Artificial Intelligence
Pretrained ASR Pseudo-labeling for Noisy Police Audio
Kaavya Chaparala, Su Huang, Stephen L. Morgan et al. (+2 more)
Pretrained ASR systems perform poorly on noisy Broadcast Police Communication (BPC), hindering efforts to understand police decision-making. Pseudo-labeling offers an unsupervised path to improve ASR without expensive human labels, but the efficacy of this approach on very noisy domains is not known. In this work, we systematically assess the opportunities and limits of pseudo-labeling to adapt foundation ASR models (Whisper and Qwen3-ASR) to noisy BPC domain corpora from Baltimore and Chicago. We demonstrate that existing internal confidence metrics (log-probabilities and
STAR scores) fail to distinguish between high and low quality BPC pseudo-labels, and we introduce an external LLM-as-a-judge filtering paradigm that leverages parametric knowledge to discard contextually implausible transcripts. Our LLM-judging filters more aggressively than internal metrics and significantly reduces WER of the pseudo-labeled training sets across the Baltimore and Chicago BPC corpora, though a substantial gap remains relative to an oracle filter. We also introduce a new cross-model pseudo-labeling paradigm where one model is finetuned with pseudo-labels from the other, and we identify this method as a promising direction for future pseudo-labeling work.
Read the full abstract ▾
All authors Kaavya Chaparala, Su Huang, Stephen L. Morgan, Rhiannon N. Miller, Anjalie Field
Performance
SR-Gadgets: Make Scan-Resistant Caching Practical
Yunjia Zheng and Juncheng Yang
Block caches commonly serve scan-heavy I/O workloads, motivating extensive studies on scan-resistant eviction
algorithms. Many of these
algorithms adopt a multi-queue structure. However, they focus primarily on one-time scans and do not handle repeated scans well. Two important challenges from repeated scans are miss-ratio cliffs, where a small increase in cache size sharply reduces the miss ratio, and Belady's anomalies, where increasing the cache size increases the miss ratio. In this paper, we first develop two quantitative metrics to measure these behaviors. With these metrics, we find that LIRS is the only multi-queue
algorithm that is scan-resistant (almost cliff- and anomaly-free). Contrary to conventional wisdom, we show that stack distance is not the secret sauce that makes LIRS scan-resistant. Instead, regulating the queues are the key to its scan resistance. Based on these insights, we design the \gadgetprefix Gadgets, easy-to-integrate augmentations that make existing
algorithms scan-resistant without changing their eviction heuristics or queue structures. We implement the \gadgetprefix Gadgets in five
algorithms: S3-FIFO, SIEVE, ARC, 2Q, and TinyLFU, and make them scan-resistant. Evaluated on 5,538 production traces, all augmented
algorithms outperform their base versions, reducing miss ratios by up to 23.1% while consistently reducing cliffs and Belady's anomalies across the production traces.
Read the full abstract ▾
Computation and Language
Where Does Retrieval-Based Open-Ended Evaluation Fail? Automatic Taxonomy Induction from Long-Form Medical Answer Factuality Verification
Heyuan Huang, Jirui Dai, Alexandra DeLucia et al. (+5 more)
Retrieval-based factuality evaluation, where LLM-generated claims are verified against evidence from authoritative medical corpora, has become the dominant paradigm for scalable
hallucination detection in high-stakes clinical settings. Despite the urgency of reliable and transparent medical fact verification, most systems measure performance with aggregate metrics like F1, which obscure where and why failures occur. Existing RAG diagnostics require gold answers or annotated gold evidence, neither of which exists in this regime. We introduce two comprehensive taxonomies, grounded in a case study on the open-ended MedExpert
dataset and 3 closed-ended
datasets, decomposing failures into
retrieval-stage errors along five quality dimensions, and verifier-
reasoning errors into six consecutive steps. We adapt an automatic pattern induction pipeline using LLM-as-Judge to label evidence quality and classify verifier
reasoning errors at scale, and then stress-test our findings across 4
retrieval methods and 6 frontier verifier models. Our analysis reveals that scaling model size, adding
reasoning effort, expanding to authoritative web sources, and applying medical fine-tuning do not resolve these failure modes, demonstrating that they represent fundamental limitations of the retrieve-then-verify paradigm in open-ended medical settings rather than artifacts of outdated systems. We release our code and data at https://anonymous.4open.science/r/Medical_RAG_eval-4AB5 for the full reproducibility of our results.
Read the full abstract ▾
All authors Heyuan Huang, Jirui Dai, Alexandra DeLucia, Sonal Joshi, Mahsa Yarmohammadi, Jie Gao, Bernal Jiménez Gutiérrez, Mark Dredze
Human-Computer Interaction
A Benchmarking Framework for Context-aware XR Interfaces
Hyunsung Cho, Sarah Yewon Yun, Nancy Ruonan Sun et al. (+7 more)
Everyday Extended Reality (XR) systems aim to provide context-aware access to the right functionalities at the right time and place, with minimal manual reconfiguration as users switch context. Yet these interfaces are hard to evaluate: current prototyping and user-study workflows offer no systematic, repeatable way to compare adaptation methods across users and scenarios. We present ContextXR, a novel benchmarking framework for context-aware XR interfaces. ContextXR represents an XR application as a connected graph of functional facets, each a semantically coherent group of related capabilities that together support a shared user intent. On this representation, we build MineXR++, a
dataset augmenting prior XR interface data with facet-level annotations, and formulate three canonical tasks of context-aware suggestion: context factor analysis, initial facet suggestion, and next facet suggestion. Our evaluation protocol scores suggestion methods by a simulated interaction metric, the
navigation and search cost of reaching the desired functionality. Through experiments benchmarking global popularity, relational
retrieval, and LLM-based methods, we demonstrate that ContextXR enables the systematic, reproducible evaluation of context-aware XR interfaces.
Read the full abstract ▾
All authors Hyunsung Cho, Sarah Yewon Yun, Nancy Ruonan Sun, Ben Lafreniere, Mark Parent, Kashyap Todi, Tanya R. Jonker, Hrvoje Benko, Sherry Tongshuang Wu, David Lindlbauer
Machine Learning
RAZOR: Pruning Replaceable Experts in LLMs
Mingyang Song and Mao Zheng
Mixture-of-experts (MoE) models activate only a few experts per
token but store the entire expert pool.
Pruning this pool requires identifying experts whose removal preserves model behavior. Routing frequency and output magnitude do not fully describe deletion damage, which also depends on how the surviving and replacement experts compensate for the removed output. We introduce RAZOR, a training-free
pruning method based on consensus residuals, the deviations of expert outputs from their original weighted mixture. At a fixed layer input, these residuals give the exact output change for a single deletion under survivor renormalization and router refill. RAZOR aggregates this damage by conditional root mean square and selects experts under a layerwise budget using forward computation alone, without gradients, subset search, or recovery training. Against frequency, activation-norm, and REAP baselines on GLM-4.7-Flash and Qwen3.6-35B-A3B at 25% and 50% expert removal, it attains the highest macro average over nine
reasoning-intensive tasks in all four model-budget settings, gaining 2.12-5.59 points over REAP and lowering reverse KL in all four. On DeepSeek-V4-Flash-0731 and Hy3, it also achieves the highest macro average among the three residual criteria. Local exactness does not guarantee better joint
pruning. Generation analyses show changes in diversity, formatting, and termination despite higher task scores.
Read the full abstract ▾
Theoretical Economics
The Robustness of Sequential Implementation under Information Perturbations
Ian Ball, Drew Fudenberg and Stephen Morris
Bayesian monotonicity is a necessary condition for full implementation in Bayes-Nash
equilibrium. Under the sequential
equilibrium refinement, however, multi-stage mechanisms can expand the scope for implementation. We show that this additional implementation power is fragile. Whenever a multi-stage mechanism sequentially implements a social choice correspondence that is not
Bayesian monotone, we construct arbitrarily small "information perturbations" and show that there are sequential
equilibria under these perturbations that yield inadmissible decisions with probability bounded away from zero. Under these perturbations, players' types are not independent conditional on the state and some player is sometimes uncertain of their payoff type. We show by example that the general result fails if perturbations are required to satisfy conditional independence or "known payoff types."
Read the full abstract ▾
High Energy Astrophysical Phenomena
Infrared Cavity Surrounding T Pyxidis: Evidence of a Nova Super-remnant?
Michael W. Healy-Kalesh and Matthew J. Darnley
Nova super-remnants (NSR) are shell-like structures spanning tens of parsecs that are predicted to grow around all recurrent novae as a result of repeated ejecta sweeping up the surrounding material. After the first NSR was discovered in the Andromeda
Galaxy around M31N 2008-12a, a number of
Galactic NSRs have been uncovered, as has the first NSR in the Large Magellanic Cloud. Here we report an apparent infrared cavity surrounding the
Galactic recurrent nova, T Pyxidis, which we suggest may be a nova super-remnant and discuss the implications.
Read the full abstract ▾
Robotics
Policy-Calibrated DAgger: Offline Calibrated Noise Injection for Imitation Learning
Jenny Wang and George Kantor
Policies trained with imitation learning can accumulate errors over time, causing the
robot to drift outside the training distribution. Existing methods mitigate this
covariate shift by collecting additional data where the policy fails or is likely to fail. The first places the
robot in unsafe conditions and the second requires choosing an appropriate noise distribution to collect new expert demonstrations under that noise. We propose Policy-Calibrated DAgger, a method that makes use of the properties of recent generative policies to estimate the policy's noise offline by using its own predicted action distribution. We measure a
diffusion policy's spread of predicted actions at observations along the expert trajectory and measure its closed-loop error relative to a recorded trajectory. To address issues with measuring error in a
multimodal action space, we guide the policy towards the trajectory during closed-loop control through partial denoising, and use properties of a
diffusion model to unnormalize the measured error as if we did not guide it. We experiment in a scenario where a
robot is tasked to reach an engine lever in a cluttered and narrow environment and show results in a 3D photorealistic simulator and a 2D planar reacher environment. We show that our method surpasses policies trained with
dataset aggregation without noising and matches the performance of the best noise level in hindsight, without requiring a sweep over noise levels.
Read the full abstract ▾
Robotics
DGT-Map: Directional Global Traversability Mapping Utilizing Multi-Task Learning for Heterogeneous Vehicles
Jaskrit Singh, Kashif K. Noori, Jing Xiao et al. (+1 more)
Off-road traversability is direction-dependent and vehicle specific, yet most global maps assign a single isotropic cost to each location. Existing learned
estimators are also commonly trained independently for each vehicle; this preserves vehicle-specific behavior but prevents vehicles from sharing common terrain representations. DGT-MAP addresses both limitations through a self-supervised framework that learns global, directional, and vehicle-conditioned traversability costmaps from RGB-D observations and locomotion signals. A shared multi-task backbone learns common terrain features across training vehicles while vehicle-specific prediction heads preserve platform-dependent responses. At inference, DGT-MAP produces a heading-indexed costmap that can be used by a direction-aware planner. We evaluate DGT-MAP in simulation by integrating it into a Hybrid A*
navigation stack and measuring downstream task success on challenging terrains, including slopes that are traversable downhill but not uphill and a ridge obstacle that is traversable by some vehicles, but not by others. Across evaluated tasks, DGT-MAP achieves the highest or tied-highest
navigation success rate when compared against
geometric, binary, and learned direction-agnostic baselines.
Read the full abstract ▾
All authors Jaskrit Singh, Kashif K. Noori, Jing Xiao, Constantinos Chamzas
Robotics
Realizability Is Not Enough: Encoding, Liveness, and Auditing of Synthesized Robot Supervisors
David C. Conner, Joshua Luzier, William J. Doyle et al. (+7 more)
High-level robotic supervisors coordinate capabilities whose reported outcomes determine the
robot's next action. Reactive synthesis can generate such supervisors with formal guarantees, but deployment requires more than proving a Generalized Reactivity (1) (GR(1)) specification realizable. Designers must encode failure-prone capabilities, choose liveness assumptions that match retry intent, audit strategies, and translate them into
robot software. We present an open-source pipeline for
Robot Operating System (ROS) 2 Flexible Behavior Engine (FlexBE) supervisors that generates capability-based GR(1) specifications, analyzes assumptions before synthesis, audits strategies, reduces states with a behavior-preservation proof, and emits executable state machines. Across four case studies (six comparisons), including hardware on two quadcopter platforms, we compare enumerated and one-hot encodings and two liveness formulations. Under the tested backend, enumerated encoding usually synthesizes faster, although fewer propositions do not reliably predict smaller controllers or lower symbolic cost. System-Goal without pending
memory is the only liveness treatment confirmed to yield executable controllers under both encodings across the reported grid; Fair-Outcome can permit realizable cycles without designer-intended completion. For this backend and model, we recommend enumerated encoding with System-Goal and
auditing every realized strategy, since proposition count and realizability do not measure deployability. The auditor is sound and complete for four structural defect classes (protocol violations, deadlocks, bounded-failure violations, goal-unreachable traps) but is not a general liveness verifier, and the reduction preserves capability-level behavior. Together, these stages narrow the gap between formal realizability and controllers that pass protocol and structural-progress checks.
Read the full abstract ▾
All authors David C. Conner, Joshua Luzier, William J. Doyle, Emma R. Faith, Aubrie B. Kooiker, Andrew J. Farney, Sebastian Fox, Evangelina Grimes, Ian G. Conner, Kyle Bloom
Robotics
VkVIO: Cross-platform GPU Acceleration for Visual-Inertial Odometry with Vulkan
Ole Hoffmann, Mateo de Mayo and Daniel Cremers
Perception in robotics and XR fundamentally relies on good state estimation. Visual-inertial odometry (VIO) and Simultaneous Localization and Mapping (VI-SLAM) are proven ways of achieving this goal in a cost-effective and accurate manner. Efficiency in these systems allows for smaller, cooler, and lighter devices. GPU acceleration is a natural approach for reducing latency, thanks to their wide availability in platforms like embedded computers, mobile phones, and XR headsets. However, previous works in the literature have limited themselves to the use of CUDA for this task, significantly reducing deployment options to a single vendor. We instead leverage the vendor-agnostic Vulkan API, originally designed for the strict performance requirements of 3D graphics applications. In this work, we present VkVIO, the first, to the best of our knowledge, cross-platform GPU-accelerated VIO method. We provide state-of-the-art accuracy with causal estimates required for real-time operation. We deploy VkVIO on a diverse range of devices spanning a workstation, a laptop, and an extremely inexpensive single-board computer, while outperforming CUDA-based systems on the same hardware. VkVIO enables possibilities for low-latency, low-power, and low-cost VIO in robotics and XR.
Read the full abstract ▾
Materials Science
Hydrogen-stabilized multimodal high-index twin network in iron
Mehrab Lotfpour, Haoran Cui, Yan Wang et al. (+2 more)
Hydrogen significantly affects plasticity in bcc iron, but its role is commonly attributed to dislocation-mediated mechanisms, leaving the influence of hydrogen on phase transformation and deformation twinning poorly understood. We use a density-functional-theory-trained deep-neural-network interatomic potential and large-scale molecular dynamics to study pure bcc Fe and Fe containing 10% hydrogen. It is found that hydrogen lowers the yield stress but, more importantly, increases the persistence of {112} twin variants and suppresses detwinning. This effect stabilizes an interconnected {332}-{10 9 3}
multimodal twin network and produces pronounced post-yield hardening. Higher temperature promotes the initial transformation but weakens the persistence of the high-index
multimodal twin network. Moreover, twinning follows distinct loading-dependent pathways: compression generates {332} and {10 9 3} boundaries through co-zone and non-co-zone twin-twin interactions, whereas tension produces {7 4 1} boundaries through non-co-zone interactions. These results establish a pathway-based picture in which the intermediate phase determines the accessible twin modes and variant crystallography, while hydrogen and temperature control their kinetic survival and the emergence of high-index twin networks.
Read the full abstract ▾
All authors Mehrab Lotfpour, Haoran Cui, Yan Wang, Eduardo Vitral, Lei Cao
Analysis of PDEs
Isothermal hydrodynamic limit for kinetic flocking models with nonlinear velocity alignment
Roman Shvydkoy and Changhui Tan
We study the hydrodynamic limit of kinetic flocking models with nonlinear velocity
alignment and strong local Fokker–Planck forcing. In the limit of small Knudsen number the kinetic densities converge to a local Maxwellian centered around macroscopic quantities satisfying the compressible Euler system with isothermal pressure $p = \s \rho$, while macroscopic
alignment is obtained by convolution of the microscopic velocity law with the standard Gaussian, resulting in a different nonlinearity. Such discrepancy has been observed already in \cite{black2025hydrodynamic}. In this note we rigorously justify the limit for admissible weak kinetic solutions and non-vacuous limiting macroscopic solution. Our analysis provides a quantitative rate of convergence in terms of relative
entropy: $\cH(f_\e | \mu) \lesssim \e$ provided initially $\cH(f_\e(0) | \mu(0)) \leq \e$. In the linear case we recover the known result \cite{karper2015hydrodynamic}.
Read the full abstract ▾
Artificial Intelligence
Shai Dickman, Mert Cemri, Landon Butler et al. (+1 more)
Discrete
diffusion models have become leading architectures for inverse
protein folding. Test-time
alignment guides these models during inference to generate
proteins with desirable attributes.
Alignment methods for discrete
diffusion models often treat inference as a unidirectional process, lacking effective mechanisms for revisiting undesirable
token selections. We introduce Spectral Feedback, an
algorithm that selects edit-positions in a feedback loop, allowing the model to iteratively correct its own generations. This approach leverages the masking structure of discrete
diffusion models by re-masking and re-sampling
tokens. While prior
alignment methods focus on what
token labels to assign to maximize a target reward, we instead treat which
tokens to revisit as the central
alignment problem. Selecting edit-positions is challenging because edit effects are interdependent: the impact of modifying one
token depends on which others are edited simultaneously. We define an edit-set as a set of positions to edit by re-masking and re-sampling the corresponding
tokens in a sequence. Motivated by prior work on sparse interactions in biological systems, we find empirically that edit-set value functions for
protein inverse folding admit sparse
Fourier representations. This structure enables Spectral Feedback to efficiently learn and optimize the value functions for edit-position selection. Spectral Feedback is model-agnostic and can be applied to pre-trained, test-time aligned, and fine-tuned
diffusion models without modifying the underlying generative processes. Applied to inverse folding with a
protein stability reward oracle, Spectral Feedback increases the proportion of stable
proteins by 32.3 percentage points over a pre-trained model, 24.8 over Best-of-10, and 5.8 over a state-of-the-art RL fine-tuned
diffusion model.
Read the full abstract ▾
All authors Shai Dickman, Mert Cemri, Landon Butler, Kannan Ramchandran
Statistical Mechanics
Critical escape dynamics of an active particle moving on curved surfaces
Maxim Root, Hartmut Löwen, Peter Sollich et al. (+2 more)
On the macroscopic scale, active (self-propelled) objects moving through complex environments are often subject to effects arising from both the
curvature of the surrounding surface and external influences such as
gravity. We study the Kramers escape problem for active Brownian particles in two settings: motion in an external potential driven solely by gradient forces, and motion on a curved
manifold subject to an additional constant gravitational force. In the absence of translational noise, a sharp dynamical transition occurs when the self-propulsion velocity $v_0$ reaches a critical velocity $v_c$ required for an escape. Above this threshold, the escape rate $k$ follows a universal scaling form, $k \propto \exp\left[-\mathrm{const}(v_0-v_c)^{-\gamma}\right]$, with an exponent $\gamma=3/2$ for the escape in a potential and $\gamma=1/2$ for the escape on a curved
manifold in the presence of
gravity. These distinct exponents reveal how the interplay of activity, surface
curvature, and
gravity determines the critical escape dynamics. Our theoretical predictions are verified by numerical simulations.
Read the full abstract ▾
All authors Maxim Root, Hartmut Löwen, Peter Sollich, Lorenzo Caprini, Alexander P. Antonov
Machine Learning
Auditing System-1 Models on Biosecurity-Relevant Benchmarks: Calibration, Selective Prediction, and Permutation Instability in a Non-Generative Model
Kimon Antonios Provatas and Ilias Georgakopoulos-Soares
Non-generative "System-1" models return structured probabilistic decisions in a single forward pass, without autoregressive decoding, at a small fraction of the inference cost of a generative model. This makes them of interest as inexpensive components in larger pipelines, but their reliability on biosecurity-relevant tasks has not been systematically examined. We audit one commercial System-1 model on 6,020 multiple-choice items drawn from the Weapons of Mass Destruction Proxy (WMDP), a paraphrase-robust WMDP-Bio variant, and six LAB-Bench subtasks, measuring accuracy, calibration, error detection, selective prediction, and sensitivity to the order in which answer options are presented. Accuracy is strongly task-dependent. Once the vendor's uncertainty field is correctly interpreted, the model is reasonably well calibrated (pooled expected calibration error 0.034) and its top-1 probability separates correct from incorrect predictions (pooled AUROC 0.820), though both degrade substantially on the weaker tasks. Under four cyclic rotations of the answer options, 37.4% of WMDP-Cyber items receive different answers; a control using byte-identical repeated calls attributes most of this to option order rather than run-to-run variation. Averaging probabilities across rotations improves WMDP-Cyber accuracy by 3.8 percentage points, and applying it only to low-confidence items recovers most of that gain at well under the cost of averaging every item.
Read the full abstract ▾
High Energy Physics (Theory)
de Sitter Spacetime: Geometry, Causal Structure, and Quantum Fields
Azadeh Maleknejad and Ethan Richardson
de~Sitter spacetime plays a central role in modern cosmology, providing an excellent approximation to both inflation and the present accelerated expansion. It also offers a particularly simple setting in which
gravity, spacetime
symmetry, and
quantum field theory depart qualitatively from their Minkowski-space counterparts. These lecture notes introduce de Sitter spacetime, from its geometry and
symmetries to
quantum fields and perturbations on a fixed \(3+1\)-dimensional de Sitter background. We cover coordinate systems, causal structure,
unitary representations, particle production in time-dependent backgrounds, and cosmological perturbations, with emphasis on physical interpretation, applications to cosmology, and how
curvature reshapes the notions of mass,
vacuum, particles, and late-time observables.
Read the full abstract ▾
Plasma Physics
Demonstration of H-mode Error Field Identification in a Single Discharge via Island Healing
E.M. Bursch, A. Xie, J.L. Barr et al. (+15 more)
Non-disruptive compass scan error field identification via island healing is demonstrated on KSTAR in a single H-mode discharge. This method can allow for up to a 75% reduction in the operational time necessary for a complete compass scan by reducing the required discharges from four to one. Previous experiments on DIII-D, JET, and MAST-U have demonstrated the technique in Ohmic and L-mode but required multiple discharges and strong real-time n=1 magnetics diagnostics close to the
plasma that are unlikely to be viable for future devices. The results presented here were made possible by implementing a new set of triggering
algorithms relevant to fusion pilot plant operation into the KSTAR
plasma control system, along with KSTAR's strong RMPs and long pulse lengths. This represents a significant step towards the disruption-free error field identification method being demonstrated for deployment on future disruption-averse tokamaks, including ITER and fusion pilot plants. Steps needed to close remaining gaps to viability are addressed.
Read the full abstract ▾
All authors E.M. Bursch, A. Xie, J.L. Barr, S.K. Kim, N.C. Logan, S.M. Yang, J.G. Bak, W. Choi, Q. Hu, J.W. Juhn, J. Kim, J. Lee, N. Leuthold, J.K. Park, C. Paz-Soldan, L. Piron, G. Seo, J. Seo
Systems and Control
Practical Algebraic Parameter Estimation for Noisy Data via Gaussian Process Regression
Oren Bassik, Alexander Demin and Alexey Ovchinnikov
Parameter estimation for ordinary differential equation (ODE) models is a fundamental task that is often complicated by the limitations of conventional optimization-based methods. In theory, differential-algebraic approaches offer an appealing alternative: they reduce the problem to
polynomial system solving and do not require user-supplied initial guesses for parameter values. In practice, however, algebraic methods have been limited by their sensitivity to measurement noise, because they require accurate derivatives of observed outputs. In this work, we integrate Gaussian Process Regression (GPR) into the differential-algebraic method and derive a first-order error analysis in terms of noise level and algebraic sensitivity. We evaluate the method across several noise levels on a
benchmark of 25 dynamical systems arising in applications including mechanical engineering and systems biology. The proposed method achieves the highest aggregate performance among the methods considered, recovering all sought parameter values and initial conditions to within 10% relative error in 88.5% of runs. These results demonstrate that robust derivative estimation can make differential-algebraic parameter estimation practical for dense, noisy synthetic data while retaining key advantages of the algebraic formulation.
Read the full abstract ▾
Computer Vision
LensDesigner: A Self-Improving Agent for Optical Lens Design
Lei Sun, Haoran Liang, Dannong Xu et al. (+6 more)
Optical lens design is a complex, non-convex optimization challenge that relies heavily on human experience and intuition. Existing optimized-based automatic lens design methods struggle to navigate this vast parameter space without meticulous manual tuning. In this paper, we present LensDesigner, an autonomous
agent framework that mirrors the problem-solving workflow of expert opticians. To overcome the initial cold start problem, we construct LensLib100K, an extensive optical lens library, and employ Optics-Aware
Retrieval to supply physically valid structural seeds. Within an interactive physical simulation environment, the
agent executes macroscopic orchestration while receiving immediate optical feedback. Furthermore, we introduce a continuous self-evolving mechanism guided by a curriculum
agent. By iteratively solving design tasks with progressively increasing difficulty, the
agent autonomously extracts, accumulates, and reuses design heuristics, effectively evolving its optical lens design expertise over time. At the evaluation level, we introduce LensArena, a standardized evaluation
benchmark comprising $120$ diverse optical design tasks, covering extreme configurations. Extensive experiments on this
benchmark demonstrate that LensDesigner significantly outperforms publicly available baseline
algorithms, achieving superior success rates and optimization efficiency. We hope this work sheds light on the emerging field of intelligent optics. The code will be publicly available.
Read the full abstract ▾
All authors Lei Sun, Haoran Liang, Dannong Xu, Yao Gao, Yuyu Geng, Jinjin Gu, Kaiwei Wang, Danda Pani Paudel, Luc Van Gool
Human-Computer Interaction
Who Acts When the User Is Gone? Digital Remains, Survivor Claims, and Post-Mortem Governance
Supriya Khadka, Dhiman Goswami and Sanchari Das
Digital systems continue to govern accounts, devices, data, and recovery channels after an account holder dies, leaving survivors to manage digital remains through mechanisms built around a living user. We examine post-mortem digital governance as a sociotechnical problem of cooperative and contested work, focusing on who acts when the user is gone, what claims they make, and what barriers shape recovery, preservation, closure, and protection. We conducted a content analysis of $800$ Reddit posts about post-mortem digital
privacy and security, coding posts across assets, actors, actions,
privacy tensions, access barriers, policy gaps, emotional contexts, and risks. Findings show that phones/devices often act as gateways to other digital remains, socially connected actors make most claims, and data loss emerges as a central harm. We synthesize these findings into a Post-Mortem Digital Governance Framework for designing mechanisms that support survivor coordination while limiting access by purpose, asset, actor, and context.
Read the full abstract ▾
Performance
To Store or To Regenerate? A Cost Model for AI-Generated Content at Scale
Yunjia Zheng, Zirui Wang, Haoran Ni et al. (+4 more)
AI-generated content is becoming a rapidly growing class of digital artifacts. Because these artifacts accumulate over time, their exponential growth creates a substantial storage, energy, and infrastructure cost for operators and society. At the same time, GPU compute cost continues to fall rapidly with each hardware generation. This divergence raises a fundamental question: when does on-demand regeneration become cheaper than persistent storage? This paper develops a cost model for comparing persistent storage and on-demand regeneration for AI-generated artifacts. The model accounts for corpus growth, HDD and tape price trends, drive replacement, electricity, request skew, caching, generator FLOPs, and future GPU price-performance improvements. For image generation, our analysis shows that
prompt-based regeneration does not become cheaper than storage until around 2040, because every cache miss must still rerun the full
prompt-to-artifact generation pipeline. We observe that widely used
diffusion-based generation models operate in latent space, creating an alternative point in the cost tradeoff: instead of storing the final artifact or only the
prompt, operators can store a compact intermediate representation (IR) and perform cheap on-demand decoding. Our analysis shows that caching combined with IR-based regeneration substantially reduces both storage and compute cost, making it at least 2x cheaper than both full-object storage and
prompt-based regeneration even today. On a production image trace with 2.07 billion requests, the same conclusion holds:
prompt-based regeneration is over 100x more expensive than storage, while IR-based regeneration reduces total cost to roughly half that of full-object storage while preserving interactive miss latency.
Read the full abstract ▾
All authors Yunjia Zheng, Zirui Wang, Haoran Ni, Tingfeng Lan, Zhaoyuan Su, Yue Cheng, Juncheng Yang
Rings and Algebras
The Hua identity implies alternativity
Pasha Zusmanovich
The well-known Hua identity $a - (a^{-1} + (b^{-1} - a)^{-1})^{-1} = aba$ holds in any associative division ring. We prove, in a sense, a converse: if a (nonassociative) ring with unit and multiplicative inverses satisfies the Hua identity, then it is alternative.
Artificial Intelligence
Predicting Transmembrane Protein Topology from 3D Structure
Sitong Chen and Xiaopeng Mao
This paper presents a novel approach to infer
protein topology using the state-of-the-art graph neural network (GNN), SchNet. The model is trained on the same
dataset used to develop the recent DeepTMHMM model with 5-fold cross-validation. Unlike the conventional approaches based on using only the
protein sequences or the $\alpha$-carbons as features, we have decoded our classifier in this way, so all atom-level
embeddings are used. Without applying any pre-trained weight, the final results have shown great potential that GNNs can be used for topological predictions.
Read the full abstract ▾
Machine Learning (Statistics)
Bayesian Uncertainty Quantification for fMRI Functional Connectivity via Simulation-Based Inference
Simon Carter, Zeming Kuang, Lilianne R. Mujica-Parodi et al. (+1 more)
Optimizing fMRI scan duration and spatial resolution is critical for experimental design, yet traditional correlation-based approaches cannot quantify uncertainty or disentangle scanner measurement noise from true neural variability across subjects. Without principled uncertainty bounds, researchers cannot know whether a protocol is long enough to reliably estimate connectivity, or whether between-subject differences reflect biological variation or noise. We present a
Bayesian framework modeling BOLD dynamics as coupled Ornstein-Uhlenbeck processes, using Sequential Neural Posterior Estimation to obtain connectivity posteriors while accounting for frequency-independent measurement noise across the BOLD spectrum. Applied to N = 28 healthy controls (55 scans) at 7T using a functional network atlas (65 DMN regions), the framework quantifies uncertainty across its sources: scanner noise, subject variability, and acquisition length. Spatial analysis identifies a mean of 46 voxels per ROI, roughly half of typical region sizes, as sufficient to achieve 90% of asymptotic precision. At the single-subject level, 7T reaches its within-session precision plateau in approximately 7 minutes versus 10 minutes for 3T, a 40% reduction in required scan time, providing the first direct, model-based quantification of the scan-time advantage conferred by higher field strength. At the population level, 3T requires roughly 37 times more per-subject scan time than 7T for the pooled curves to converge, confirming a consistent advantage of higher field strength at every timescale. Together these findings provide concrete, scanner-specific guidance for protocol optimization, with direct implications for reducing acquisition costs and improving the reliability of connectivity-based clinical biomarkers. We provide code enabling researchers to derive these bounds from their own data.
Read the full abstract ▾
All authors Simon Carter, Zeming Kuang, Lilianne R. Mujica-Parodi, Helmut H. Strey
Category Theory
Graduated categories of presheaves
Jiri Adamek and Lurdes Sousa
We study conditions on a small category $\mathcal{A}$ under which every finitely generated object of its presheaf category is finitely presentable. Example: this holds for a group $\mathcal{A}$ iff $\mathcal{A}$ is Noetherian. For ordinals $\mathcal{A}=\alpha$ this holds iff $\alpha\leq \omega$, whereas this always holds for $\mathcal{A}=\alpha^{op}$. Various important properties of set functors also apply to endofunctors on locally finitely presentable categories which are graduated. This means that every finitely presentable object $X$ carries a grade (in $\mathbb{N}$) and grades respect subobjects and strong quotients of $X$. We characterize presheaf categories $\mathbb{Set}^{\mathcal{A}^{op}}$ which are graduated for three types of small categories $\mathcal{A}$. If $\mathcal{A}$ is a poset, all down sets of elements must be finite. For a group $\mathcal{A}$, a finite bound on the length of a chain of subgroups must exist. In the case of cartesian categories $\mathcal{A}$, every object must carry a finite number of sieves. Example: all finitary set functors form a graduated category, here $\mathcal{A}$ is the dual of finite sets. A closely related concept is a locally finitely presentable category with the descending chain condition (a DCC category): every finitely presentable object has only finite descending chains of subobjects or strong quotients. The presheaf category on a group $G$ is $DCC$ iff $G$ has no infinite chain of subgroups. Example: presheaves on the group $\mathbb{Z}$ are not $DCC$, but finite generation implies finite presentation.
Read the full abstract ▾
General Relativity and Quantum Cosmology
Quasinormal modes and shadow of Schwarzschild-Tangherlini black hole with GUP
J. A. V. Campos, M. A. Anacleto, F. A. Brito et al. (+2 more)
In this work, we investigate quasinormal modes and the shadow of a Schwarzschild-Tangherlini black hole in $(d+1)$ dimensions, incorporating corrections arising from the Generalized Uncertainty Principle (GUP). Using the WKB approximation up to sixth order, we calculate the quasinormal frequencies for $l=1, 2$ and $d=3, 4, 5, 6$, and analyze the effects of the dimensionless linear and quadratic GUP parameters $\bar{\alpha}$ and $\bar{\beta}$, respectively. We find that the linear correction increases the oscillation frequency and the damping rate, whereas the quadratic correction produces the opposite behavior within the parameter range considered. Using the finite difference scheme, we obtain the time-domain
evolution and find that higher dimensions lead to faster damping and that the visibility of GUP-induced modifications to the time-domain waveform depends on the excitation profile of the initial perturbation, whereas the underlying quasinormal spectrum remains determined by the black hole geometry.Furthermore, we investigate the shadow and establish its correspondence with the quasinormal spectrum in the eikonal limit through the properties of the unstable circular null geodesics. Finally, we confront the predicted shadow size with Event Horizon
Telescope observations of Sagittarius A* and use these constraints to investigate the allowed GUP parameter space.
Read the full abstract ▾
All authors J. A. V. Campos, M. A. Anacleto, F. A. Brito, E. Passos, Amilcar R. Queiroz
Algebraic Geometry
Symmetry and solvability: Galois groups of equivariant enumerative problems
Thomas Brazelton, Alberto Landi and Sidhanth Raman
At the genesis of Galois theory at the end of the 19th century, enumerative geometers were concerned with the solvability of various problems: for instance whether one can express lines on a cubic surface in radicals in terms of the coefficients that define it. By work of Hermite and a celebrated
theorem of Harris, this is equivalent to determining whether the monodromy group of a certain finite étale cover is solvable. This problem, as well as the related problem of computing bitangents to plane quartics, are both unsolvable. When one restricts to the locus of $G$-symmetric cubic surfaces, monodromy will drop and can become solvable. In this setting there are four
flavors of monodromy one can consider: the one from restricting the classical cover, the one over the GIT quotient after modding out by projective transformations, and two related notions of "monodromy" coming from the stacky cover of $G$-symmetric lines on $G$-symmetric cubic surfaces - unlike in the non-symmetric setting these are all different. In this paper we characterize the relationships between all four notions of monodromy in the general setting of quotient stacks, and use these connections to compute all these monodromy groups for all 11 possible
automorphism groups of smooth cubic surfaces, as well as for all 12 possible
automorphism groups for smooth planar quartics. We demonstrate that, in the presence of any
symmetry whatsoever, the problem of solving for lines on cubic surfaces or bitangents to plane quartics is solvable.
Read the full abstract ▾
Astrophysics of Galaxies
S. Fujimoto, Q. Fei, G. B. Brammer et al. (+41 more)
We report a long-lived
nuclear transient in GNz7q, a red
quasar at $z=7.19$ powered by a $\sim3\times10^{7}\,M_\odot$ black hole and hosted by a compact, dusty starburst
galaxy. The transient was identified in more than two decades of imaging with HST, Spitzer, and JWST. After a rapid rise, the source faded smoothly by $\Delta m\simeq1.1$ mag in the rest-frame ultraviolet over $\sim2$ rest-frame yr, with coherent, wavelength-dependent
evolution across 1–5 $\mu$m that places GNz7q above the 99.9th percentile of the SDSS
quasar variability distribution and is absent at $z\gtrsim5$. The duration and energetics disfavor superluminous
supernovae, and
stochastic quasar variability reproduces the multi-band
evolution with a probability of $\lesssim10^{-5}$. A panchromatic model combining a thermal continuum with a $t^{-5/3}$ decline reproduces the light curves, yielding a peak bolometric luminosity of $\simeq3\times10^{45}$ erg s$^{-1}$ and a radiated energy of $\simeq1.5\times10^{53}$ erg, which imply the
tidal disruption of a
star of a few
solar masses and place the event among the most energetic
tidal disruption events (TDEs) known. A redshifted, extremely broad Balmer-line component in independent JWST/NIRSpec spectroscopy is consistent with a transient, non-virialized broad-line region. JWST/MIRI photometry further reveals delayed mid-infrared emission from $\simeq1500$ K
dust at a sub-parsec radius, consistent with a
dust echo of the
flare. The relatively low black-hole mass and dense
star-forming
nucleus of GNz7q are conditions under which TDEs are expected to be most efficient. These observations provide a time-domain view of episodic black-hole fueling and its dusty
nuclear environment 700 million years after the Big Bang. Wide-field surveys with Roman and Euclid in the near-infrared, complemented by LSST at lower
redshifts, may uncover such transients in large numbers.
Read the full abstract ▾
All authors S. Fujimoto, Q. Fei, G. B. Brammer, K. Inayoshi, V. Kokorev, L. C. Ho, R. Li, F. Walter, V. Bromm, L. Colina, P. Dayal, S. L. Finkelstein, M. Ginolfi, Z. Liu, G. C. K. Leung, G. E. Magdis, J. Matthee, R. P. Naidu, P. Oesch, M. Onoue, P. G. Pérez-González, D. Watson, J. Álvarez-Márquez, J. Antwi-Danso, Y. Asada, G. Barro, C. M. Casey, J. Chisholm, S. Gillman, R. Marques-Chaves, J. P. U. Fynbo, T. R. Greve, I. Labbe, R. A. Meyer, F. Rizzo, A. Robbins, R. Schneider, C. L. Steinhardt, S. Toft, A. Trinca, F. Valentino, R. Valiante, M. Vestergaard, M. Xiao
Astrophysics of Galaxies
SN Helios: A Multiply Imaged Type II Supernova Opening Time-Delay Cosmography beyond redshift of 3
Seiji Fujimoto, Conor Larison, Justin D. R. Pierel et al. (+63 more)
We report the discovery and spectroscopic confirmation of SN Helios, a strongly lensed, multiply imaged Type II
supernova (SN) at $z=3.34$ behind the
galaxy cluster RXC J0018.5+1626. The transient was identified as a red F150W-dropout source (F277W = 26.1 mag) in JWST/NIRCam imaging taken on 2026 July 3 for the VENUS survey, with no counterpart in 2024 December imaging. Director's Discretionary Time observations on 2026 August 17 added second-epoch NIRCam imaging and a deep NIRSpec prism spectrum. The spectrum shows a strong, spectrally resolved H$\alpha$ emission line (FWHM $\approx 7900$ km/s) with a P-Cygni absorption blueshifted by 8800 km/s, together with corresponding H$\beta$ and He I $\lambda$10830 features, establishing a hydrogen-rich photosphere at $z=3.34$. Template fits classify SN Helios as a Type II SN, and light-curve comparisons with well-studied local SNe II yield a phase of $27.5\pm2.8$ rest-frame days past B-band maximum. Independent cluster lens models magnify the observed image by $\mu\approx 10$-20 and predict another image within $\simeq 2$-6 yr, offering a rare opportunity for a time-delay measurement of $H_{0}$ with a source at $z>3$. The host
galaxy is undetected in deep rest-frame ultraviolet and optical imaging, in emission lines, and in ALMA 2 mm
dust continuum, leaving a heavily obscured counterpart unlikely. Its de-lensed $M_{\rm UV}\gtrsim -13$ places it far below blank-field limits and adds support to an elevated core-collapse rate per unit
star formation in ultra-faint
galaxies. Without the multi-epoch data, SN Helios would masquerade as a convincing $z\approx 16$ Lyman-break
galaxy, with
galaxy-template fits strongly disfavoring any low-
redshift solution even with multiple medium bands. Its spectrum thus provides an empirical SED template with which high-
redshift searches can vet similar dropout candidates.
Read the full abstract ▾
All authors Seiji Fujimoto, Conor Larison, Justin D. R. Pierel, Lukas J. Furtak, Masamune Oguri, Adi Zitrin, Jose M. Diego, Keren Sharon, David A. Coulter, Gabriel Brammer, Vasily Kokorev, Chelsea Nash, Pedram Abedi, Christa DeCoursey, Armin Rest, Andreas L. Faisst, Brenda L. Frye, Tiger Yu-Yang Hsiao, Kohei Inayoshi, Kotaro Kohno, Jorryt Matthee, Matteo Messa, Yoshiaki Ono, Fengwu Sun, Eros Vanzella, Rogier A. Windhorst, Abdurro'uf, Aadya Agrawal, Joseph F. V. Allingham, Yoshihisa Asada, Hakim Atek, Maruša Bradač, Larry D. Bradley, Mateusz Bronikowski, Dan Coe, Christopher J. Conselice, Pratika Dayal, Eiichi Egami, Michael Engesser, Gavin Scott Farley, Qinyue Fei, Ori D. Fox, Yoshinobu Fudamoto, Miriam Golubchik, Yuichi Harikane, Gourav Khullar, Tomokazu Kiyota, Paulo A. A. Lopes, Ray A. Lucas, Guillaume Mahler, Ashish Kumar Meena, Rohan P. Naidu, Gautham Narayan, Gaël Noirot, Pascal A. Oesch, Masami Ouchi, José María Palencia, Massimo Pascale, Luke Robbins, M. R. Siebert, Roberta Tripodi, Francesco Valentino, Darach Watson, Hayley Williams, Hiroto Yanagisawa, Anita Zanella