Earth and Planetary Astrophysics
A JWST Study of Stardust. II. Polycyclic Aromatic Hydrocarbons in Comet 81P/Wild 2 and Implications for Primitive Solar System Carbon
Nathan X. Roth, Stefanie N. Milam, Diane H. Wooden et al. (+11 more)
We report observations of comet 81P/Wild 2, target of the Stardust sample return mission, on UT 2023 March 20 and 24 at a heliocentric distance ($r_H$) of 1.85 au using the NIRSpec and MIRI integral field unit spectrographs on board the James Webb Space
Telescope (JWST). This study is the first compositional comparison between JWST remote-sensing spectroscopy of a
solar system object against terrestrial analysis of its returned samples. Recent work determined contributions of molecular emission from coma volatiles and thermal emission from the
nucleus and coma
dust grains to these spectra. Subtracting the molecular and thermal emission models produces a residual spectrum with a clear resonance near 3.37 $\mu$m and multiple others spanning the $6 - 12$ $\mu$m region. Analysis of these residuals provides strong evidence for the presence of polycyclic aromatic hydrocarbons (PAHs) in the coma of 81P/Wild 2. The resulting PAH populations reflect material incorporated into the
nucleus of 81P/Wild 2 that was inherited from the diffuse interstellar medium, the interstellar medium, and the protoplanetary disk. The PAHs sensed by JWST are compared to those detected in laboratory analyses of the Stardust mission returned samples.
Read the full abstract ▾
All authors Nathan X. Roth, Stefanie N. Milam, Diane H. Wooden, Els Peeters, Charles E. Woodward, Dominique Bockelee-Morvan, Michael S. P. Kelley, Steven B. Charnley, Simon J. Clemett, Martin A. Cordiner, Perry A. Gerakines, David E. Harker, Ella Sciamma-O'Brien, Geronimo L. Villanueva
Statistical Methodology
Forecast-Necessary Causal Discovery for Nonlinear Political Panel Data: Feedback, Functional Form, and the Dynamics of Democratization
Michael Coppedge, Dmitry Zaytsev and Valentina Kuskova
A non-significant coefficient in a dynamic panel model need not imply the absence of a relationship. It may instead reflect heterogeneous effects averaged toward zero, reciprocal dynamics overlooked by a recursive specification, or relationships masked by the omission of correlated
covariates. Standard linear
estimators cannot distinguish among these possibilities. We develop an inferential workflow for political panel data that resolves this ambiguity by combining flexible autoregressive estimation, forecast-necessity testing, functional characterization, and same-data linear benchmarking. The workflow first identifies relationships required for out-of-sample prediction, then characterizes their functional form across political contexts, and finally, distinguishes differences arising from
estimator flexibility from those due to model specification. Applied to the causal sequence model of democratization on the V-Dem panel of 113 countries, the workflow reproduces the model's central finding - the protective belt of civil society, the rule of law, and institutionalized parties - while recovering reciprocal relationships from democracy to its institutional supports that a linear model cannot detect. Most importantly, three weak published direct effects, of which two are null, and one is marginally significant, receive three different diagnoses: one dissolves under the full specification, one reflects heterogeneous effects averaged toward zero, and one was masked by the reduced variable set. The workflow corrects the published record in both directions, removing one relationship and recovering two. More broadly, the workflow provides a framework for evaluating dynamic political theories under a model class capable of representing nonlinear and reciprocal mechanisms while preserving relationship-level interpretation and explicit inferential standards.
Read the full abstract ▾
Machine Learning
Evolution of fairness in multi-objective reinforcement learning framework
Jingyi Zhang, Xin Ou, Guozhong Zheng et al. (+3 more)
Fairness, as a fundamental social norm, continues to pose a longstanding puzzle regarding its emergence. Traditional game-theoretic models largely rely on the assumption of Homo economicus, wherein individuals are purely rational and self-interested, acting solely to maximize material payoffs. Such accounts, however, overlook the multidimensional nature of human decision-making, which is often shaped also by other considerations beyond economic incentives. To address this gap, we propose a multi-objective reinforcement learning framework that models the
evolution of
fairness as a dynamic trade-off between material payoff maximization and
fairness-driven moral behavior, regulated by a
fairness pressure coefficient. Using simulations of a two-objective Q-learning ultimatum game, we find that increased
fairness pressure promotes fair outcomes, as expected. Strikingly, however, under moderate pressure, responder behavior reverses: responders become “forgiving" by accepting low offers – a pattern in line with our daily experience. Microscopic analyses reveal that this strategy reversal stems from competition between payoff-maximizing and
fairness-oriented preferences. We further extend our framework to an asymmetric setting, where proposers and responders assign different weights to the two objectives. Overall, our work expands the reinforcement learning paradigm from a single-objective to a multi-objective formulation, offering a versatile tool for elucidating a broader range of human social behaviors.
Read the full abstract ▾
All authors Jingyi Zhang, Xin Ou, Guozhong Zheng, Shengfeng Deng, Jiqiang Zhang, Li Chen
Artificial Intelligence
GUITAR: Structured Failure Diagnosis of GUI Agents via State Transitions
Shaoqing Zhang, Kehai Chen, Xuefeng Bai et al. (+4 more)
Understanding where and why Graphical User Interface (GUI)
agents fail is essential for building more reliable systems, yet current evaluation relies on step accuracy, a metric that treats each screen independently and overlooks the underlying structure of GUI environments. This leads to two critical blind spots: (1) functionally equivalent screens are evaluated in isolation, obscuring systematic failure patterns across shared screens; and (2) the long-tailed GUI distribution renders failures on rare but critical screens invisible under standard metrics. To address these issues, we propose GUITAR, a state-centric diagnostic framework that performs structured failure analysis over both states and transitions, using a State Transition Graph (STG) by mapping visually diverse screens to shared functional states. Across 8
agents and 6 tasks from AndroidControl and Mind2Web, GUITAR reveals that 60.4% of failures occur in 20% of states, localizing errors to a small set of bottlenecks. Bottleneck-targeted guidance improves SR by 2.8% and retains a 1.88% average gain across 7
agents under three-fold trajectory-held-out evaluation with fully automatic STGs. These findings demonstrate the diagnostic and actionable value of structure-aware evaluation within the evaluated mobile and web tasks. Code is available at https://github.com/sqzhang-lazy/GUITAR
Read the full abstract ▾
All authors Shaoqing Zhang, Kehai Chen, Xuefeng Bai, Zhuosheng Zhang, Pengfei Zhang, Yang Xiang, Min Zhang
Computation and Language
Unknown is not normal: separating language-model extraction from rule-based decision logic for clinical risk scores
Nicolás Vera Zúñiga
Large
language models (LLMs) are increasingly used to compute clinical risk scores from free-text notes. Notes are often incomplete, and treating undocumented findings as normal can silently misclassify patients. We test whether separating three-state extraction (present, absent or unknown, by an LLM) from decision logic (deterministic code computing score bounds over unknown inputs) lets a system ask only questions that can change the decision. On 1,200 synthetic emergency cases across six calculators (HEART, CURB-65, qSOFA, PERC, Wells, Cockcroft-Gault), with a simulated clinician answering questions, we compared this bounds policy with asking for every missing input, a missing-equals-normal schema, and an end-to-end LLM
agent (Claude Opus 5.5). With Claude Haiku 4.5 as extractor, the bounds policy matched ask-all accuracy (99.4% vs 99.4%) with half the questions (0.92 vs 1.78 per case) and no irrelevant ones. Treating missing as normal dropped accuracy to 91.2% and under-triaged 8.5% of patients (95% CI 7.1-10.2), and under-triage persisted under messy notes and a noisy clinician. The
agent was equally accurate under ideal conditions (99.6%) but 9.5% of its questions were irrelevant; with a noisy clinician it was less accurate than the bounds policy (83.5% vs 87.0%, p<0.001) and committed prematurely in 2.7% of cases (bounds: 0%). A 9B local model as extractor reached oracle-level accuracy (99.8%). In 584 real case reports from MedCalc-Bench, only 52% contained enough information to determine the category (HEART 13%). Routing decisions through code that reasons explicitly about unknowns avoids premature commitment and irrelevant questions, halves the questions asked, and works with small local models.
Read the full abstract ▾
Artificial Intelligence
SpecRegMatch: Robust Semi-Supervised Regression for Vehicle Interior Noise Prediction
Sejin Sim, Jinsoo Bae and Seoung Bum Kim
The rapid advancement of artificial intelligence has observed increased application in predicting vehicle interior noise levels within the automotive industry. However, the collection of labeled data for training models in this context involves significant costs. Previous studies in semi-supervised regression (SSR) have effectively mitigated the reliance on labeled data by incorporating unlabeled data. Nonetheless, these approaches often introduce a high computational cost due to the training of multiple models and data sampling. This study introduces SpecRegMatch, a novel SSR method aimed at addressing the computational cost associated with training by leveraging a single model, thus eliminating the need for multiple data samplings. SpecRegMatch integrates consistency regularization and information maximization to robustly train the model, achieved through various augmentations applied to both the
embedding vectors and predicted values. Experimental results demonstrate that SpecRegMatch achieves state-of-the-art performance across various scenarios, even when using a single model. It attains a remarkable performance, as indicated by an R^2 score of 0.434. This is especially noteworthy in scenarios where labeled data is scarce. You can access the code for our proposed method at https://github.com/sejin-sim/SpecRegMatch.
Read the full abstract ▾
Functional Analysis
Endpoint-Safe Direct Characterisations of Weighted Bilinear Hardy Inequalities on Arbitrary Intervals
Saikat Kanjilal
We study the weighted bilinear Hardy inequality for the same-direction product $H_I f H_I g$ on an arbitrary real interval $I$, with $0<q<\infty$, $1\le p_1,p_2\le\infty$, and measurable weights allowed to take the values $0$ and $+\infty$. The characterisation is formulated through directed compact restrictions $J\Subset I$. On each compact interval, endpoint-safe input profiles retain the $p_i=1$ boundary contribution as a Stieltjes atom, while the lower-triangle reduction separates a closed Hardy part from a strict Copson part so that common atoms are counted exactly once. Upper and mixed regimes follow from exact freezing and compact linear Hardy estimates. The lower regimes are obtained by power lifting and a measure-valued Hardy-Copson
theorem. After simultaneous regularisation of the weights, the compact optimal constant is equivalent, with exponent-only constants, to an operational local characteristic, and the global constant satisfies $C_I\asymp\sup_{J\Subset I} A_J^{\mathrm{op}}$. In the regular interior range, the resulting conditions recover the classical $A_1,\ldots,A_7$ characterisations.
Read the full abstract ▾
Systems and Control
Integrated Thermal and Power Management for Wave-Powered Subsea Data Centers via Nonlinear Model Predictive Control
Wanqun Yang and Jun Chen
This paper develops an integrated modeling and nonlinear model predictive control (NMPC) framework for coordinating thermal management, flexible workload scheduling, wave-power utilization, and battery operation in a wave-powered subsea data center. Realistic data center workloads are constructed from job-level CPU,
memory, and GPU measurements from the MIT Supercloud
dataset and divided into interactive and delay-tolerant flexible jobs. Thermal behavior is represented by a three-node lumped model of the IT equipment, recirculating nitrogen, and pressure hull with surrounding seawater as the thermal boundary. The NMPC jointly optimizes the flexible workload power budget and cooling command subject to thermal, battery, and workload constraints. Closed-loop simulations under different workload, thermal, battery, and renewable-generation conditions demonstrate that the proposed framework maintains thermal safety while adapting cooling operation and flexible workload execution to wave-power availability and battery state-of-charge. The parametric studies show that battery capacity and wave-generation capacity strongly affect battery availability and flexible-workload queue accumulation, while excessive renewable generation capacity may lead to increased energy curtailment. Monte Carlo and distance-correlation analyses further show that flexible-job delay is relatively insensitive to the investigated system parameters, whereas terminal battery state-of-charge is primarily influenced by battery energy capacity and wave generation capacity.
Read the full abstract ▾
Geometric Topology
Local Combinatorial Criteria for Geometric Hyper-ideal Triangulations via Combinatorial Ricci Flow
Soichiro Uemura
We give local combinatorial criteria for realizing prescribed ideal triangulations by nondegenerate
hyperbolic truncated tetrahedra. A local bichromatic criterion allows valence-7 edges when tetrahedra meeting low-valence edges use at most two quotient-edge classes. Using the extended combinatorial
Ricci flow, we obtain parameter-dependent criteria for triangulations of minimum valence 6. Symmetric and pair-min angle estimates yield explicit valence conditions, including mixed conditions on entire edge
stars. The proofs use analytic angle inequalities and rigorous interval evaluations. Explicit face pairings realize the criteria and distinguish their scope, while cyclic constructions and finite covers give infinite families. Each criterion yields a unique zero-
curvature hyper-ideal metric and exponential convergence from any positive initial length vector.
Read the full abstract ▾
Earth and Planetary Astrophysics
Dynamical Impacts of Accretion Streamers on Protoplanetary Disks
Han-Gyeol Yun, Jeong-Eun Lee, Jaehan Bae et al. (+1 more)
Recent observations suggest that
accretion streamers are common in protoplanetary disks, yet their dynamical impact on disk
evolution remains poorly understood. Using three-dimensional hydrodynamic simulations with mass infall rates of $10^{-8}$–$10^{-6} \ M_{\odot}\ \mathrm{yr}^{-1}$, we investigate how streamer
accretion influences the structure and
evolution of protoplanetary disks. We find that a prograde streamer can excite disk eccentricity globally to values as high as $\sim0.4$. The resulting eccentric disk develops prominent spiral arms and crescent-shaped overdensities whose spatial structures agree remarkably well with analytic eccentric-disk theory. It also undergoes significant warping, and exhibits enhanced and variable
stellar accretion. In contrast, retrograde streamer
accretion efficiently removes disk angular momentum, producing a compact disk and, in extreme cases, triggering disk breaking. Using synthetic ALMA molecular-line observations, we show that streamer-driven perturbations generate observable kinematic signatures, including Doppler flips in moment maps and wiggles in position–velocity diagrams. Remarkably, these signatures can persist for up to $100\ \mathrm{kyr}$ after infall has ceased, suggesting that some kinematic disturbances observed in disks without currently detected streamers may be relics of past infall events. Finally, we discuss the implications of streamer-driven disk
evolution for
planet formation and
planet–disk interactions.
Read the full abstract ▾
All authors Han-Gyeol Yun, Jeong-Eun Lee, Jaehan Bae, Woong-Tae Kim
Computer Vision
The Devil is in the Spectrum Bias: Spectrum-Balanced Feature Matching for Robust Representation Distillation
Kuniaki Saito and Yoshitaka Ushiku
Large visual foundation models have demonstrated remarkable transferability across a wide range of downstream tasks. To deploy such models efficiently, feature matching has become a popular knowledge
distillation approach that transfers teacher representations to smaller student models without requiring labeled data. However, we show that the conventional feature matching objective with L2-distance is inherently biased toward reconstructing dominant spectral directions of the teacher representation, while under-optimizing low-variance directions that often contain task-relevant information. To address this, we propose Spectrum-Balanced Feature Matching, SpecMatch, a simple objective that adaptively emphasizes under-optimized spectral directions while preserving the relative importance of dominant directions. SpecMatch is easy to implement and introduces negligible computational overhead. Extensive experiments on image recognition demonstrate that SpecMatch consistently improves downstream adaptation across diverse tasks, including image classification, anomaly detection, medical image analysis, and domain generalization. In particular, SpecMatch outperforms conventional feature matching in 40 of 42 teacher–student and training-setting combinations, while consistently improving over the original student model in all settings. We further demonstrate that the proposed objective generalizes beyond
vision, improving downstream performance across six
protein understanding tasks.
Read the full abstract ▾
Optics
Stimulated Electro-optic Scattering
Violet Workman and Gaurav Bahl
Stimulated Brillouin Scattering (SBS) couples light to acoustic waves and underpins applications ranging from sensing and signal processing to
quantum photonics. In solids, this interaction is generally attributed to photoelasticity and the motion of dielectric boundaries. Here we show that piezoelectricity opens an additional, previously unrecognized pathway that can substantially reshape Brillouin gain. This contribution arises from the combined action of linear electro-optic “Pockels” scattering and its reciprocal $\chi^{(2)}$-mediated optical drive, which couple optical fields to the electric fields that accompany mechanical motion in piezoelectrics. We formally incorporate these effects into the modern theory of SBS and apply it to monolithic lithium niobate waveguides. In addition to large enhancement and suppression of the Brillouin gain, we surprisingly find occurrences of large gain where conventional theory would otherwise have predicted zero Brillouin interaction. These instances are the first identification of purely electro-optic SBS, a process we term stimulated electro-optic scattering, as it only involves interactions between electromagnetic fields even though a mechanical excitation is present. Our results establish the electro-optic tensor as a key governing parameter for Brillouin interactions in piezoelectric media and provide a new degree of freedom for engineering
photon-
phonon coupling independent of photoelasticity and index contrast.
Read the full abstract ▾
Mesoscale and Nanoscale Physics
(Sub)nanoscale Visualization of Reconstruction-Driven Moiré Exciton Localization and Delocalization
Medha Dandu, Sriram Sankar, Giovanny Espitia et al. (+12 more)
Spectral fingerprints in optical absorption and emission have typically been used as a signature of
exciton localization in twisted moiré
bilayers. However, the mechanism by which
excitons become confined to specific stacking sites and their associated optical signature is experimentally unresolved. Here, we directly visualize in real space how tuning the change in extent of structural reconstruction leads to localization and delocalization of moiré
excitons in the WSe2/WS2 moiré superlattice. Using cryogenic monochromated electron energy loss spectroscopy, together with first-principles GW-Bethe Salpeter equation calculations and optical spectroscopy, we uncover the physical mechanism that drives the correlation between twist-angle-dependent structural transformations, the real-space localization of moiré
excitons, and their optical signatures. Surprisingly, and in contrast to the prevailing understanding, we show that the emergence of new moiré
exciton resonances in the optical spectra alone is insufficient to establish
exciton localization. Instead, the extent of structural reconstruction and external strain drives
exciton localization, leading to new design principles for engineering moiré
excitons and strain-aware
quantum optoelectronic devices.
Read the full abstract ▾
All authors Medha Dandu, Sriram Sankar, Giovanny Espitia, Rwik Dutta, Patrick J. Hays, Daria Blach, Takashi Taniguchi, Kenji Watanabe, James R. Chelikowsky, Seth Ariel Tongay, Peter Ercius, Jordan A. Hachtel, Mit H. Naik, Archana Raja, Sandhya Susarla
Artificial Intelligence
A Differentiable Optimization Framework for Registering Sequential Bounding Boxes with Point Cloud Stream
Xuesong Li, Jinguang Tong and Jie Hong
Refining a sequence of coarse 3D bounding boxes against a
LiDAR point-cloud stream demands tracks that are geometrically accurate (high IoU) and temporally coherent (low roughness), preferably without training data. The usual recipe keeps the two concerns apart: register each frame independently, then smooth the trajectory afterwards with a Kalman~RTS or Savitzky–Golay filter. Smoothing displaces boxes from a
geometric optimum and never re-optimises, so it trades accuracy for smoothness. We instead fold the temporal smoothness constraint into a training-free registration objective and solve for all poses jointly with L-BFGS. The payoff depends on how well the object is seen. On well-observed tracks it is large: within the low-roughness budget, the joint objective beats both post-hoc smoothers on paired multi-seed statistics and cuts roughness several-fold relative to frame-wise registration at matched accuracy. Treating visibility as an experimental variable exposes the limit. The advantage decays monotonically as views become one-sided, until it is indistinguishable from zero for near-edge-on objects and slightly negative under a ray-cast simulator with range-dependent density and ego motion, where the decoupled pipeline is in fact ahead at tight roughness budgets. We locate that boundary and trace it to one term: orientation
alignment ties yaw to the estimated velocity and fails once that estimate is noisy. A ground-truth-free rule can choose the temporal scale and keep every track inside the roughness budget.
Read the full abstract ▾
High Energy Astrophysical Phenomena
Discovery of a 62-min long-period transient with near-orthogonal rotator geometry
Scott D. Hyman, Natasha Hurley-Walker, Dale A. Frail et al. (+2 more)
We report the serendipitous discovery of J155543.7$-$563102 (hereafter LPT J1555$-$5631) with a period of 62.2 min in MeerKAT and ASKAP survey data. The source emitted bright (10-30 mJy), steep-spectrum ($\alpha\sim-1.5$), highly polarized
radio pulses during a 15-hr active window in May 2022, with no other quiescent or pulsed emission detected before or after over $\sim5$ years. The mean light curve exhibits two alternating pulses per cycle. A main pulse (MP) and an interpulse (IP) separated by $\Delta\phi=165.4^\circ$ with similar amplitudes, but different polarization properties: the wider MP (3.9%) is partially linearly and circularly polarized and shows a constant polarization position angle, while the narrow IP (1.1%) is $\sim$100% circularly polarized. We classify LPT J1555$-$5631 as a near-orthogonal rotator, for which emission from both
magnetic poles is visible during one rotation and the
magnetic and rotation axes are almost perpendicular. The 14.6$^\circ$ deviation from an antipodal separation and the nearly constant polarization position angle suggest that the emission does not arise from the standard dipolar polar-cap geometry of ordinary
pulsars, but instead originates within a broader, more complex magnetosphere. The observed properties of LPT J1555$-$5631 can be accommodated by extended emission regions, and favor coherent emission mechanisms such as the electron cyclotron maser instability, although they do not uniquely determine the progenitor or emission physics. If these inferences apply more broadly to the LPT population, they could account for why their beam widths are decoupled from rotation periods, the prevalence of flat polarization position angles, and may favor an elevated incidence of orthogonal rotators.
Read the full abstract ▾
All authors Scott D. Hyman, Natasha Hurley-Walker, Dale A. Frail, Emil Polisensky, W. D. Cotton
Optimization and Control
On the Optimal Linear Minimization Oracle Complexity of Active-Hull Methods for Minimizing Convex Hölder Smooth Functions
Renbo Zhao
We consider the convex problem $(P): {\min}_{y\in \mathbb{R}^n}\, f(Ay)+g(y)$, where $A:\mathbb{R}^n\to \mathbb{R}^m$ is a linear operator, $f$ is convex and $(M,\nu)$-Hölder smooth on $\mathbb{R}^m$, and $g$ is proper, closed convex on $\mathbb{R}^n$, and admits a (generalized) linear minimization oracle (LMO). An active-hull (AH) method calls the LMO in each iteration and outputs a point that lies in the convex hull of all the past LMO returns. We show that under the Euclidean geometry (and in the high-dimensional regime), the optimal LMO complexity for the class of AH methods to solve $(P)$ is $O\big(M^{{2}/({1+\nu})}D^2\varepsilon^{-{2}/({1+\nu})}\big)$, where $D$ denotes the diameter of the feasible region and $\varepsilon$ denotes the objective sub-optimality gap. In particular, the optimal LMO complexity is achieved by a single-loop first-order primal-dual splitting method. Since this method requires knowledge of the problem parameters (e.g., $M$, $D$ and $\nu$), we propose two meta parameter-search
algorithms that aim to resolve this issue. These
algorithms are parameter-free, but come with the price of an additional log-factor in their LMO complexities.
Read the full abstract ▾
Populations and Evolution
Migration Genealogies in Multiregional Stable Populations: First-Return Decompositions, Recurrence, and Sensitivity
Ryo Oizumi, Kensaku Kinjo and Yuki Chino
When people of reproductive age move between regions, replacement-level fertility need not produce spatially balanced population replacement. Migration genealogies connect a woman's birth region, through survival and movement, to her daughters' birth regions across generations. We decompose existing genealogical series for stable births and reproductive value into first-return paths, on which a lineage again produces a descendant in its reference region without intermediate births there. Discounting at the intrinsic growth rate yields a first-return distribution with an Euler–Lotka-type normalization. Its mean recurrence generation, the expected generations until first return, equals the normalized product of stable births and reproductive value and, at replacement, reciprocal net reproduction sensitivity to within-region reproduction. Using Japan's 2010, 2015, and 2020 censuses, vital statistics, and life tables, we compare 47 prefectures under national replacement. Based on 2020 rates, mean recurrence is approximately 10 generations for Tokyo and 500 for Akita; Fukushima varies across observation years. Similar one-generation birth contributions conceal different descendant destinations and subsequent return pathways. These period-based results reveal how reproductive pathways generate regional differences in stationary populations. Population replacement must therefore be analyzed jointly through fertility, migration, and regional conditions, with mean recurrence identifying the strength of closed reproductive pathways.
Read the full abstract ▾
Optimization and Control
Constrained Nonconvex Stochastic Optimization with One Projection
Yuyang Deng, Mohammadreza M. Kalan, Eitan J. Neugut et al. (+1 more)
Constrained nonconvex optimization has seen increasing application in modern machine learning, such as safe LLM
alignment/finetuning, transfer learning and rank-constrained continual learning. The commonly used approach is Projected SGD which requires projection at every iteration. However, projection onto a functional constraint can cost substantially more than a
stochastic first-order update. We study whether this operation can be deferred until the end of nonconvex
stochastic optimization, and only do it once. For weakly convex, possibly nonsmooth objectives with regular convex constraints, we give a penalized proximal method that uses $\widetilde O(\epsilon^{-4})$
stochastic subgradients and one projection onto the constraint set. The returned point is exactly feasible and has a small Moreau-envelope stationarity measure; for smooth objectives, the same
algorithm controls the projected-gradient mapping. For smooth nonconvex constraints, we assume a global lower bound on the infeasible constraint slope which permits arbitrary, possibly infeasible initialization. An exact-penalty variant attains the same
stochastic-oracle order and returns an exactly feasible point within $O(\epsilon)$ of an $\epsilon$-KKT point. All intermediate updates use only projections onto a simple Euclidean ball. These are oracle-complexity guarantees: the cost of the single terminal projection is separate, and no efficient projection
algorithm for a general nonconvex set is assumed to follow from regularity.
Read the full abstract ▾
All authors Yuyang Deng, Mohammadreza M. Kalan, Eitan J. Neugut, Mehrdad Mahdavi
Machine Learning
Beyond Correctness: Evaluating Semantic Knowledge in Cross-Table Transfer
Seokyong Sheem, Hochang Lee, Suyeong Lee et al. (+1 more)
Semantic knowledge is increasingly used to bridge heterogeneous schemas in tabular learning, but how much does that knowledge actually improve prediction? Studies in tabular learning commonly answer this question through semantic ablations that modify or suppress the supplied semantic knowledge. We show that these ablations can lead to misleading conclusions about predictive benefit: poor performance under altered semantics may be taken as evidence that the intended knowledge is beneficial. Across real and controlled experiments, altering semantic content can produce large performance differences even when the model gains little predictive benefit from having that semantic knowledge in the first place. To separate these effects, we distinguish two quantities: content sensitivity and predictive utility. Content sensitivity measures the change in performance when semantic content is altered, whereas predictive utility measures the benefit of the intended semantic knowledge relative to a suitable reference without that knowledge. This distinction motivates an evaluation framework in which the control is chosen according to the question being asked: altered controls assess sensitivity to semantic content, whereas claims that semantic knowledge improves prediction require a suitable reference. Even then, predictive utility is not fixed; it varies across suitable references and decreases when the reference can more easily recover the tested knowledge from other inputs or labeled examples. In a bounded audit of 25 semantic-ablation comparisons across nine studies, only one of 18 explicit predictive-utility claims is paired with a control that clearly isolates the tested semantic contribution. Together, these findings motivate a simple evaluation principle: semantic-ablation controls should be chosen and interpreted according to the question they are intended to answer.
Read the full abstract ▾
All authors Seokyong Sheem, Hochang Lee, Suyeong Lee, Daekyum Kim
Fluid Dynamics
On the Rayleigh-Taylor instability of a diffuse interface
Marildo Kola, Daniel Israel and Aaron Towne
We present a formulation for the linear stability of the miscible Rayleigh-Taylor instability, retaining viscosity, Fickian mass
diffusion, and non-Boussinesq effects. Rather than adopting the diffusive corrections of Duff, Harlow, and Hirt (1962) or relying on the viscous
eigenvalue problem of Chandrasekhar (1961), the present formulation is derived directly from the incompressible variable-density governing equations. Both approaches are recovered as limiting cases, extending Morgan, Likhachev, and Jacobs (2016). The governing equations admit a time-evolving self-similar diffusive base state in which an error-function density profile is accompanied by a nonzero vertical base velocity required by the non-solenoidal constraint. Linearization yields a reduced coupled formulation for the vertical velocity and density perturbations, suitable for both theoretical analysis and computation. Within the quasi-steady-state approximation, we show that the formulation reduces, in the infinite-Schmidt-number limit, to the classical viscous immiscible theory and recovers the Chandrasekhar
eigenvalue problem. At finite Schmidt number,
diffusion alters the growth-rate spectrum and its high-wavenumber structure. Beyond the spectral cutoff we establish the scaling $\omega_r \sim -\lambda_m k^2$, where $\lambda_m$ is determined by the competition between the locally stratified kinematic viscosity and the reference diffusive scale. By isolating the effect of
diffusion, we show that its stabilizing influence weakens with increasing density stratification. Buoyancy production becomes bimodal and shifts toward the lighter-fluid side, while pressure production grows to rival buoyancy and acts as an energy source on the heavier side of stratification. Together, these results clarify when classical diffusive corrections remain applicable and when
diffusion and non-Boussinesq coupling must be retained.
Read the full abstract ▾
Human-Computer Interaction
HiThink Turn: An Intent-Aware Turn-Taking Control Module for Full-Duplex Dialogue
Feiyang Chen, Wenhan Yang, Bohan Wang et al. (+4 more)
Full-duplex dialogue requires timely yet selective interruption handling, which end-of-turn prediction alone cannot achieve: complete utterances may need no response, while unfinished requests may warrant interruption. To address this challenge, we propose HiThink Turn, an intent-aware streaming turn-state predictor that separates response intent from semantic completeness and conditions decisions on system playback state. A key contribution is minimal intent-sufficient prefix supervision, constructed through LLM judgments and
speech alignment, while training on audio truncated at chunk boundaries improves robustness to partial
speech. These components support streaming inference with 240-ms audio chunks, enabling low-latency, accurate full-duplex turn control. Experiments show that HiThink Turn leads the compared methods in Easy Turn macro accuracy, Full-Duplex-Bench average interaction rate score (0.933), and non-target-
speech average playback resume rate (0.735). Additionally, intent-prefix triggering raises interruption success from 89% to 98% and reduces mean stop latency by 60.9%.
Read the full abstract ▾
All authors Feiyang Chen, Wenhan Yang, Bohan Wang, Xinjian Gao, Rongjunchen Zhang, Jun Wang, Xinhui Hu
Optics
Tensor-Engineered Van der Waals NbOCl2 Resonant Metasurface for Polarization-entangled Bell State Generation
Xin Zeng, Wenna Du, Yun-Kun Wu et al. (+20 more)
Polarization-entangled
photon pairs are essential resources for
quantum information technologies, yet realizing compact sources with intrinsically controllable
entanglement remains challenging. Van der Waals (vdW) nonlinear materials such as NbOCl2 provide atomically thin platforms for
quantum light generation, yet their native crystalline anisotropies as natural materials impose limitations on accessible
quantum states. Here, we develop a resonant vdW nonlinear metasurface based on NbOCl2 that exploits its intrinsic optical anisotropy to engineer polarization-dependent nonlinear responses. The anisotropic optical dispersion enables selective manipulation of resonant modes, resulting in a three-order-of-magnitude enhancement of the nonlinear response along the c axis. By further tailoring these resonant modes, we redistribute the effective second-order nonlinear susceptibility tensor between orthogonal polarization channels, balancing the spontaneous parametric down-conversion pathways along the b and c axes. This enables polarization-entangled
photon generation with a measured fidelity of up to 92%. Our work establishes metasurface-enabled nonlinear optical engineering as a strategy for enhancing and controlling
quantum light generation in vdW materials, providing a pathway toward scalable
quantum photonic platforms.
Read the full abstract ▾
All authors Xin Zeng, Wenna Du, Yun-Kun Wu, Yuyang Zhang, Zhiyong Zhang, Kwok Kwan Tang, Ruihuan Duan, Xiaotian Bao, Yiyang Gong, Yuexing Xia, Yutong Zhang, Jinhan Zhao, Yubo Tian, Yangguang Zhong, Chuanxiu Jiang, Yubin Wang, Jianhui Fu, Shuai Yue, Guankui Long, Qing Zhang, Zheng Liu, Xi-Feng Ren, Xinfeng Liu
Computer Vision
Advancing Wildlife Conservation through Multimodal Animal Re-Identification with Environmental Metadata
Yuzhuo Li, Di Zhao, Tingrui Qiao et al. (+3 more)
Identifying individual animals is crucial for effective wildlife monitoring and conservation efforts. Recent advancements in computer
vision have shown promise in animal re-identification (Animal ReID) by leveraging data from camera traps. However, existing Animal ReID
datasets rely exclusively on visual data, overlooking environmental metadata that ecologists have identified as highly correlated with animal behavior and identity, such as temperature and circadian rhythms. Meanwhile, modern
vision-
language models (VLMs) offer rich
multimodal reasoning capabilities, but existing resources underutilize their text-processing potential. To address these limitations, we propose MetaWild, a
multimodal Animal ReID
dataset comprising 20,890 images across six species, paired with environmental metadata extracted from embedded camera trap overlays and scene contexts. Additionally, to facilitate the use of metadata in existing ReID methods, we propose the Meta-Feature Adapter (MFA), a lightweight module that can be incorporated into existing VLM-based ReID methods, allowing ReID models to leverage both environmental metadata and visual information to improve ReID performance. Experiments on MetaWild show that combining baseline ReID models with MFA to incorporate metadata consistently improves performance compared to using visual information alone, validating the effectiveness of incorporating metadata in re-identification.
Read the full abstract ▾
All authors Yuzhuo Li, Di Zhao, Tingrui Qiao, Yihao Wu, Bo Pang, Yun Sing Koh
Number Theory
Density of $\mu$-ordinary Primes for K3 Surfaces
Trung Can and Yuxin Lin
By a result of Bogomolov and Zarhin, a K3 surface $X$ over a number field $L$ has ordinary reduction at a density 1 set of
primes after passing to a finite extension of $L$. In this paper, we refine this result for non-CM K3 surfaces whose transcendental Hodge structure has endomorphism field $F$ abelian over $\mathbb{Q}$. We further assume a condition on the connected components of the $l$-adic monodromy group of $X$, and, when $F$ is totally real, a parity condition on the rank of the transcendental
lattice over $F$. Under these assumptions, we prove that the set of
primes of $L$ at which $X$ has $\mu$-ordinary reduction has density $1$. As a corollary, the set of
primes of $L$ at which $X$ has ordinary reduction has density $1/[FL:L]$. We include explicit examples of K3 surfaces satisfying these conditions.
Read the full abstract ▾
Computation and Language
Evaluating Machine Unlearning in ASR
Diogo Dinis, Francisco Teixeira, Bhiksha Raj et al. (+2 more)
Machine unlearning (MU) offers a path to compliance with "right to be forgotten" regulations. While MU has received increasing
attention for
speech tasks, it remains largely unexplored for Automatic
Speech Recognition (ASR). In this work, we investigate whether existing MU
algorithms and evaluation tools are suitable for ASR. We apply several MU techniques to an ASR model, evaluating
privacy-utility trade-offs for single-subject unlearning, then assess the best
algorithm under sequential and simultaneous unlearning. Results show that gradient ascent-based
algorithms achieve strong utility-
privacy trade-offs, whereas more complex approaches over-unlearn samples, making them easier to identify as unlearned. This suggests standard
privacy evaluations based on simple Membership Inference attacks are insufficient to reliably assess unlearning success, motivating improved evaluation methods for MU in ASR. Finally, we show that both sequential and simultaneous unlearning yield worse
privacy and utility than single-subject unlearning, underscoring the need for unlearning constructions better suited to these settings.
Read the full abstract ▾
All authors Diogo Dinis, Francisco Teixeira, Bhiksha Raj, Alberto Abad, Isabel Trancoso
Machine Learning
From Perturbation Sensitivity to Sparse Autoencoder Feature Interpretation: An EEG Foundation-Model Case Study
Yucong Cao, Chenqi Li and Tingting Zhu
Sparse autoencoders (SAEs) split a model's internal activity into many sparse units, or latents, and a latent is often named after a concept when removing that concept from the input changes its activity. We identify a consequential confound in this inference: if removing the concept changes the input more than the comparison edit does, a larger response may reflect a larger edit rather than the concept.
Brain-signal (electroencephalography,
EEG) models provide a controlled test, because how much an edit changes the
EEG input can be measured directly. We measure the SAE response as the change in how often each latent is active, averaged over all latents. Across 27 combinations of model,
dataset, and layer, removing the 8–12 Hz alpha band changes this response 7.33 times as much as removing an equally wide 30–34 Hz control band, but it also changes the signal far more. Matching the two edits per trial on squared time-domain distortion lowers the mean ratio to 1.42, with 19 of 27 combinations above one. Matching thus substantially reduces, but does not eliminate, the alpha–control contrast. The residual supports relative sensitivity under this single control, not general spectral specificity or an individual alpha-representing latent. We use an evidence-to-claim map to separate what this comparison supports from stronger claims of specificity and feature semantics.
Read the full abstract ▾
Computational Geometry
Using Persistent Homology to Analyze Access to Heterogeneous-Quality Resources and Heterogeneous-Severity Nuisances
Sarah Tymochko, Gillian Grindstaff, Abigail Hickok et al. (+2 more)
We develop a framework to use multiparameter persistent homology (PH) to examine access to heterogeneous-quality resources and exposure to heterogeneous-severity nuisances in a geographic region. Persistent homology, which is a type of topological data analysis {(TDA)}, has been employed previously to examine resource coverage. Unlike prior approaches, which used one-parameter PH to study resource coverage and nuisance exposure, our method accounts for heterogeneous-quality resources. Our framework, which employs a computationally-efficient approximation of multiparameter PH, allows one to study access to any resource ({or} exposure of any nuisance) using any notion of quality (or severity). Using the city of Chicago as an example region, we employ our framework to detect clusters of poor access to public parks, overexposure to landfills, and both underexposure and overexposure to pubs and bars.
Read the full abstract ▾
All authors Sarah Tymochko, Gillian Grindstaff, Abigail Hickok, Jiajie Luo, Mason A. Porter
Programming Languages
Faultless: A Program Equivalence Technique for Validating and Evaluating Neural Decompilers
Luke Dramko, Claire Le Goues and Edward Schwartz
Neural decompilers are machine learning models which perform the process of decompilation, lifting code from a lower-level
language to a higher one. Neural decompilers offer substantial utility relative to traditional deterministic decompilers because they can probabilistically recover information discarded during lowering, like variable names, types, and control flow structuring. However, they can also make mistakes, producing code that is not equivalent to the original, making it difficult to trust their output. In this work, we introduce Faultless, a program equivalence technique for performing translation validation on neural decompilers. Faultless compares code produced by a deterministic decompiler, which has stronger correctness properties, with that of a neural decompiler. Faultless is also useful for model evaluation, a highly related task, in which the neural decompilers' prediction is compared with a reference solution. Neural decompilation introduces significant challenges to the task of program equivalence which existing techniques are not equipped to handle, including limited extrafunctional context and systematic semantic inconsistencies in decompiled code. Faultless takes a static symbolic execution-based approach with an execution model and
memory model designed to handle these challenges.
Read the full abstract ▾
Machine Learning
TRACE: Expert-Aligned ECG Representation Learning with Rigorous Benchmarking and Real-World Validation in Acute Cardiac Care
Lovely Yeswanth Panchumarthi, Andrew Lu, Saurabh Kataria et al. (+11 more)
TRACE (Text-Reinforced Analysis of Cardio ECGs) is a
multimodal electrocardiogram (ECG) representation model that learns clinically grounded signal
embeddings for downstream cardiac classification. It is designed to address the limitations of existing CLIP-style training, which often struggles with noisy clinical text and fails to leverage the complementary strengths of unimodal (from ECG) and cross-modal (between ECG and matched cardiologist reports) learning. To bridge this gap, we propose a hybrid architecture that jointly learns unimodal and cross-modal representations via uncertainty-weighted multi-task learning while utilizing an LLM-based pipeline to extract high-fidelity findings from cardiologist reports. We evaluate TRACE across a spectrum of clinical urgency, establishing robust performance on public
benchmarks for arrhythmia classification and structural abnormalities relative to existing unimodal and
multimodal ECG models. To demonstrate real-world utility, we further validate the model on acute coronary occlusion (ACO), where the prevailing ST-elevation criteria miss 25-34% of true occlusions. Utilizing a large private ACO
dataset with expert-annotated ground truth, TRACE significantly outperforms real-world clinical practice, yielding a 19.0% increase in sensitivity or a 62.6% reduction in false positive rates at the clinical baseline. This extensive evaluation confirms that TRACE delivers both strong performance on
benchmark tasks and tangible clinical impact in the most acute, high-risk cardiac scenarios.
Read the full abstract ▾
All authors Lovely Yeswanth Panchumarthi, Andrew Lu, Saurabh Kataria, Delgersuren Bold, Minxiao Wang, Runze Yan, Patricia Dykes, Brian J. Gow, Tom J. Pollard, Jessica K. Zègre-Hemsey, Dillon J. Dzikowicz, Lekshmi Kumar, Xiao Hu, Ran Xiao
Materials Science
Nanometer-scale operando electric field mapping in oxide junctions by correlative STEM-EBIC and 4D-STEM
Yining Xie, Eoin Moynihan, Xingyao Li et al. (+5 more)
Understanding local electric fields at
oxide interfaces is essential for linking interfacial electrostatics to device functionality, yet conventional electrical measurements infer these fields only indirectly. Here, we combine operando scanning transmission electron microscope electron-beam-induced current (STEM-EBIC) with four-dimensional-STEM (4D-STEM) to quantitatively reconstruct the bias-dependent electric field spatial distribution across an
oxide Schottky junction at nanometer scale. Using La0.67Sr0.33MnO3/Nb:SrTiO3 (LSMO/NSTO) as a model system, we fabricate an electron-transparent junction on a MEMS biasing platform and verify that it retains the rectifying transport behavior. STEM-EBIC provides complementary information addressing several key limitations in quantitative 4D-STEM field mapping for heterojunctions. The reconstructed electric field profiles show a pronounced nonlinear decay within the depletion region and an extended penetration into the LSMO. These results directly reveal the deviations of
oxides Schottky junction from the ideal conventional Schottky depletion model, providing experimental signatures of non-classical interfacial electrostatics. Our correlative approach enables quantitative nanoscale electric field mapping in operating
oxide heterojunctions, providing a basis for linking interfacial electrostatics to macroscopic transport and guiding
oxide devices design.
Read the full abstract ▾
All authors Yining Xie, Eoin Moynihan, Xingyao Li, Richard Beanland, Aditya Singh, Marin Alexe, Mingmin Yang, Ana Sanchez