before classification. Studies that test such models on held-out subjects seldom report where the normalization statistics come from, yet statistics computed from the held-out subject's own recording give the model information that a device does not have when it is first worn. We asked how this choice alters the measured benefit of pretraining. A compact convolutional network, SAFE-EDA, was pretrained on expert artifact annotations from 43 subjects and compared with the same network trained from scratch on the Wearable Stress and Affect Detection (WESAD)
dataset (15 subjects, leave-one-subject-out), with two normalization sources crossed with four window hops. When the statistics came only from training subjects, pretraining raised macro-F1 by 0.078 to 0.227; when they came from the held-out user's full recording, the gain fell to between 0.020 and 0.050 and was no longer significant. Artifact supervision was far more useful than self-supervised pretraining on the same recordings (0.078 versus 0.008). Across 13 configurations in two
datasets, the pretrained network was better in 12, but on the second
dataset (26 subjects) per-user normalization increased the gain instead of reducing it, so the interaction depends on the data. Only five of 50 published WESAD studies state which data were used for normalization. Reporting this choice is necessary to separate first-use performance from performance after calibration.
Read the full abstract ▾
All authors Haochen Chai, Xinbi Luo, Zining Liu, Fangfang Jiang
Group Theory
Rotational Trajectory Synopsis Groups
Gregory S. Chirikjian and Taeyoung Lee
We construct finite-dimensional Lie-group summaries of based rotational paths that retain the exact terminal attitude and selected iterated angular-velocity integrals. These summaries, called rotational trajectory synopses, compose exactly under path concatenation. Building on the classical works of Magnus and Chen, we establish endpoint-retaining body, spatial, and hybrid synopsis groups at every finite order and give a uniform
matrix realization for the evaluated Magnus hierarchy. We distinguish this hierarchy from the full Chen signature from order three onward and prove \chg{that the exact endpoint is independent of every finite truncation of the body, spatial, or combined angular-velocity signature: the synopsis maps are surjective at every finite order}. An underactuated satellite example connects the retained coordinates to
geometric phase in spacecraft attitude dynamics.
Read the full abstract ▾
Analysis of PDEs
Nonlinear stability of the Guderley imploding shock wave in radial symmetry
Giorgio Cialdea
We prove nonlinear stability of the Guderley imploding shock under radial perturbations for the three-dimensional compressible Euler equations in the adiabatic range $1<\gamma\leq\frac{5}{3}$. The perturbed solutions develop an asymptotically self-similar implosion in finite time. Our result allows perturbations with small, nonzero constant pressure in the quiescent region ahead of the shock. The proof combines weighted $L^\infty$ estimates along characteristics with a computer-assisted proof of stability inequalities for the Guderley profile.
Read the full abstract ▾
Analysis of PDEs
Local separation profiles for the stationary Prandtl system
Ky Giao Duong, Tej-Eddine Ghoul, Sameer Iyer et al. (+1 more)
We construct solutions to the two-dimensional stationary Prandtl system under a constant adverse pressure gradient exhibiting a family of distinct separation laws. More precisely, for every integer $\ell\ge1$, we construct smooth initial data for which separation occurs at a finite location $x=x^\ast$, with \[ \partial_y u(x,0) =\mathbf C_\ell(x^\ast-x)^{\ell/2}\bigl(1+o(1)\bigr), \qquad x\uparrow x^\ast, \] where $\mathbf C_\ell>0$ depends on the initial data. We determine the local asymptotic structure of the solution near separation through matched inner and outer expansions based on linearization around the ground state in von Mises variables. This description addresses Open Problem~5 posed by Oleinik and Samokhin~\cite{OleinikSamok-book-99}. We also prove that the square-root separation regime, corresponding to $\ell=1$, is dynamically stable under small admissible perturbations of the initial data in a suitable
topology.
Read the full abstract ▾
All authors Ky Giao Duong, Tej-Eddine Ghoul, Sameer Iyer, Yasunori Maekawa
Computation and Language
Compound interpretation is based on analogy
Tian Shen and Harald Baayen
How compound meanings are best predicted from constituent meanings remains a central question in computational models of lexical semantics. Comparing different computational models provides a way to evaluate alternative accounts of how semantic information is combined during compound comprehension. We propose a new model, the Compound Analogy Model (CAM), that predicts a compound's
embedding by adding its constituent
embeddings together with the average shift vectors of the two constituents' compound families. The resulting model is parameter-free and exploits local analogical structure in the semantic space. We evaluated CAM against the CAOSS model on Mandarin Chinese compounds. CAM consistently achieved higher prediction accuracy than CAOSS on both training and held-out data, with the exception of three-character compounds, for which analogical generalization is constrained by both small constituent families and a pronounced imbalance in family size between the two constituents. The advantage of CAM remained when evaluation was based on frequency-defined train-test splits that better approximate generalization from familiar to novel compounds. To assess the cognitive plausibility of the two models, we further examined whether model-derived semantic measures predict visual lexical decision latencies for two-character compounds. Predictors derived from CAM provided improved prediction for response latencies compared to predictors derived from the CAOSS model. These findings indicate that compound meaning is better characterized as local analogical generalization than as the application of a learned global linear transformation, and demonstrate that analogical semantic structure provides a cognitively plausible basis for compound comprehension.
Read the full abstract ▾
Computer Vision
Architectural Sampling: Test-Time Scaling via Computational Diversity in Frozen Vision-Language Models
Akshit Singh, Shyam Marjit, Wei Lin et al. (+2 more)
Test-time scaling often seeks better answers by sampling multiple responses from a frozen model, yet conventional temperature sampling generates every candidate along the same fixed computation path. We introduce architectural sampling, a training-free method that generates candidates through distinct forward computations by reusing selected blocks of decoder layers. Varying the block location and repetition count introduces computational diversity without updating model weights or adding auxiliary parameters. Across five Qwen checkpoints and twelve
multimodal benchmarks, architectural sampling improves pass@9 over standard-path temperature sampling by 6.58 percentage points on average at the same nine-candidate budget. Reusing early layers yields the strongest gains, and the improvement in candidate coverage persists even under greedy decoding. The resulting candidates show lower lexical overlap and improve accuracy when used as rollouts for label-free test-time reinforcement learning. These findings extend the benefits of our architectural sampling beyond candidate coverage, demonstrating more effective learning from a model's own outputs.
Read the full abstract ▾
All authors Akshit Singh, Shyam Marjit, Wei Lin, Leonid Karlinsky, M. Jehanzeb Mirza
Medical Physics
AnatomIQ: An Open-Source Toolkit for Automated Background Detection in Medical Imaging
Rafael Carballeira, Hayley A. Cash and Marthony L. Robins
Manual background selection for contrast-to-noise ratio (CNR) calculations in CT image quality assessment is time-consuming, operator-dependent, and compromises reproducibility. Advanced metrics such as Noise Power Spectrum (NPS) and Task Transfer Function (TTF) traditionally require dedicated phantom acquisitions, adding cost, radiation exposure, and workflow disruption. We developed AnatomIQ, an open-source Python toolkit that automates background detection and performs phantom-free NPS and TTF analysis directly on patient anatomy. Automated background detection uses tissue-specific Hounsfield unit thresholding and morphological filtering, paired with a web interface for click-to-select lesion identification from PACS screenshots. Reported metrics include CNR, signal-to-noise ratio, detectability indices, and dose-normalized figures of merit with automated optimization recommendations. NPS quantifies noise texture from automatically detected uniform tissue regions, while TTF leverages natural circular anatomical structures (e.g., trachea, vessels) as edge phantoms, together characterizing the resolution-noise trade-off without dedicated phantom scans. Analysis takes 2-3 minutes per case, with CNR computed in under 10 seconds and NPS/TTF available on demand. Automated detection reliably identified valid regions across anatomical sites, including head-and-neck scans, a near-worst-case test of uniform-region availability. Both NPS and TTF pipelines were validated against phantom-derived references, confirming that in vivo measurements reflect genuine anatomical texture and resolution rather than pipeline artifacts. By eliminating operator variability and phantom requirements, AnatomIQ enables practical continuous quality monitoring and evidence-based reconstruction optimization. The toolkit is open-source, with optional commercial licensing for academic and clinical use.
Read the full abstract ▾
Machine Learning
MiLoop: Selective Memory Propagation for Neural Combinatorial Optimization
Changliang Zhou, Yuanyao Chen, Rongsheng Chen et al. (+2 more)
Constructive neural combinatorial optimization (NCO) has emerged as a promising paradigm that learns to construct solutions to combinatorial optimization problems (COPs) step by step, which reduces reliance on handcrafted rules and enables fast inference. While many methods with dynamic
embeddings generalize well, they typically rebuild subproblem representations from scratch at each step using deep
attention stacks. Many high-performing methods in this category rely on solution labels or pseudo-labels for efficient training, or on aggressive search space
pruning during reinforcement learning (RL). To address these limitations, we propose
Memory-in-the-Loop (MiLoop), a purely RL-based constructive framework that leverages the multi-step computation already required by a rollout for selective
memory propagation. Each rollout provides solution-quality feedback for learning while propagating historical representations, thereby enabling a shallow policy to learn effective dynamic
embeddings without external solution labels or training-time search-space
pruning. Specifically, MiLoop fuses current
embeddings with historical
memory before the
attention layers and applies adaptive gated updates afterward. The updated representations support both current decisions and stepwise reuse. Extensive experiments across four COPs demonstrate that MiLoop consistently produces high-quality solutions on instances ranging from 100 to 10 million nodes, highlighting its strong generalization ability.
Read the full abstract ▾
All authors Changliang Zhou, Yuanyao Chen, Rongsheng Chen, Zhiyun Lin, Zhenkun Wang
Quantum Physics
Mentor-Initiated Asymmetric Bidirectional Quantum Teleportation Protocol for Arbitrary Qubit States
Sugata Adhya and Benchao Yang
In this paper, a mentor-initiated bidirectional asymmetric
quantum teleportation protocol is proposed. Unlike existing schemes that restrict the two-
qubit state transmitted from Bob to Alice to a Bell-like form $\alpha|00\rangle+\beta|11\rangle$, our protocol allows an arbitrary two-
qubit state to be teleported from Bob to Alice within the mentor-initiated framework. The required entangled resources are constructed using standard
quantum gates, and the protocol is illustrated through its
quantum-
circuit implementation in Qiskit. The proposed protocol is further simulated using Qiskit, and the simulation results show probability distributions very close to the theoretically expected values for the chosen input states.
Read the full abstract ▾
Quantum Physics
A Code-Agnostic Graph Neural Network Decoder from the Detection Error Model
Federico Alberto Astolfi and Guido Pupillo
We present POLYMECHANON, a graph neural network (GNN) decoder for
quantum error correction whose only input is the detection error model (DEM) of a
quantum code under a given noise model. We represent the DEM as a tripartite graph of detectors, error mechanisms and logical observables, where every input feature is computed by using the
quantum code as data rather than design choice. In this way, the same architecture decodes in principle any stabiliser code, under any noise model that can be expressed as a detection error model, once trained on the DEM. We test this approach along four directions. Firstly, on the rotated surface code the decoder outperforms correlated MWPM under both phenomenological and
circuit-level noise, with up to $25\%$ fewer logical failures. Secondly, on a family of high-rate qLDPC codes it matches BP+OSD on the smaller codes and surpasses it on the larger ones, with up to $16\%$ fewer logical failures on $[\![130,4,6]\!]$. Thirdly, its decoding time does not depend on the physical error rate, whereas that of BP+OSD grows with it: on $[\![130,4,6]\!]$ its effective cost per shot is of the order of a few ms on a single GPU against a few tens of milliseconds for BP+OSD on a single CPU core, and its fixed computational graph makes it a candidate for real-time decoding on neutral-atom processors. Finally, a single model trained across codes of different families outperforms uncorrelated MWPM on the graph-like codes seen during training and stays within $10\%$ of BP+OSD on the others, while it tends to fail to generalize to unseen codes, particularly larger ones. Its probabilistic output enables confidence-based post-selection, lowering the logical error rate by more than an order of magnitude on the surface code while keeping more than $90\%$ of the shots. Since only the DEM changes, new codes and noise models can be decoded without redesigning the decoder.
Read the full abstract ▾
Robotics
Beyond Leaderboard Scores: A Deployment-Focused Protocol for Interpretable Tracking Evaluation in Pedestrian-Centric Environments
Dominik Wojcikiewicz and Diego Paez-Granados
Mobile
robots operating among pedestrians need trajectories that become available quickly, remain spatially credible through missed observations, preserve identity, and fit within an embedded computing budget. Aggregate tracking scores provide limited insight into when and how trajectories fail, while varying detector inputs can confound tracker and detector quality. We present a deployment-focused, tracker-only evaluation protocol that uses shared detections to isolate tracker behavior and directly evaluates initialization, detector-gap continuation, identity recovery, close-neighbor association, and load-dependent tracker-step runtime, while Higher Order Tracking Accuracy (HOTA) is retained as a complementary aggregate measure. We apply the protocol to the JackRabbot
Dataset and
Benchmark (JRDB) using six open-source trackers and our lightweight Pedestrian Reference Tracker (PedRefTrack), together with a GT-assisted variant that estimates the remaining tracker-side gap under idealized association and motion. Under fixed detections, the non-GT trackers span only 24.26%-29.67% HOTA yet exhibit markedly different capability profiles. After 1.0 s without detector support, no tracker without GT assistance maintains spatially correct, same-identity output in more than half of eligible cases, making missing-observation continuation the dominant limitation among the tested properties. Close-neighbor failures are smaller and increase mainly at the shortest separations. Tracker-step runtime on an NVIDIA Jetson Orin is heavy-tailed and load-sensitive, causing several trackers to fall below the 10 Hz real-time target in crowded frames. The protocol provides a reproducible way to characterize tracker behavior and deployment suitability in pedestrian-centric environments. Code and evaluation scripts are released at https://github.com/SCAI-Lab/tracker_eval.
Read the full abstract ▾
Computer Vision
When Text-to-Image Helps Editing: The Effects of Conditioning During Denoising
Lidia Troeshestova, Alexander Ustyuzhanin and Sergey Kastryulin
Unified models are trained for both instruction-based image editing and text-to-image (T2I) generation, but standard editing pipelines keep source-image conditioning throughout denoising. We ask whether editing can benefit from T2I, and study how the effects of conditioning vary across edits and denoising stages. In pure editing, source
attention declines for some edits over the sampling trajectory. This observation led us to task switching, which lets the model draw on its T2I capabilities. Across three unified editors and four
benchmarks, switching to the T2I task for bounded intervals improves edit quality, while mean perceptual preservation remains close to pure editing across all three models. Unified editors therefore benefit from using both conditioning modes they are trained for, and the timing of the switch sets the balance between quality and preservation.
Read the full abstract ▾
Systems and Control
Token Economy Design for Fair and Efficient Highway Congestion Management with Express Lanes
Leonardo Pedroso, Juan Pablo Bertucci, W.P.M.H. Heemels et al. (+1 more)
We study the design of a
token economy for highway lane allocation that aims to improve
fairness without sacrificing traffic efficiency. Motivated by the San Mateo 101 Express Lanes Project, we consider a setting in which high-occupancy vehicles have unrestricted access to an express lane, while the remaining users can alternate between regular and express lanes by earning and spending nonmonetary
tokens. We model the resulting interaction as a finite-population dynamic congestion game with heterogeneous time preferences and limited-information evolutionary policy revisions. Building on a mean-field approximation, we derive
token prices that enforce the system-optimal lane split while inducing
fairness over time through turn-taking. The scheme is evaluated in a microscopic traffic simulation with real-world demand data. The results show that the proposed prices yield nearly the same average travel time as a baseline scenario in which no lane is reserved as an express lane, while substantially reducing urgency-weighted perceived travel time. These findings highlight
token economies as a promising alternative to monetary congestion pricing for fairer management of scarce road capacity.
Read the full abstract ▾
All authors Leonardo Pedroso, Juan Pablo Bertucci, W.P.M.H. Heemels, Mauro Salazar
Solar and Stellar Astrophysics
Ensemble pulsational characteristics of eclipsing binaries containing $\beta$ Cep stars observed by the TESS mission
C. I. Eze, G. Handler, M. Vanrespaille et al. (+4 more)
Massive pulsators in eclipsing binaries provide precise, model-independent
stellar parameters that, combined with asteroseismic constraints, provide powerful probes of interior properties and angular-momentum transport in massive
stars. This study investigates the pulsational characteristics and their impact on the structure and
evolution of $\beta$ Cep pulsators in eclipsing binaries. We analyse an ensemble of 65 $\beta$ Cep
stars in eclipsing binaries, identify rotationally split modes where possible, and statistically investigate the dependence of pulsational properties on binary orbital parameters. We derive their interior properties using MESA model grids and associated StORM pulsation frequencies, together with an asteroseismic routine that tracks
stellar age and rotation across the $\beta$ Cep instability region. Interior rotation rates are derived for 14 systems that exhibit rotational mode splitting. While differential rotation is tentatively inferred for HD 112485, the other systems, except CZ Vel, are consistent with quasi-rigid rotation. The seismic and evolutionary properties of HD 112485 and EK Cru are derived through detailed forward asteroseismic modelling. HD 112485 is one of the least evolved $\beta$ Cep
stars asteroseismically modelled to date, with a core hydrogen mass fraction $X_{c}= 0.501_{-0.013}^{+0.009}$. The two reproduced zonal modes in HD 112485 are the first radial-order dipole pressure mode and the second radial-order quadrupole
gravity mode, while the zonal modes in EK Cru are the second-order quadrupole mixed mode and the third order dipole pressure mode. A multivariate regression shows a negative correlation of the pulsation amplitude with pulsation frequency, and no statistically significant relationship with the orbital period. In addition, no significant relationship is observed between the pulsation frequency and the orbital period in this sample.
Read the full abstract ▾
All authors C. I. Eze, G. Handler, M. Vanrespaille, A. Kemp, A. Tkachenko, D. M. Bowman, C. Aerts
Data Structures and Algorithms
Safe Hypergraph Contraction via Capacity-Aware Repair Certificates
Yu Deng, Xinyi Yang and Keren Zhu
Multilevel partitioners shrink
circuit hypergraphs through vertex contractions, yet a contraction that satisfies block capacity can still eliminate every optimal balanced bipartition. We develop certified safe coarsening (CSC) to identify contractions that preserve an optimum without computing that optimum. CSC certifies a repair for any feasible partition that splits a candidate group: the repair must respect the fixed block capacities and must not increase the cut-net objective. Its bounds exclude hyperedges that capacity constraints force to be cut. A pair certificate checks individual merges, while a directed minimum-cut test certifies groups whose savings emerge only when vertices move together. We prove that certified disjoint batches and successive rounds with recertification retain at least one globally optimal feasible partition for hypergraphs with positive integer vertex and net weights. Experiments on exactly solvable instances confirm optimum preservation for every tested configuration; integration with KaHyPar lowers the sum of per-instance best cuts on
circuit benchmarks, with additional runtime.
Read the full abstract ▾
Dynamical Systems
From trends to optimized forecasts: Quantifying the skill and advancing the utility of dynamics-based early warnings for tipping events
Franco Du Plessis, Victoria Volodina, Chris A. Boulton et al. (+5 more)
In this paper we explore and extend the state of the art for understanding the skill and utility of early warnings of critical transitions (tipping points) in forced nonlinear systems. We highlight some of the challenges and opportunities of transforming estimates of a dynamics-based early warning system, based on trends in stability or resilience into forecasts with high skill. We highlight the importance of (a) quantifying consequences of possible actions in response to warnings that may be false negatives or positives (b) considering finite time horizon predictions to give verifiable predictions (c) assumptions necessary for valid extrapolations of trends. We evaluate approaches to improving the skill and utility of the forecast.
Read the full abstract ▾
All authors Franco Du Plessis, Victoria Volodina, Chris A. Boulton, Muhammed Fadera, Sneha Kachhara, Timothy M. Lenton, Paul D. L. Ritchie, Peter Ashwin
Operator Algebras
On C*-nuclearity of graph operator systems
Adam Dor-On and Travis Russell
We classify the set of finite simple undirected graphs whose corresponding graph operator system is C*-
nuclear. We prove that the operator system associated to the graph is C*-
nuclear if and only if the graph is 2-chordal, as defined by McCullough. In particular, we present the first known example of a non-chordal graph whose graph operator system is C*-
nuclear, as well as the first known examples of graph operator systems which are not C*-
nuclear.
Read the full abstract ▾
Solar and Stellar Astrophysics
Delayed, Line-dependent Reorganization of Pore Oscillations Following a Compact C8.2 Flare Observed by Sunrise III/SUSI
Shahin Jafarzadeh, David B. Jess, Marco Stangalini et al. (+34 more)
Magnetic oscillations permeating the
solar atmosphere can be altered by
flares. This could be due to changes in the atmospheric
magnetic waveguide, the thermodynamic stratification, or both. To investigate this, we examine oscillations in a
magnetic pore crossed in part by the ribbon of a C8.2
flare, using observations from the Sunrise Ultraviolet Spectropolarimeter and Imager (SUSI) aboard Sunrise III. We track the core intensities of eight spectral lines between 327-329 nm and show that oscillatory power in the 4.2-7.0 mHz range, measured around 30 minutes after the
flare peak, exceeds the pre-event value in all eight lines by factors of 1.60-8.71. Conversely, in the 7.0-11.5 mHz band, the post-event intensity power is lower - approximately 0.13-0.41 of the original power in seven of the observed lines. Line-center velocity measurements support the lower-frequency enhancement. This delayed, line-dependent shift toward the lower-frequency band is consistent with
flare-induced changes in wave transmission, reflection, or resonant response, although thermodynamic and
magnetic effects cannot yet be separated.
Read the full abstract ▾
All authors Shahin Jafarzadeh, David B. Jess, Marco Stangalini, Lyndsay Fletcher, Ryan O. Milligan, Paulo J. A. Simões, Peter H. Keys, Sami K. Solanki, H. N. Smitha, Andreas Lagg, Achim Gandorfer, Alex Feller, Francisco A. Iglesias, Tino L. Riethmüller, Bianca Grauf, Johannes Hoelken, José Carlos del Toro Iniesta, Yukio Katsukawa, Pietro Bernasconi, Thomas Berkefeld, Alberto Álvarez-Herrero, Masahito Kubo, David Orozco Suárez, Michael Carpenter, Alexander Bell, Valentín Martínez Pillet, Francisco Javier Bailén, Julian Blanco Rodríguez, Juan Sebastián Castellanos Durán, Edvarda Harnes, Ryohtaroh T. Ishikawa, Yusuke Kawabata, Takuma Matsumoto, Takayoshi Oba, Azaymi L. Siu-Tapia, Hanna Strecker, Dušan Vukadinović
Machine Learning
Invent a Dataset: Measuring Dataset Generation Abilities with Zero Seed Data
Shivalika Singh, Andrija Djurisic, Gbemileke Onilude et al. (+2 more)
Building
datasets remains one of the most manual and brittle parts of AI development. In this technical report, we focus on the most extreme but also most prevalent setting real world practitioners face: a zero data regime. Here, practitioners don't have any data for the capability they want to learn. We introduce Invent-A-
Dataset which is a
prompt based system to go from
dataset description to realistic and large scale post-training
datasets. We evaluate Invent-A-
Dataset against five frontier model APIs including Anthropic, Google, Open AI, DeepSeek, Zai. Across eight task types and
dataset sizes up to 20K samples, Invent-A-
Dataset significantly outperforms with both the highest quality (17% relative gains) while simultaneously producing the most diverse samples (19% relative gains). Its diversity advantage widens with scale of training
dataset size (from parity at 200 samples to 37% relative gains at 20K samples). This translates into considerable downstream training gains, resulting in far more performant post-trained models. Invent-A-
Dataset fine-tune consistently ranks higher compared to other generator fine-tunes across different post-trained model architectures.
Read the full abstract ▾
All authors Shivalika Singh, Andrija Djurisic, Gbemileke Onilude, Sudip Roy, Sara Hooker
Optimization and Control
Simplifying the computation of weak-$\star$ second subderivatives of convex functionals via $\Gamma$-convergence
Gerd Wachsmuth
We prove (under some mild assumptions) that the strong-strong twice epidifferentiability of a convex functional is equivalent to the weak-$\star$ twice epidifferentiability of its conjugate functional. Further, the corresponding subderivatives are (up to scaling) conjugates of each other. This result can be used to facilitate the computation of weak-$\star$ second subderivatives, which are crucial ingredients in no-gap second-order optimality conditions. We also show that weak-$\star$ second subderivatives are invariant under suitable changes of the underlying function space setting.
Read the full abstract ▾
Statistical Mechanics
Scalar field theory for (chiral) active Brownian particles: bottom-up derivation revisited
Yuta Kuroda and Thomas Speck
The active Brownian particle (ABP) model is one of the most widely used particle-based models in studies of scalar active fluids. A collection of interacting ABPs is known to exhibit intriguing phenomena that are absent in
equilibrium counterparts. A prominent example is spontaneous phase separation, known as motility-induced phase separation (MIPS), which occurs even in the absence of explicit attractive interactions. In parallel, the large-scale behavior of scalar active fluids is often modeled using continuum descriptions, namely, scalar field theories. Such scalar active field theories are usually formulated in a top-down manner, and establishing their connections to particle-based models remains challenging, with these connections not yet fully understood. To advance the microscopic derivation of scalar active field theories, we revisit the connection between ABPs interacting via a two-body potential in two dimensions and scalar field theories that include all possible terms up to fourth order in spatial gradients. Our approach is based on a pressure expansion and the renormalization group (RG) method in the context of singular perturbation theory. The RG method provides a systematic way to eliminate fast variables and identify the dynamics on the slow invariant
manifold. Within this framework, scalar field theories can be obtained as RG flow equations. We also apply this method to a
chiral variant of the ABP model, the
chiral ABP (cABP) model, in which a constant torque biases particle rotation to the left or right. We show that the scalar field theory corresponding to the cABP model contains two “odd” terms in addition to those present in the scalar field theory for ABPs. The method developed here may also be useful for deriving hydrodynamic descriptions of other types of active matter systems.
Read the full abstract ▾
Category Theory
Ternary semi-direct products in semi-abelian categories
Arnaud Duvieusart
We study the construction of ternary semi-direct products in semi-abelian categories, following a definition of higher semi-direct product previously introduced by Carrasco and Cegarra for groups and Lie
algebras. We show that these objects are determined by actions and equivariant morphisms satisfying a certain compatibility condition in the ambient category, and that actions by semi-direct products are a special case of this construction. We also show how their structure can be further simplified in algebraically coherent or locally algebraically cartesian closed categories. We also give a concrete description of the structure of ternary semi-direct products in concrete categories, and how our categorical interpretation of the necessary structure relates to the group and Lie-theoretic versions given by Carrasco and Cegarra.
Read the full abstract ▾
Computer Vision
Do MLLM Judges Judge the Edit? Auditing Bias in Image Editing Evaluation with Verified Quality Preservation
Yuan Huang, Zirui Song and Xiuying Chen
Multimodal large
language models (MLLMs) are increasingly used as automated judges for instruction-based image editing and as reward signals for model training. However, systematically
auditing whether these judges are influenced by cues irrelevant to editing quality is challenging because visual interventions may themselves alter the quality being evaluated. A judgment shift can therefore be attributed to bias only when the intervention is verified to preserve the underlying editing quality. To address this challenge, we introduce EditJudgeBias, a counterfactual
benchmark with verified quality preservation, comprising 1,196 real editing samples and 13 cues injected across four evaluation sites. We verify quality preservation for the requested edit using calibrated
multimodal validators, controls, and human inspection. We then audit five MLLM judges along three complementary dimensions: invariance to quality-preserving cues, agreement with human judgments, and stability of pairwise preferences. Importantly, observed shifts are evaluated against each judge's own zero-dose and re-query noise floors rather than against zero. Experiments show that quality-preserving cues move every judge beyond its own noise. Fabricated majority opinions increase ratings, irrelevant visual elements cause larger shifts than whole-image manipulations, and swapping candidate order reverses up to 60.9% of pairwise decisions. Edit-region cues also tend to reduce human agreement. The three measures characterize judges differently, showing that robustness cannot be captured by a single metric.
Read the full abstract ▾