Cosmology and Nongalactic Astrophysics
Lensing in the Blue IV: The First Weak Gravitational Lensing Maps from the Stratosphere
Sayan Saha, Maya Amit, Jacqueline E. McCleary et al. (+42 more)
The Super-Pressure Balloon-borne Imaging
Telescope (SuperBIT) completed its science flight in April 2023, obtaining imaging data for 30
galaxy clusters. In this work, we present the first weak-
lensing convergence maps from these observations, using six clusters selected from the 30 targets. We construct background-
galaxy catalogs using SuperBIT's three-band photometry and a color-color-based method to separate foreground and background
galaxies, and use the calibrated
galaxy shapes presented in
Lensing in the Blue III to reconstruct the weak-
lensing convergence field using Kaiser-Squires inversion. We compare the convergence signal-to-noise maps for the six clusters with independent tracers, including X-ray emission and red-sequence cluster-member number-density maps. These results demonstrate the first successful weak-
lensing mass mapping with imaging from a sub-orbital astronomical observatory.
Read the full abstract ▾
All authors Sayan Saha, Maya Amit, Jacqueline E. McCleary, Georgios N. Vassilakis, Emaad Paracha, Giulia Cerini, Spencer W. Everett, Leila Ohashi, Eric M. Huff, Gavin Leroy, Leo W.H. Fung, Richard Massey, Steven J. Benton, William C. Jones, Thuy Vy T. Luu, Ajay S. Gill, Mohamed M. Shaaban, Philippe Voyer, Adam Field, Adam Aoua, Anthony M. Brown, Paul Clark, Matthew Craigie, Christopher J. Damaren, Tim Eifler, David Harvey, Eric Habjan, John W. Hartley, Bradley Holder, Mathilde Jauzac, David Lagattuta, Jason S.-Y. Leung, Lun Li, Johanna M. Nagy, C. Barth Netterfield, Susan F. Redmond, Jason D. Rhodes, Andrew Robertson, L. Javier Romualdez, Jürgen Schmoll, Ellen Sirks, Sut Ieng Tam, Karuna Tarafdar, André Z. Vitorelli, Alfredo Zenteno
Computer Vision
Hybrid Cross-Modal Attention Network for Early Breast Cancer Detection in Low-Resource Clinical Settings
Simon Hadush Nrea, Filimon Gidey Gebremichael, Gebrekirstos Hagos Gebrekirstos et al. (+1 more)
Breast cancer is the leading cause of cancer-related mortality among women in Sub-Saharan Africa, where delayed diagnosis results from limited radiology expertise and fragmented clinical data systems. Although deep learning models have demonstrated strong performance in mammographic analysis, most rely solely on imaging data and are trained on Western populations, limiting their applicability in African healthcare settings. This paper presents a Hybrid Cross-Modal
Attention Network (HCMAN) that integrates mammogram images with structured clinical data using
transformer-based cross-modal
attention mechanisms. The model was developed and validated using a locally collected
dataset of 2,560 mammogram images from 1,024 patients across four Ethiopian referral hospitals, with biopsy-confirmed ground truth labels. The proposed framework achieves 97.8% accuracy, 97.2% sensitivity, 98.3% specificity, and an AUC of 0.987, significantly outperforming image-only baselines. The system demonstrates robustness to low-quality images typical of resource-limited settings, with only 3.2% performance degradation compared to 8.7% for image-only models. Cross-modal
attention analysis reveals clinically appropriate behavior: higher reliance on clinical features for ambiguous cases such as dense breasts and young patients. The model's lightweight architecture enables deployment on standard hospital workstations (<2 seconds inference on CPU). This work advances sustainable, context-aware AI solutions for equitable breast cancer diagnostics in Africa.
Read the full abstract ▾
All authors Simon Hadush Nrea, Filimon Gidey Gebremichael, Gebrekirstos Hagos Gebrekirstos, Yaecob Girmay Gezahegn
Quantum Physics
Quantum Machine Learning for Few-Shot Impersonation Detection in Digital Account Opening
Javiera Canales, Jose Ortega and Maximiliano Araya
Identity impersonation fraud may emerge only after digital account opening, requiring an early post-onboarding approach based on behavioral and transactional signals observed shortly after activation. Confirmed cases remain rare, creating extreme class imbalance and material operational risk for digital banks. We evaluate
quantum machine learning (QML) against classical one-class methods in this setting using a real 2025 cohort of 3,419 digital account openings from Banco de Creditos e Inversiones (BCI), Chile, including 8 confirmed impersonation-fraud cases (0.23% prevalence). The data were provided through an industry-research collaboration between BCI and CoreDevX LABTAV under operational constraints. The best-performing configuration used only 2
qubits and was executed on CoreDevX's SpinQ Triangulum II, a 3-
qubit nuclear magnetic resonance (NMR) device, enabling hardware validation beyond classical simulation. The
quantum one-class classifier based on Automatic
Quantum Feature Mapping (AQFM) achieves 88% recall (7 of 8 frauds detected) and 9.21% precision, corresponding to approximately 70% fewer false positives than the best-performing classical baseline at comparable recall. When its 2025-calibrated decision threshold is applied unchanged to the independent 2026 cohort, AQFM detects 20 of 31 frauds (64.5% recall) with 8.81% precision and remains the best-performing model among those evaluated.
Read the full abstract ▾
High Energy Astrophysical Phenomena
Radial Modes of Hot Neutrino-Trapped Hybrid Stars
Peter B. Rau, Andrea Sabatucci and Armen Sedrakian
We study the fundamental radial oscillation mode of hot hadronic and hybrid
stars under conditions relevant to proto-
neutron stars and
neutron-
star merger remnants. The
stellar backgrounds are constructed from finite-temperature equations of state with a covariant density functional for hadronic matter and three-
flavor 2SC
quark matter at high density. We consider two isentropic configurations in both
neutrino-transparent matter and with trapped electron
neutrinos at fixed electron-
lepton fraction. For trapped matter we compare a Gibbs-type construction with global electron-
lepton-number conservation to a locally constrained Maxwell-type prescription; in both cases the fixed-
entropy stellar equation of state contains an extended coexistence region and no sharp interface requiring conversion rate-dependent junction conditions. We find that the fundamental radial frequency follows the corresponding purely hadronic sequence up to the onset of deconfinement and then decreases relative to it.
Neutrino trapping shifts this dynamical signature to larger
stellar masses and central pressures, reflecting the delayed appearance of
quark matter. The Maxwell-type prescription moves the onset of deconfinement to still higher central pressure and produces a more rapid decrease of the mode frequency near the stability limit. We quantify the deconfinement effect through the fractional frequency shift between hybrid and hadronic
stars at equal gravitational mass. These results show that the fundamental radial mode provides a sensitive dynamical probe of the interplay among
entropy,
lepton trapping, phase-transition construction, and
quark deconfinement in hot compact
stars.
Read the full abstract ▾
Statistics Theory
Posterior Consistency under dependent errors in non-linear inverse problems with applications to time evolution equations
Aditya Jambhale
Let $\theta \to \mathcal G(\theta)$ be a map from a parameter to the solution of some non-linear partial differential equation modelling a dynamical system. We provide conditions to show posterior consistency for $\theta$ under observations of $\mathcal G(\theta)$ with dependent noise, modelling potential time-correlations of the sensors used to make the measurements. We provide two examples, namely the Navier-Stokes equations with unknown viscosity and initial condition, as well as a Reaction-
Diffusion type equation with diffusivity coefficient that is varying (heterogeneous) in space, which requires novel stability estimates for a Darcy transport operator with accumulation in time.
Read the full abstract ▾
General Economics
Optimal Retirement of European Fossil Fuel Power Plants and the Cost of Delay
Imke Rhoden
Mitigation pathways require fossil power plants to retire well before the end of their technical lives, yet retirement decisions are typically derived from cost-minimizing system models rather than from the decision facing the plant's owner. This study investigates the carbon price threshold at which closure becomes more advantageous for an operator, as well as the ongoing costs associated with continued emissions. A finite-horizon Markov decision process is solved plant by plant for the European fossil fleet of 850 units, with a carbon price that moves stochastically and plants that age deterministically. The price process is constructed so that widening its range leaves the expected price unchanged, isolating the effect of uncertainty from the effect of price level. Closure thresholds fall below recent EU allowance prices for all fuels considered, and the sensitive range lies well beneath them, so further price increases are unlikely to be the binding instrument for retiring this fleet. Because fuel costs rather than emissions determine how much margin a carbon price must erode, the least carbon-intensive capacity retires first. Uncertainty raises the thresholds by about a tenth at an unchanged expected price: a credibly stable carbon price retires the same capacity as a volatile price roughly a tenth higher. Policy credibility therefore substitutes for price level.
Read the full abstract ▾
Optimization and Control
The Cost of Differential Privacy in Linear-Quadratic Dynamic Games
Chih-Yuan Chiu and Matthew Hale
Multi-
agent coordination often requires strategic
agents to share sensitive information about their states or objectives, creating a tension between performance and
privacy. Our paper studies this tradeoff in
stochastic linear-quadratic (LQ) dynamic games with heterogeneous
agent objectives. In our framework,
agents share noise-perturbed state and reference information with a cloud computer that computes feedback Nash
equilibrium strategies, with the injected noise calibrated to provide differential
privacy. We derive an analytical expression for each
agent's infinite-horizon steady-state cost of
privacy relative to the non-private game. Then, we prove that when
agents' objectives are sufficiently aligned, the injection of
privacy noise necessarily incurs a positive performance cost. In contrast, we characterize a class of games with sufficiently misaligned objectives across
agents for which an
agent's cost of
privacy can be strictly negative. Thus, counterintuitively, noise can simultaneously protect
privacy and improve the
equilibrium performance of an
agent when the objectives of interacting
agents are sufficiently misaligned. Finally, we present numerical experiments which corroborate our theoretical contributions.
Read the full abstract ▾
Artificial Intelligence
SPECTRUM: Proximal Spectral Modulation for Looped Self-Distillation
Yunbo Long, WenJie Chen, Jiaquan Zhang et al. (+4 more)
A model that learns from its own outputs inherits more than their correctness: it inherits which solutions it produces. We formulate Looped Self-
Distillation, a self-
evolution framework for code generation in which a model repeatedly generates and learns from its own raw outputs, under a fixed information budget, without ongoing external assessment or test-based selection of the generated samples. We identify a consequential separation: correctness can improve while the breadth of correct implementations contracts. We introduce SPECTRUM, which re-estimates loss-sensitive key/value geometry from a fixed reference anchor at each round and converts it into full-rank proximal spectral modulation. All generated completions train a single student, whose subsequent inference requires no intervention. After five rounds of experiments on MBPP, SPECTRUM retains 89.9% of the initial model's 64-sample correct AST richness, compared with 66.4% for Vanilla self-
distillation and 65.5% for a subspace-projection control. The advantage persists at matched correct-sample counts. Without further training or recalibration, the resulting student also achieves higher matched-correct richness than Vanilla SD on HumanEval+ and APPS Intro, demonstrating transfer of the diversity benefit. These findings establish correct-solution retention as a complementary objective of recursive self-improvement (RSI) and show that generation-time intervention can improve the solution repertoire retained by subsequent students.
Read the full abstract ▾
All authors Yunbo Long, WenJie Chen, Jiaquan Zhang, Guangya Hao, Zihang Zeng, Pengze Li, Xi Chen
Fluid Dynamics
Well-posed by Design: Learning Constitutive Laws from Velocity Data using Convex Neural Network Potentials
Gonzalo G. de Diego and Georg Stadler
Learning constitutive laws of complex fluids from velocity data (i.e. indirect observations) is a PDE-constrained inverse problem in which expressive neural parameterizations risk breaking the well-posedness of the forward physics model. We address this tension by learning, rather than the constitutive law itself, the dissipation potential: a scalar function whose convexity, frame-indifference, and dissipativity propagate into the underlying continuum mechanics and guarantee the structural properties needed for a well-posed and generalizable forward PDE. We introduce ICNNE, an input-convex neural architecture that exactly enforces convexity, frame-indifference, and evenness in the second strain-rate invariant via symmetrization, alleviating the numerical instabilities that occur at small strain rates in standard formulations. By weakly enforcing zero gradients in the origin of the potential via penalization, ICNNE also satisfies the dissipativity property. The loss objective and its gradient are computed using a combination of finite element and neural network methods, coupling the Firedrake and PyTorch libraries. We evaluate on four problems spanning compressible and incompressible regimes: compressible Navier-Stokes, Herschel-Bulkley yield-stress flow, Hibler's viscous-plastic sea-ice model, and discrete-element-method data with no known constitutive law. We show that the learned potentials (i) recover ground-truth physics where it is known, (ii) transfer accurately to geometries unseen during training, and (iii) succeed in regimes where unstructured methods diverge. These results suggest that
embedding mathematical structure into the parameterization, rather than into the loss, is a robust path to learning physics from indirect data.
Read the full abstract ▾
Optics
Recoil-heating modulation in levitated optomechanics through optical wavefront shaping
Hugo Marot, Joel Love Nanfack, Murad Abuzarli et al. (+3 more)
Optical levitation relies on light forces to trap mesoscopic objects in a
vacuum and assemble pristine oscillators, whose mechanical properties are commonly tamed through
laser-intensity modulation–e.g., to provide leverage over the stiffness or engineer
quantum states. When approaching the
quantum regime at low pressure, the oscillator's dynamics is governed by
recoil heating, which stems from the backaction of scattered
photons and whose control is highly desirable. Here, we demonstrate that spatial degrees of freedom accessible within the trapping-light field can be harnessed to modulate
recoil heating. Specifically, we introduce an optimization routine that spatially shapes the optical field's wavefront to either maximize or minimize the
recoil-heating rate independently of other parameters like the mechanical stiffness. Importantly, this approach can serve to devise sequences of patterns providing smooth and deep modulations. Implementable into common experimental setups, this work leads the way to the dynamic modulation of optomechanical coupling in optical levitation.
Read the full abstract ▾
All authors Hugo Marot, Joel Love Nanfack, Murad Abuzarli, Yann Louyer, Mathias Perrin, Nicolas Bachelard
High Energy Physics (Theory)
Geometry of massive loops in gravitational scattering: Maximal cut, Riemann's bilinear relations and genus drop
Arpan Bhattacharyya and Saptaswa Ghosh
We study a representative two-loop family of worldline
quantum field theory integrals relevant for classical gravitational scattering with a massive mediator and investigate the
geometric structure of its maximal cut. We find that, unlike the massless case, the massive loops already give rise to an
elliptic maximal cut at two loops, which degenerates into a nodal curve in the massless limit and in the static limit. In the global Baikov representation, two successive integrations reduce the maximal cut to a period of an
elliptic curve. We further establish five primary consequences of this study: (a) The integration-by-parts method generates five master integrals, which, under reflection, split into a three-dimensional
elliptic block and a two-dimensional polylog block. We derive the differential equation in two ways, one by Laporta reduction and subsequently by intersection theory in two cocycle bases, and construct the rational map between them. (b) The period
matrices in arbitrary dimension have been computed in terms of Appell and Beta functions. Furthermore, we derive twisted Riemann period relations that yield nine quadratic identities among them. (c) We study the global monodromy relation of the differential system that satisfies Riemann's bilinear relation. (d) We explicitly derive the function space of the iterated integrals in terms of
elliptic multiple polylogs. (e) In impact-parameter space, the
Fourier-transformed cut is a Bessel-twisted differential on the
elliptic curve, which we show has extra involution
symmetries. Thus, the parity of the differential further decides which quotient - rational or
elliptic its periods live on. All results presented concern the maximal cut and, therefore, the homogeneous solutions, but they open an important pathway to tackling the full massive two-loop integral.
Read the full abstract ▾
High Energy Physics (Phenomenology)
Toward N$^3$LO CT global PDF analyses: The SACOT-$\chi$ scheme in DIS and DGLAP evolution at partial N$^3$LO
Marco Guzzi, Yao Fu, Casey Hampson et al. (+5 more)
We discuss three aspects of the ongoing efforts of the CTEQ–TEA group toward global parton distribution function (PDF) analyses beyond NNLO accuracy in QCD. First, we present a realization of the SACOT-$\chi$ general-mass variable
flavor number scheme for deep-inelastic scattering (DIS) at the partial next-to-next-to-next-to-leading order (pN$^3$LO) in QCD. We numerically illustrate how this SACOT-$\chi$ theory prediction captures the dependence of DIS structure functions on masses of heavy
quarks with inclusion of the available $O(\alpha_s^3)$ contributions. Second, we discuss a preliminary pN$^3$LO CT25 fit that implements such contributions. Third, we extend the Candia
algorithm for solving DGLAP
evolution equations for unpolarized parton distribution functions (PDFs) in $x$ space to N$^3$LO in QCD. The
algorithm is based on logarithmic expansions of the solution and can be extended to all orders in QCD. An expansion equivalent to the exact DGLAP solution at N$^3$LO is presented in the non-singlet sector. We show results for approximate N$^3$LO PDFs evolved using the most recent approximations to the 4-loop splitting functions in QCD.
Read the full abstract ▾
All authors Marco Guzzi, Yao Fu, Casey Hampson, Pavel Nadolsky, Maximiliano Ponce-Chavez, Tanishq Sharma, Keping Xie, C.-P. Yuan
Machine Learning
Benchmarking Time Series Foundation Models for Load Forecasting Under Covariate Uncertainty
Tomas Kaljevic, Ivan Arzola and Yu Zhang
Accurate short-term load
forecasting (STLF) is essential for the reliable and efficient operation of modern power systems. While time series foundation models (TSFMs) have recently demonstrated remarkable performance across a wide range of
forecasting tasks, their effectiveness for STLF under realistic operational conditions remains largely unexplored. In this paper, we present a comprehensive
benchmark of four trained-from-scratch (TFS) models and four TSFMs across three real-world load
forecasting datasets under operational scenarios that differ in the availability and quality of future
covariate information. Our results show that Chronos-2 consistently achieves state-of-the-art performance in both zero-shot and fine-tuned settings when future
covariates are available or accurately forecast. However, its performance degrades as
covariate forecasts become increasingly noisy, whereas TimesNet exhibits greater robustness under severe
covariate uncertainty. These findings demonstrate the effectiveness of
covariate-informed TSFMs for STLF while highlighting the critical role of robust
covariate modeling in real-world
forecasting applications.
Read the full abstract ▾
Robotics
Monocular Navigation Relative to Unknown Spacecraft Using a Transformer-Aided Kalman Filter
Pol Francesch Huc and Simone D'Amico
This work presents a novel learning-based pipeline for pose estimation of unknown spacecraft using only
monocular images from a single servicer. The approach combines a
transformer-based neural network with a Multi-State Constraint Kalman Filter (MSCKF) to estimate the pose (i.e., position and orientation of the target spacecraft relative to the camera) throughout rendezvous and proximity operations. Unlike existing
vision-based methods that require prior knowledge of the target shape or inertia properties, rely on additional sensing modalities such as depth,
lidar, or stereo, or only recover translation up to scale, the proposed pipeline generalizes to previously unseen spacecraft using a single
monocular camera. The
transformer network estimates the odometry, the change in pose between images up to scale, from SuperPoint features matched by LightGlue. The MSCKF uses these pseudo-measurements along with an orbit and attitude kinematics model to estimate the pose of the target. In particular, the relative orbit elements, the target's attitude with respect to the servicer's camera, and the associated angular velocity are estimated directly by the filter. Given the
monocular approach and short distance to the target, the full observability of the range to the target is recovered via attitude maneuvers by the servicer. The method is trained and evaluated on a re-rendered high-resolution version of the SPE3R
dataset, which includes synthetic images of 103 spacecraft. Eleven of these spacecraft are held out during training to evaluate the generalization to unseen targets. Monte Carlo simulations are then used to evaluate the
navigation pipeline on rendered trajectories of the held out spacecraft. The results demonstrate that learned
vision pipelines as a front-end for Kalman filters provide median errors of 3.7{\deg} in attitude and 2.2% of range in ROE when navigating about unknown targets.
Read the full abstract ▾
Networking
Ofan: Optimal Load Balancing for AI Training
Sarah McClure, Evyatar Cohen, Jakob Krebs et al. (+5 more)
The extreme collective completion time (CCT) demands of AI workloads challenge existing packet spraying
algorithms, which can have trouble efficiently load-balancing workloads that are sent at full line rates. We trace this to a structural cause: on a fat tree, once a packet picks its upward path, the downward path to its destination is unique, so destination-oblivious schemes cannot undo the imbalance it creates. We prove that such schemes can suffer from $\Theta(\sqrt{m})$ queueing for messages of size $m$, thus eventually triggering rate reductions by the congestion control. Instead, we suggest Ofan, a switch-based destination-aware LB scheme that can reach $O(1)$ queueing. We also present its pOfan variant that fits the pipe-based architecture of current switches. Our P4 implementation shows that it consumes modest resources. An end-to-end FSDP2 evaluation with Llama-3 405B-parameter models shows that Ofan cuts CCT inflation by $16$–$39\times$ when compared to existing
algorithms.
Read the full abstract ▾
All authors Sarah McClure, Evyatar Cohen, Jakob Krebs, Alexander Shpiner, Mark Silberstein, Scott Shenker, Sylvia Ratnasamy, Isaac Keslassy
Machine Learning
Conditional Flow Matching for Transport Between Markov Processes
Syamantak Kumar, Dheeraj Nagaraj, Saptarshi Roy et al. (+1 more)
Motivated by sequence-to-sequence transport in the context time-series domain adaptation, we study the problem of transportation between trajectories of Markov processes. Given a limited number of trajectories from source distribution and the target distribution, we formulate a flow matching based
algorithm which learns a transport map from the source to target trajectory distribution, while preserving the Markov structure. We show that this is consistent in the population limit and derive finite-sample error bounds under mixing time assumptions, following the analysis of classical statistical problems including regression (Nagaraj et al., 2020), principal component analysis (Kumar and Sarkar, 2023), and
matrix concentration (Neeman et al., 2024) in the Markov setting. We complement that with a lower-bound construction showing that a mixing-time dependent sample complexity is unavoidable even with regular Gaussian conditional transitions. We evaluate on synthetic and real-world data. For image
retrieval from electroencephalography (
EEG) on THINGS-
EEG2 (Gifford et al., 2022), the task is to identify the viewed image from
EEG signals captured from human subjects, which suffers from high inter subject variability. We augment the ENIGMA decoder (Kneeland et al., 2026) with a conditional flow before its subject-specific temporal map. This improves mean top-5
retrieval accuracy from 43.87% to 49.05%, an 11.82% relative improvement.
Read the full abstract ▾
All authors Syamantak Kumar, Dheeraj Nagaraj, Saptarshi Roy, Purnamrita Sarkar
Machine Learning (Statistics)
How Inefficient Is Natural Gradient Descent? From Exact Optimality to \Theta ( \sqrt{ \log d } ) Divergence
Guni Sharon and Alan Kuhnle
Natural gradient descent (NGD) underlies common methods in ML. For dually flat families, idealized NGD on the forward Kullback–Leibler objective follows the mixture geodesic which is often longer than the shortest Fisher–Rao path. We quantify this overhead by the inefficiency ratio \(R \ge 1\), the Fisher length of the mixture geodesic divided by the Fisher–Rao distance, and bound its supremum over endpoint pairs as a function of the parameter dimension \(d\). A tensor criterion identifies the regime (I) families, with \(R=1\) everywhere: exactly those with quadratic potential or dimension one, such as fixed-covariance Gaussians. For non-quadratic families, we prove two further regimes: (II) bounded third-order skewness plus finite Fisher–Rao diameter yields a dimension-independent bound; and (III) for products of scale families—including Gaussian covariances and Gamma rates—\(R\) grows as \(\Theta(\sqrt{\log d})\), unbounded in \(d\). Under a per-step Fisher-chord budget, \(R\) translates to a practical computational cost: NGD requires asymptotically at least \(R\) times as many steps as an optimizer following the Fisher–Rao geodesic. Experiments confirm all three regimes: \(R=1\) to machine precision for quadratic-potential families (I), the categorical bound \(\pi/(2\sqrt{2})\) is approached but not attained (II), and sampled scale-product \(R\) grows with \(d\), reaching \(R \approx 1.5\) for long, high-dimensional moves (III).
Read the full abstract ▾
Number Theory
Systems of equations with elements of finitely generated groups and with linear recurrence sequences
Subham Bhakta, Alina Ostafe and Igor E. Shparlinski
We obtain new and essentially optimal bounds on the number of solutions of diagonal systems of $k$
polynomial equations in $s$ variables from an arbitrary finite subset of a finitely generated group in a field of characteristic zero. Our bounds generalise and sometimes improve previous results of J. Bourgain, M. Z. Garaev, S. V. Konyagin, and I. E. Shparlinski (2014), for $k=1$, and of A. Manning, A. Ostafe and I. E. Shparlinski (2026), for $k=2$. In turn, this improvement leads to sharpening a result of A. Manning, A. Ostafe and I. E. Shparlinski (2026) concerning arithmetic statistics of certain
matrices. We also obtain similar results for $s$-tuples of linear recurrence sequences, in the case of linear equations and quadratic equations.
Read the full abstract ▾
Machine Learning
Minimal Witness Reinforcement Learning
T. Y. Tsui, Zihao Ye, Pengxiang Cai et al. (+3 more)
“What are the irreducible conditions that are sufficient to produce an outcome?” is one of the most common questions that recur across computation and science. Its answers, the minimal sufficient witnesses, are what we mean by explanations, mechanisms and reasons. These problems usually ask for multiple minimal witnesses, yet standard RL methods may reveal only one solution or redundant ones. We formalize this problem as minimal-witness identification and introduce Minimal-Witness Reinforcement Learning (MWRL). MWRL takes the union of the sets certified by successful proposals sampled from the policy and credits each proposal for the coverage the group union would lose without that proposal. This credit assignment, derived directly from the problem definition, unifies the demands for minimality and recovery of alternatives from a single black-box verifier bit. Under this principle, we derive a value iteration planner that recovers the entire family of witnesses and a policy gradient method that can scale to large
language models. Across different experimental settings, MWRL recovers most minimal witnesses, while other methods return redundant supersets or a single witness. By making witness families learnable from verifier feedback, MWRL expands the scope of reinforcement learning beyond single-solution optimization. Our code is available at https://github.com/TSUITUENYUE/MWRL.
Read the full abstract ▾
All authors T. Y. Tsui, Zihao Ye, Pengxiang Cai, Yanchao Li, Yuqiang Li, Zhehong Ai
Analysis of PDEs
The Dirichlet problem with boundary data of fractional smoothness in domains with Ahlfors regular boundary
Ariel Barton, Svitlana Mayboroda and Alberto Pacati
We establish well posedness of the Dirichlet problem in weak local John domains $\Omega\subset\R^\dmn$ with Ahlfors regular boundaries of possibly non integer dimension $0<\bdmn<\dmn$, for linear second order
elliptic equations with real coefficients, and with data in Besov spaces with a very broad range of acceptable parameters.
Read the full abstract ▾
Computation and Language
TIDE 2.0: an open, model-agnostic engine for keyed de-identification of clinical notes
Jose D. Posada, Somalee Datta and Priya Desai
Clinical notes capture most of what is documented about a patient's care, but they cannot be used for research until protected health information (PHI) is removed. De-identification is often treated as a detection problem. Detection alone is not sufficient: redaction strips clinical content along with identifiers, date blanking destroys the temporal intervals needed for longitudinal analysis, and assigning a fresh random surrogate at each occurrence breaks links between a patient's notes. We present TIDE 2.0, an MIT-licensed engine with two separable stages: an interchangeable recognizer and a keyed anonymizer. Both run on hardware the institution owns. Surrogates are generated cryptographically with no stored linkage table. Dates shift by a per-patient, interval-preserving offset; each value receives the same surrogate across all occurrences under a given key; and a release produced under a new key cannot be linked to earlier releases. We also release TIDE2-Sentry, a recognizer distilled from a large
language model. On two gold-annotated corpora from two institutions, the default configuration reached span-level recall of 0.88 in-domain and 0.77 on the second institution's corpus, at precision 0.88 and 0.87. We report recall and precision per category alongside these aggregates. The engine is open source, and the recognizer is available under a gated research-use agreement, so institutions can run, inspect and extend both within their own environments.
Read the full abstract ▾
Computer Vision
Deep Learning Based Illegal Bowling Action Detection
Debopom Sutradhar, Niful Islam, Sudipto Mondal et al. (+3 more)
Cricket, often referred to as the "gentleman's game," adheres to a strict rule set for both batsmen and bowlers, where each delivery can significantly impact the match outcome. Detecting illegal bowling actions is crucial for maintaining fair play, yet it remains challenging for umpires to monitor in real time. Existing sensor-based solutions have limitations in live match scenarios, making real-time assessment difficult. This paper proposes a computer
vision-based deep learning solution to detect illegal bowling actions in live cricket matches. To develop and evaluate our approach, we compiled a
dataset of 62
videos featuring 11 male bowlers, capturing both legal and illegal bowling actions from multiple angles-front, back, and side. However, the
dataset predominantly comprises right-handed bowlers with conventional actions. The proposed system identifies two key frames, the shoulder frame and the release frame from
video footage of a bowler's delivery and analyzes the change in the bowling arm's angle between these frames. If the angle difference exceeds a predefined threshold (e.g., 15 degrees), the delivery is flagged as potentially illegal. We evaluated the system on a custom
dataset and achieved a high true positive rate, suggesting the system's potential effectiveness in real-time match settings. However, further research is required to validate the system across diverse environmental conditions and larger
datasets to ensure generalizability and robustness in various live match scenarios. To the best of our knowledge, this is the first AI-based computer
vision method for detecting illegal bowling actions in cricket.
Read the full abstract ▾
All authors Debopom Sutradhar, Niful Islam, Sudipto Mondal, Tasmima Hossain Jamim, Jubaer Muhammad Shufol, Swakkhar Shatabda
Materials Science
Effect of Source-to-Substrate Distance on the Optical and Electrical Properties of Thermally Deposited CdTe, CdSe and Sb$_2$S$_3$ Thin Films
Himanshu Sharma Pathok and Prasanta Kumar Saikia
In this work, cadmium telluride (CdTe), cadmium selenide (CdSe), and antimony sulfide (Sb\(_2\)S\(_3\)) thin films were deposited using the thermal evaporation technique by varying the source-to-substrate distance (SSD) from 15 to 24 cm (15, 18, 21, and 24 cm). The influence of the SSD on the structural, compositional, optical and electrical properties of the deposited films was systematically investigated. XRD analysis revealed the polycrystalline nature of the CdTe and CdSe thin films, whereas the Sb$_2$S$_3$ thin films exhibited an amorphous nature. EDX spectroscopy confirmed the presence of the constituent elements in the deposited films. Optical analysis revealed a marginal variation in the optical band gap with increasing SSD, while the film thickness, determined using the envelope method, decreased with increasing SSD. The optical band gap energies of CdTe and CdSe were found to be below 2 eV, whereas Sb$_2$S$_3$ exhibited a relatively higher band gap. The relatively low band gap energies and high absorption coefficients of the CdTe and CdSe films make them promising candidates for absorber layer applications in thin film
solar cells (TFSCs). The films exhibited relatively low optical transmittance, which is desirable for efficient light absorption in absorber layers. Electrical characterization confirmed the semiconducting nature of all the deposited thin films. Overall, the results demonstrate that the optical and electrical properties of CdTe, CdSe and Sb$_2$S$_3$ thin films can be effectively modified by varying the SSD. The findings indicate that, in addition to the well established CdTe absorber, CdSe thin films show potential for application as absorber materials in TFSCs.
Read the full abstract ▾
Quantum Physics
Open-System Probe Selection and Optimal Measurements for Dispersive Coupling Metrology
Achal Vinod, Da-Wei Luo and Ting Yu
We investigate how
photon loss and environmental
memory affect probe selection and measurement sensitivity for estimating the dispersive coupling strength in a cavity
qubit system. Using density operator and
quantum state
diffusion approaches, we examine the
quantum Fisher information for representative separable and entangled initial states in Markovian and non Markovian regimes. The results show that closed system generator variance alone does not determine open system performance. The survival of parameter sensitive coherences and the influence of environmental
memory also play important roles. Within the single
photon preparations examined, a separable probe can outperform the entangled family considered. Comparisons of selected coherent and squeezed
vacuum preparations at equal mean
photon number further illustrate how
photon number statistics and
qubit preparation affect sensitivity. For the parameters studied, finite environmental
memory can enhance and prolong sensitivity and yield nonzero QFI for a probe whose Markovian QFI vanishes. We also numerically identify a fixed Hermitian observable whose error propagation uncertainty remains close to the
quantum Cramer Rao bound over a prescribed coupling interval in the Markovian regime. These findings clarify the interplay between probe preparation, environmental dynamics, and measurement choice in dispersive
quantum metrology.
Read the full abstract ▾
Machine Learning
Data, Numbers, and Geometry: Three Tutorials on Numerical Methods, Machine Learning, and Evaluation
Jessica N. Howard, Yidi Qi and Tomás S. R. Silva
We present three practical tutorials on numerical computation and machine learning for mathematical research, developed for the DANGER: Data, Numbers, and Geometry workshop held at the Banff International Research Station in April 2026. The first develops a numerical approach to exterior calculus from pointwise evaluations of differential forms, using a flux formulation of the exterior derivative. Examples in Euclidean space and on the sphere illustrate
geometric identities, topological features, and the effects of approximation and finite precision. The second examines how mathematical structure guides neural network design through examples involving
elliptic curves, quivers, and a boundary value problem. It explores how architectural choices affect learning and uses interval arithmetic to bound the residual of a trained network over the full interval of the boundary value problem. The third addresses the evaluation and presentation of machine learning results, covering performance metrics, statistical uncertainty, classification thresholds, receiver operating characteristic curves, and accessible figure design. Throughout, the tutorials distinguish numerical agreement, predictive accuracy, structural guarantees, and rigorous bounds as different forms of evidence. Each contribution can be read independently, with accompanying notebooks and exercises that allow readers to reproduce the examples and adapt the methods to other problems.
Read the full abstract ▾
Artificial Intelligence
Cascadia: Resident 975B MoE Inference on Eleven AI PCs
Tate Berenbaum, Matias Parij and Muthaiah Venkatachalam
Mixture-of-experts models make nearly trillion-parameter capacity accessible with sparse per-
token computation, provided that the serving system can distribute the weights and coordinate their execution. We present Cascadia's resident execution of Inkling, a 975B-total/41B-active-parameter model, on eleven Intel Core Ultra X7 358H AI PCs, each with 64 GB of
memory, Arc B390 integrated graphics and gigabit Ethernet. We contribute a custom resident MoE engine that preserves Inkling's routing rules, constructs compressed graphs for OpenVINO's fused iGPU primitives, and coordinates FP16 expert computation with FP32 output restoration. The engine fits six consecutive decoder layers per machine and represents dense feed-forward blocks as all-active expert slices, reducing measured dense-layer call time from approximately 8.1 to 4.5 ms. A streaming pipeline coordinates concurrent generation, while captured-state draft evaluation measures agreement with the deployed numerical path. Paired measurements at fifteen concurrency levels from 1 to 176 streams reach 60.29 aggregate decode
tokens/s at 88 streams, with 46.87
tokens/s over the complete serving phases. At fifteen streams, median first-
token latency is 6.05 s. Raising the context budget from the 1,024-position default, real
prompts of 1k to 64k
tokens recover the embedded code in all 19 measured answers, with first-
token time growing as $aN+bN^2$ and decode latency growing approximately linearly, both bounded by a single-threaded CPU
attention loop rather than by
memory, which holds 512k positions per stream. Evaluation on captured fleet states separates the effects of vocabulary selection and weight quantization on draft agreement. Together, these contributions establish an execution and evaluation approach for large sparse models on distributed client systems with shared CPU-GPU
memory.
Read the full abstract ▾
Machine Learning
Constant-Curvature Sliced Gromov-Wasserstein for Heterogeneous Cross-Curvature Alignment
Shanglin Li, Wenjing Lu, Muyang Li et al. (+2 more)
Recent advances in representation learning have highlighted the utility of constant-
curvature models, such as
hyperbolic and spherical spaces, for modeling complex data. Mixed-
curvature models further enhance this by integrating multiple constant-
curvature components. However, these models typically learn each component space independently because spaces with different
curvatures are inherently heterogeneous and lack a unified metric. Consequently, they lack explicit mechanisms to enforce
geometric consistency across various spaces. Moreover, the problem of comparing probability distributions across mixed-
curvature spaces remains unexplored. To compare distributions on heterogeneous spaces, Gromov-Wasserstein (GW) distances provide a principled framework by aligning their intra-space geometries. Building on this, we propose constant-
curvature sliced Gromov-Wasserstein (CCSGW), a novel divergence for aligning distributions supported on heterogeneous constant-
curvature spaces. We first introduce the missing geodesic-based one-dimensional projections for spherical spaces, and then extend sliced GW to constant-
curvature spaces, enabling efficient and principled comparison across
manifolds with different
curvatures. This formulation preserves intrinsic
geometric relationships while avoiding the high computational cost. We provide theoretical analysis showing that CCSGW controls intrinsic
geometric discrepancy across heterogeneous spaces, promoting distribution-level
geometric consistency. By integrating CCSGW into existing mixed-
curvature learning tasks, including graph anomaly detection, graph node classification, and
multimodal learning, we observe consistent performance gains across diverse settings.
Read the full abstract ▾
All authors Shanglin Li, Wenjing Lu, Muyang Li, Nicu Sebe, Ziheng Chen
Robotics
RoboCap: A New Platform for Egocentric Robot Learning
Qasim Ali, Nikash Bhardwaj, Kheri Hughes et al. (+15 more)
Despite its promise for scaling
robot learning,
egocentric manipulation data is still scarce today. Collection at scale requires vertically integrating ergonomic hardware with centimeter-precise 3D
algorithms, at a precision that has not been publicly demonstrated. To address this gap, we introduce RoboCap, a 250\,g six-camera dual-IMU hat designed for in-the-wild
egocentric data capture, and the Grounded API, a suite of device-agnostic 3D
algorithms tuned for RoboCap. In this report, we demonstrate how hardware, calibration, and 3D
algorithms interact to achieve state-of-the-art performance on the public
benchmarks: our SLAM across diverse settings and rigs, our depth estimation on
egocentric settings, and our hand tracking when adapted to third-party devices.
Read the full abstract ▾
All authors Qasim Ali, Nikash Bhardwaj, Kheri Hughes, Sam Cho, Ethan Lee, Paul Luo, Shinn Chong, Sunny Yu, Owen Akhibi, Alex Proshkin, Haotian Zhan, Niresh Dravin, Vivek Sharma, Jonathan Victor, Amanda Young, Michael Cho, Ademi Adeniji, Vincent Liu
Sound
Exposing and Mitigating Neural Codec Vulnerabilities in Audio Deepfake Detection
Abdullah, Awais Khan and Khalid Mahmood Malik
Existing audio deepfake detection (ADD)
datasets and detectors are primarily built for vocoder-based synthesis, evaluated against traditional post-hoc perturbations such as MP3/AAC compression or additive noise, applied independently of generation. However, recent
speech synthesizers, particularly ALM-based systems, use neural audio codecs both for compression and as the resynthesis reconstructing waveforms from generated
tokens, producing artifacts distinct from post-hoc compression. Neural codecs thus play a dual role: some are designed for pure compression under low-bandwidth communication, while others serve as resynthesis components. Despite this dual role, robustness to codec-based compression, unlike post-hoc compression, remains largely unexplored. We expose this gap, showing that state-of-the-art (SOTA) ADD models degrade drastically on codec-compressed
speech; in particular, systems trained on Codec Resynthesized data as a proxy for codec-based generation prove most vulnerable, with legitimately compressed bona fide
speech often misclassified as fake. To investigate this, we construct the Audio Neural Codec-Spoof
dataset by applying seven neural codec
algorithms to existing ADD
benchmarks, isolating codec-induced resynthesis artifacts as a controlled proxy for codec-based generation. As baseline mitigation, we propose PCL-NET (Pairwise Consistency Learned Network), fine-tuning a pretrained XLS-R (300M) encoder with a pairwise consistency objective that minimizes the representation distance between an utterance's uncompressed and codec-compressed versions, disentangling codec artifacts from the real-versus-fake decision. As a result, PCL-NET reduces average EER under neural codec compression from 28.67% to 12.77%, while preserving competitive CoSG-based deepfake detection performance. We will also make the
dataset publicly available on Hugging Face upon acceptance.
Read the full abstract ▾
Plasma Physics
An Implicit, Moment-Conserving Low-Rank Tensor Method for the Hybrid Kinetic-Ion Fluid-Electron Plasma Model
Stephen R. White, Adam J. Stanier, Luis Chacón et al. (+2 more)
Quasi-neutral hybrid models for simulating
plasmas, with kinetic ions and fluid electrons, have shown great promise for approaching the multi-scale nature of
plasmas. In this paper we present a novel hybrid scheme for electromagnetic settings that conserves mass, momentum, and energy, employing a low-rank tensor method utilizing the Tensor Train decomposition to avoid the \enquote{curse of dimensionality.} Electrons are simulated as a continuous fluid, while the kinetic ions are modeled as a distribution function governed by the Vlasov equation. The Vlasov equation is updated with the implicit midpoint method. We use a modified version of a conservative truncation method to guarantee conservation of mass, momentum, and energy throughout the method, which corrects the truncated distribution function without needing to advance a separate set of fluid-moment equations. A series of electrostatic and electromagnetic test problems is used to demonstrate the correctness of the method and its conservation properties.
Read the full abstract ▾
All authors Stephen R. White, Adam J. Stanier, Luis Chacón, Will T. Taitano, Isaac A. Castro