Instrumentation and Detectors
Papers
Fresh from arXiv, typeset properly. Highlighted words trade on the market.
All of arXiv
Wednesday 9 September
This paper proposes a novel inertial sensor based on a solid diamagnetic cubic test mass passively suspended by a symmetric configuration of six magnetic quadrupoles. Designed specifically for microgravity environments, this system eliminates the mechanical noise and hysteresis inherent in elastic suspensions, as well as the complexity associated with electrostatic suspension systems. We provide an analysis of the magnetic intrinsic noise floor imposed by eddy current damping, alongside ground-based preliminary experimental results demonstrating potential sensitivities in the order of the $10^{-10} \text{ m/s}^2/\sqrt{\text{Hz}}$ regime.
Astrophysics of Galaxies
The magnetic field in M16: new results from the JCMT BISTRO survey
We present a study of the magnetic field (B-field) in the Eagle Nebula (M16) using data from the James Clerk Maxwell Telescope (JCMT) B-fields In STar-forming Region Observations (BISTRO) survey. We extend our previous study of the well-known `Pillars of Creation' region of the nebula to study two other structures in the region, the `Spire' and the bright-rimmed cloud (BRC) known as `SFO30'. These structures show similar B-field morphologies to the Pillars. We calculate B-field strengths using the Davis-Chandrasekhar-Fermi (DCF) method and thereby estimate the magnetic potential energy in each of these regions for comparison with the local gravitational, turbulent and thermal energies, as well as the pressure energy of the shocks from nearby O-type stars. We find approximate equipartition between the energies supporting the regions and those tending towards collapse, with the B-field support playing a significant role. This agrees with our earlier hypothesis that B-fields help support pillar-like structures in molecular clouds and make them longer-lived than they would otherwise have been.
Read the full abstract
Machine Learning
Development and Validation of a Physics-Guided Machine Learning Extrapolation Framework Using a Classical Transient Diffusion Benchmark
Machine learning models used in engineering are typically trained within limited operating ranges, yet reliable predictions are often required beyond these domains. Consequently, the primary challenge is extrapolation rather than interpolation. Rigorous validation is hindered by the scarcity of data outside the training range. To address this limitation, a novel extrapolation framework is integrated with established machine learning architectures to enable accurate and physically consistent predictions beyond the training domain. The framework is established by systematically evaluating two physics-guided architectures: a Bidirectional Long Short-Term Memory (BiLSTM) network and a Physics-Informed Neural Network (PINN). A classical one-dimensional transient diffusion problem is adopted as a benchmark because its exact analytical solution provides unlimited, reliable data across the spatio-temporal domain, enabling rigorous quantitative validation. The problem is particularly challenging because the solution evolves from an initial singularity through a strongly nonlinear transient regime before approaching a steady-state linear profile. When training data are confined to an intermediate portion of this evolution, backward extrapolation toward the singularity becomes especially demanding. To improve reliability, physics-guided coordinate transformations, boundary-aware learning strategies, and stability-enhancing temporal marching are incorporated. Extrapolation is evaluated using a train-predict-validate-extend strategy, in which validated predictions are recursively added to the training set to progressively extend the prediction horizon. The results demonstrate accurate and physically consistent predictions beyond the training domain, highlighting the framework's potential for engineering applications where data availability is limited.
Read the full abstract
Numerical Analysis
The geometric convergence of a parallel domain decomposition method on manifolds
This paper establishes a convergence theory for a continuous domain decomposition method for elliptic equations on manifolds. This method originated in \cite{lions2} in the setting of Euclidean domains and was later adapted and generalized to manifolds by \cite{qin_wang_wang}. Although its convergence was known, whether the convergence is geometric has remained open. We prove its geometric convergence and provide various estimates on the convergence rate.
Read the full abstract
Machine Learning
Structure and Regularization of Fused Objectives in Kernel Discriminant Analysis for Small-Sample-Size Problems
When the number of training samples is smaller than the dimension, discriminant analysis faces a singular within-class scatter matrix, and fused objectives that combine the total variance C, the between-class scatter S_b and the within-class scatter S_w are one way to handle it. Using the seven variants of Kernelized Linear Principal Component Discriminant Analysis (KLPCDA) as a platform, we derive the structure of such objectives in this small-sample-size (SSS) setting. The training projections split into a class-mean part and a within-class part, the two scatter operators satisfy WB = 0, and all between-class information lies in the null space of S_w. A single fusion coefficient therefore cannot change the solution, and the variants, KPCA, the regularized kernel Fisher discriminant and the null space methods lie on the boundary of one family with two parameters, the fusion ratio t and a ridge parameter mu. On six tasks from five application domains, the regularized members were more accurate than the best of the seven variants chosen on validation data on all six tasks, by 0.8 to 5.7 points, also under added noise, whereas the fusion ratio gave no gain, including in controlled experiments on dimension ratio, class imbalance and class separation. Selecting a variant per task could gain less than one point, while a fixed choice made on other tasks transferred to every held-out task. The analysis thus gives a principled default for SSS kernel discriminant analysis, with diagnostic quantities, the null space dimension and the class-size spectrum, that can be computed before training.
Read the full abstract
Computer Vision
Interpreting Object-Dependent Concept Brittleness in Text-to-Image Diffusion Models
Although text-to-image diffusion models generally exhibit strong prompt-following ability, we identify a persistent and previously underexplored failure pattern in which a small subset of prompts differing only in the object consistently fails to realize the same target concept under identical generation settings. We term this phenomenon object-dependent concept brittleness. Such cases suggest systematic internal blind spots rather than random sampling noise. In this paper, we present an interpretability-oriented framework to audit and minimally correct these failures. Our key idea is to analyze denoising trajectories in a step-wise sparse autoencoder (SAE) space, where abstract style and attribute concepts become more separable than in the raw denoising representation. This sparse space enables us to compare successful and failed generations, identify concept dimensions whose evidence is missing, weakened, or temporally delayed, and construct class-level concept prototypes from reliable class-consistent samples. Based on this audit process, we introduce a lightweight inference-time correction strategy that interpolates denoising features toward the corresponding prototype in SAE space. Rather than serving as a task-specific retraining method, this intervention acts as a validation of the diagnosed concept deficiency. We evaluate the proposed framework on style and attribute failure cases across multiple diffusion backbones, with significant improvements in concept consistency, text fidelity, and repair success. Further analyses show that deeper denoising representations provide clearer concept structure, while early-stage intervention offers the strongest correction leverage. Code is available at https://github.com/Metecade/Object-Dependent-Concept-Brittleness.
Read the full abstract
Plasma Physics
High-charge, highly polarized positron beams generated from a laser-driven nanowire-array target
The generation of high-charge, highly polarized positron beams in the interaction of a linearly polarized laser pulse with a nanowire-array target is investigated. Here, laser-driven electrons emit high-energy photons through nonlinear Compton scattering (NCS), which subsequently produce electron–positron pairs through the nonlinear Breit–Wheeler (NBW) process. We model this interaction using two-dimensional spin-resolved quantum electrodynamics particle-in-cell (QED-PIC)} simulations. At positron birth, the sign of $S_z$ is statistically correlated with that of the local $B_z$. The spatiotemporal field structure arising from the laser–nanowire interaction strengthens the correlation between the birth spin sign and the direction of the subsequent transverse Lorentz impulse, thereby limiting cancellation between opposite-spin contributions at a given angle. The results show that the average polarization degree reaches $|\bar S_z|\approx0.46$, and the positron charge satisfying $|\bar S_z|>0.3$ is approximately $308\,\mathrm{nC}$. Parameter scans reveal that the high-polarization positron charge is maximized at intermediate target densities and nanowire periods. Such a source could enable polarization-sensitive studies of strong-field QED and spin-dependent phenomena in high-energy and materials physics.
Read the full abstract
Machine Learning
Meta-LinEXP3: Online-within-Online Learning for Adversarial Linear Contextual Bandits
Meta-learning has emerged as an effective paradigm for transferring knowledge across sequential bandit tasks. While substantial progress has been made for stochastic bandits and non-contextual adversarial bandits, meta-learning for adversarial linear contextual bandits (ALCBs) with random action sets remains largely unexplored. To address this problem, we propose Meta-LinEXP3, an online-within-online algorithm that constructs a predictable task-level prior from completed tasks to guide the inner LinEXP3 learner. For known context distributions, we develop a policy-centered estimator that achieves an intrinsic-dimension $\mathcal{O}(\sqrt{n})$ per-task regret bound. For unknown distributions, we introduce a past-only regularized moment estimator with an $\mathcal{O}(n^{2/3})$ leading regret term and explicit finite-sample error. We further establish a direct connection between prior accuracy and transfer regret, showing that increasingly accurate priors yield sublinear transfer-dependent regret across tasks. Experiments demonstrate the effectiveness of Meta-LinEXP3, including its application to structured hyperspectral tensor sampling.
Read the full abstract
Probability
Stable and Gaussian Fluctuation Limit of a Lévy-Driven Slow-Fast System
We study fluctuation limit of a slow-fast system driven by $\alpha$-stable Lévy noise with $1<\alpha<2.$ The slow component is generated by an odd polynomial function $f(y):=y^q,$ while in the fast component, the drift is $g(y):=-|y|^p\operatorname{sgn}(y)$ for some $p>0.$ Although the noise is given, the fluctuation limit is either a stable process or a Brownian motion, depending on both $p$ and $\alpha.$ The critical line between stable limit and Brownian limit is $q+1-p=\alpha/2.$
Read the full abstract
Machine Learning (Statistics)
FlowCPO: A Unified Divergence View of Preference Alignment for Flow Models
Preference alignment for flow and diffusion models now spans online reinforcement learning and offline preference optimization, but the relation between these methods remains unclear. In particular, existing forward-process alignment methods require fresh samples from the current model, while offline methods based on fixed preference pairs rely primarily on positive-only fine-tuning or DPO-style likelihood-ratio surrogates. We organize these approaches through a divergence-based framework and introduce FlowCPO, an offline forward-KL objective that uses both preferred and dispreferred samples without online rollouts. For linear interpolation, we show under explicit regularity conditions that the forward-KL objective is bounded by a contrastive flow matching loss, yielding a tractable surrogate on fixed data. We further show that this loss is nonnegative, whereas the signed regression loss of simplified FlowDPO can be unbounded below. In the in-domain setting, FlowCPO achieves higher mean GenEval and OCR scores than the evaluated baselines, reaching 0.84 and 0.87 versus 0.81 and 0.74 for FlowDPO at CFG 3.0. In the out-of-domain setting, the results are mixed, with the best GenEval result but lower reward scores than RFT on several metrics.
Read the full abstract
Machine Learning
Beyond Conventional Federated Learning via High-Order Regularization
Federated clients that perform several local optimization steps can return parameter displacements with widely different magnitudes. The quadratic regularization of FedProx grows linearly with displacement and therefore offers limited control over the contrast between ordinary and unusually large client movements. We here introduce HiFedProx, which replaces the quadratic penalty with a scale-matched power-type regularizer indexed by $p\geq2$. All powers have the same regularization-gradient magnitude at a reference displacement $R$, while every $p>2$ gives a weaker response below $R$ and a stronger response above it. An exact affine reference calculation shows that increasing $p$ compresses relative displacement disparities, although very large powers approach fixed-radius behavior and increase local curvature. HiFedProx combines this geometry with finite-budget stochastic client optimization and same-minibatch Armijo backtracking. In paired five-seed experiments on a frozen 60-writer FEMNIST subset, a common-parameter study over $p\in\{2,3,4,5,6,7,8\}$ shows similar clean-training performance but substantial gains under composite stress. The lowest moderate- and severe-stress losses occur at $p=7$ and $p=6$, improving over $p=2$ by $11.44\%$ and $23.16\%$, respectively. Although displacement-tail ratios continue to decrease through $p=8$, predictive performance peaks in an intermediate range and Armijo trial cost increases with $p$. These results indicate that the exponent should be calibrated rather than maximized. In our experiments, $p=5$–$7$ provides the most useful range.
Read the full abstract
Audio and Speech Processing
SphereVAE: Hyperspherical Latent Autoencoders for Robust Autoregressive Speech Representation Modeling
With the rapid development of speech generation technology, discrete codec representations have been widely used because they provide a stable prediction paradigm. In expressive speech generation, however, the quantization bottleneck of discrete codecs results in information gaps in fine-grained prosody, timbre, pronunciation, and frame-to-frame continuity. Continuous representations (e.g., VAE latents), by eliminating this constraint, have emerged as a more effective alternative for autoregressive modeling. Yet when continuous representations are used as autoregressive prediction targets, prediction errors can accumulate along the generation chain, causing latent drift and degrading long-form stability. To mitigate this problem, we propose SphereVAE, which constrains the VAE latent space to the unit hypersphere. SphereVAE defines a Power Spherical posterior on the hypersphere and regularizes the latent distribution toward a uniform prior, so that information is encoded mainly by directional variation, providing a bounded geometric target for autoregressive prediction and reducing the risk of norm drift. SphereVAE underperforms the standard VAE on reconstruction metrics due to reduced latent freedom. However, when integrated into VoxCPM for zero-shot TTS and long-text generation, it yields lower content error rates with comparable speaker similarity, and shows more stable long-range speaker consistency. These results indicate that an appropriate latent geometric constraint can effectively mitigate autoregressive error accumulation and drift in speech generation.
Read the full abstract
Computation and Language
Contrastive Projection: Reading Transformer Internals by Differencing Logit Lenses
Reading a transformer's internal states in token space is easy to do and hard to trust: a logit lens on a single hidden state is dominated, at intermediate layers, by the generic tokens the model would predict for almost any input. We read the difference instead. Subtracting two closely matched prompts' hidden states and projecting through the unembedding cancels the shared component and surfaces what separates them, an operation equivalent to reading a RepE/ActAdd steering vector through a logit lens. Built into a training-free tracer that reads at every position, sub-layer, and head and averages over designed baselines, it traces a compound- noun MLP->attention chain in Phi-2, confirmed there by activation patching, with the same distinction recovered across three architectures by readout and probe rather than by patching; it reads what retrieval surfaces for real versus fictional entities, and reads metaphor as a set of domain-to-domain mappings rather than a single figurativity feature. A cross-seed control marks the boundary: across five networks differing only in initialization, the same distinction surfaces as almost entirely different tokens (top-10 overlap 0.08). What a computation looks like in token space is network-specific; the distinction it draws is not
Read the full abstract
Computation and Language
Deep and shallow biases in language models
Large language models often repeatedly select the same answer even when many alternatives are plausible. Prior work treats this concentration as bias, but it does not distinguish stable model preferences from responses that depend on a particular prompt wording. We introduce a bias depth score that measures both how strongly a model prefers its top answer under direct prompting and whether that answer survives scenario reframing. Across 4,442 opinion prompts and four large language models, only about a quarter of the concentrated preferences survive reframing. We call these persistent cases Deep biases, and the remaining prompt-dependent cases Shallow biases. Our results show that Deep biases are more often inherited from pretraining and preserved through SFT. Under both continued fine-tuning and prompt-based debiasing for diversity, Deep biases are consistently harder to remove than Shallow biases. Bias depth therefore separates stable learned biases from prompt-wording artifacts that single-prompt metrics conflate. Code, models, and data are available at deepbias.github.io.
Read the full abstract
Analysis of PDEs
On the rotation-two-component $b$-family shallow-water system: derivation and wave-breaking analysis
In this paper, we derive the rotation-two-component $b$-family shallow-water wave system from the Green–Naghdi type equations incorporating the effects of the Coriolis force and constant vorticity. The system admits, among its special cases, the two-component Degasperis–Procesi (2DP) system as a reduction, thereby confirming its genuine role as a rotational extension of the $b$-family. We then focus on a specific reduction, the rotation-two-component Degasperis-Procesi system, whose particular interest lies in the fact that it combines two independent sources of difficulty: unlike the two-component Camassa-Holm type system, it possesses no conserved $H^1\times L^2$-type quantity, and in contrast to the standard 2DP system, it contains a higher-order nonlinear coupling term induced by the Coriolis force. To overcome this double difficulty, we introduce an auxiliary variable and establish a local $L^\infty$-estimate for $u$ without using any conservation law, based on which we derive two distinct blow-up criteria. When the Coriolis force vanishes, our results recover known criteria for the 2DP system and also provide new criteria for both the 2DP system and Degasperis-Procesi equation. The results extend naturally to the periodic setting and to relevant special cases of the rotational $b$-family system.
Read the full abstract
Machine Learning
Strangers to Themselves: What Language Models Say About Themselves Is Generic
Language models can fluently describe how they would behave: whether they would cave to pushback, misuse a tool, or lie under pressure. Is that description actually about the model speaking? We turn self-knowledge into a prediction test. Across nine behavioral evaluations, we measure how a model behaves under different conditions, ask it to predict those rates, and compare its predictions with controls that remove the self from the question. We find that: (i) Direct self-report is weak (r = +0.04), and even showing the model the exact items only raises prediction to +0.24. Crucially, the same item-informed question about "capable AI agents in general" does just as well (+0.28), while other models' answers about themselves predict the target model at least as well as its own. (ii) Frontier scale does not detectably change this pattern: any gains in prediction are not self-specific, and are consistent with a better theory of how AI assistants behave rather than better self-knowledge. (iii) First-person framing does have one robust effect: it shifts reports in the flattering direction, understating harmful behavior relative to the same question about a generic agent. (iv) Finetuning on a model's own behavioral record can teach narrow self-predictions, but it also changes the behavior being predicted and the gains do not transfer broadly. The practical implication is simple: asking a model what it would do mostly reveals a theory of AI assistants in general, plus a favorable bias, rather than privileged knowledge of that model.
Read the full abstract
Artificial Intelligence
Grounded Evaluation and Repair for NL-to-PDDL Problem Generation
Large Language Models (LLMs) have shown promise for translating Natural Language (NL) planning descriptions into PDDL problem instances. However, standard evaluation criteria such as syntactic validity or planner success can substantially overestimate faithfulness to the described task: a generated problem may be parseable and solvable while misrepresenting the intended initial state, goal, object structure, or optimization target. This paper studies an end-to-end NL-to-PDDL pipeline that combines LLM generation, checks in terms of PDDL parsing, planning and validation, a domain-conformance checker, an LLM critic, and iterative repair. Fine-grained repair feedback is constructed from the domain description, the generated problem, the natural language problem description, and operational diagnostics. Reference-based comparisons against curated benchmark PDDL problem descriptions are used for post-hoc benchmark analysis, and these offline checks include renaming-invariant structural matching and semantic equivalence, where domain support is available. Across Planetarium, AutoPlanBench, and curated PDDL~2.1 problems, results show that operational success and benchmark-reference reconstruction can diverge substantially. Results also show that structured repair can be useful, and that PDDL~2.1 remains challenging for reference reconstruction, even when operational success improves.
Read the full abstract
Astrophysics of Galaxies
The Dense Gas Structures Around MMS 2/OMC-3 Traced by C$^{18}$O Emission
We report the Atacama Large Millimeter/submillimeter Array observations of 1.3 mm continuum, C$^{18}$O ($J=2$$-$$1$), and N$_{2}$D$^{+}$ ($J=3$$-$$2$) toward the millimeter multi-system MMS 2 in the Orion Molecular Cloud-3 region, at an angular resolution of 1.59" ($\sim$620 au). MMS 2 is in the flat-spectrum phase, with its protostars in the late stage of the main accretion phase. We detect the centrally condensed structures traced by C$^{18}$O emission associated with the 1.3 mm continuum sources MMS 2-North and MMS 2-South, which are spatially resolved and unresolved, respectively. The estimated diameter and mass of the centrally condensed structure associated with MMS 2-North are 1.64" ($\sim$640 au) and 3.0$\times$10$^{-4}$M$_{\odot}$, respectively. We also detect an extended structure traced by C$^{18}$O emission associated the circumbinary envelope at scale of 5.70" ($\sim$2240 au) with an estimated gas mass of 1.4$\times$10$^{-2}$$M_{\odot}$. In addition, we detect a filamentary structure traced by N$_2$D$^+$ emission to the south of MMS 2, which is spatially offset from the C$^{18}$O emission. This offset may be attributed to the influence of the warm surrounding environment.
Read the full abstract
Combinatorics
Metric Bases of Barycentric and Matching Subdivisions of Zero-Divisor Graphs
In this paper, we study metric bases and related metric properties for barycentric and partial matching subdivisions of the zero-divisor graph of $\mathbb Z_{pq}$, where $p$ and $q$ are distinct odd primes with $q>p$. We first recall the natural partition of the zero-divisor graph into the two prime classes and then give a detailed characterization of those subsets of $BS(\Gamma(\mathbb Z_{pq}))$ that form metric bases when $q\geq 2p-1$. The proof is expanded by separating the role of closed neighborhoods, rows of subdivision vertices, and forbidden twin configurations. We then investigate $M$-subdivision graphs obtained by subdividing selected edges of $\Gamma(\mathbb Z_{pq})$. In addition to the lower bounds for subdivisions of $p-3$ and $p-2$ edges, we prove an exact formula for matching subdivisions of arbitrary size $r$, $0\leq r\leq p-2$, namely $\dim(G_r)=p+q-r-4$. Several consequences are included to illustrate how a small matching subdivision can reduce the localization cost of the original zero-divisor network.
Read the full abstract
Computer Vision
Can We Trust Video Hallucination Detectors? VidHalLoc for Evaluating the Evaluators
Video-language models and video agents can produce hallucinations that conflict with spatiotemporal evidence. Existing benchmarks mainly evaluate model hallucinations, and heterogeneous mechanisms make detector reliability difficult to compare. We introduce VidHalLoc, a benchmark that evaluates hallucination detection methods under a unified diagnostic evaluation protocol using 2,000 adversarial hallucination samples across Video Question Answering and Video Captioning tasks, spanning Ontology and Dynamic hallucination categories. To construct VidHalLoc efficiently, we introduce VideoHALO, a Harness Engineering-informed multi-agent workflow that decomposes data construction into four executable stages supported by a memory system and a communication protocol. Evaluation of fifteen methods reveals that the four dedicated detectors peak at an Overall accuracy of only 34.63%, indicating limited reliability across video hallucination types [Dataset Repository: https://huggingface.co/datasets/wesfggfd/VidHalLoc].
Read the full abstract
Analysis of PDEs
Positive solutions for an indefinite-weight Minkowski mean curvature Neumann problem: multiplicity and asymptotic behaviour
We study the Neumann boundary value problem $$ -\operatorname{div}\left(\frac{\nabla u}{\sqrt{1-|\nabla u|^2}}\right) = \lambda a(x)g(u) \quad \text{in } \Omega, \qquad \frac{\nabla u}{\sqrt{1-|\nabla u|^2}}\cdot \mathbf n = 0 \quad \text{on } \partial\Omega, $$ where $\Omega \subset \mathbb R^N$ is a bounded convex domain, $a\in L^\infty(\Omega)$ is an indefinite weight with $\int_\Omega a(x) \, dx< 0$, and $g$ is a nonlinearity. Under suitable assumptions on $g$, we prove the existence of two positive solutions for sufficiently large $\lambda>0$, using Szulkin's theory for nonsmooth functionals: a global minimizer $u_\lambda^{(l)}$ with negative energy, and a mountain-pass critical point $u_\lambda^{(s)}$ with positive energy. Furthermore, we study the asymptotic behaviour as $\lambda\to+\infty$ of both families of solutions. For the ground states we prove that, up to a subsequence, $u_\lambda^{(l)}\to u_\infty$ uniformly, where $u_\infty$ satisfies $\|\nabla u_\infty\|_{L^\infty(\Omega)}=1$ and is characterized by being a maximizer of $\int_\Omega a(x) G(u) \, dx$ and a minimizer of the kinetic energy among all such maximizers. In the model case $g(u)=|u|^{p-2}u$, $p>2$, we also show that the mountain-pass energy level satisfies the upper bound $c_\lambda=O(\lambda^{-2/(p-2)})$, and that $u_\lambda^{(s)}\to0$ in $C(\overline\Omega)$.
Read the full abstract
Earth and Planetary Astrophysics
JWST-SUPER I: New insights into irradiated warm Neptunes atmospheres from MIRI observations of HD 106315 c
Sulphur-bearing molecules have recently emerged as powerful tracers of atmospheric photochemistry in exoplanets observed with JWST. In several warm giant planets, SO$_2$ has been detected as a product of UV-driven chemical processing, suggesting a close connection between stellar irradiation, atmospheric metallicity, and sulphur chemistry. Whether these trends extend to Neptune-mass planets orbiting hotter stars remains largely unexplored. We investigate the atmospheric composition of the warm Neptune HD 106315 c, a Neptune-mass planet orbiting an F-type host star and subjected to a strong ultraviolet (UV) irradiation environment. Here, we report the low-resolution transmission spectrum between 5 and 12 $\mu$m of HD 106315 c obtained with the MIRI Low-Resolution Spectrometer on-board JWST. Our work also includes re-analysis of archival data from HST WFC3 G141 mode of HD 106315 c, as well as contemporaneous XMM and Swift monitoring of the star in the UV and X-ray wavelength ranges. Together with the archival HST WFC3 data, we detect H$_2$O and find tentative evidence for NH$_3$, retrieving abundances of $\log_{10}$(H$_2$O)$=-1.40^{+0.40}_{-0.74}$ and $\log_{10}$(NH$_3$)$=-2.47^{+0.57}_{-0.89}$, together with an isothermal terminator temperature of $747^{+150}_{-155}$ K. We place stringent upper limits on CH$_4$, SO$_2$, and CS$_2$, finding no robust evidence for any sulphur-bearing species despite the intense irradiation received by the planet. The inferred water abundance implies a strongly metal-enriched atmosphere. Elevated intrinsic temperatures can reconcile the non-detections of CH$_4$ and CS$_2$ through carbon–sulphur coupling, while the absence of SO$_2$ points toward a reduced atmospheric sulphur inventory or a near-solar to mildly enhanced C/O ratio, and places HD 106315 c near the transition between sulphur-rich and sulphur-poor chemical regimes.
Read the full abstract
Probability
Comparison inequalities for discrete- and continuous-time infection processes
We study comparison inequalities for contact-process-type infection dynamics on $\mathbb Z^d$ with general local transmission mechanisms. The processes considered include both discrete-time oriented-percolation models and continuous-time contact processes in which infection events may transmit to random, possibly correlated sets of neighbouring sites. Our main results apply local-to-global comparison criteria beyond stochastic domination: if one local infection rule is more likely than another to hit every non-empty test set of neighbours, then the corresponding process has larger survival probabilities at all times. We compare classical independent-infection models with exchangeable fixed-budget, all-or-nothing, and burst-type infection mechanisms, deriving orderings for example of finite-time and infinite-time survival probabilities and stopping-time distributions. In continuous time, we further obtain consequences for critical survival thresholds of exchangeable infection laws. The results show that infection processes with identical or comparable mean-field infection intensity can nevertheless be rigorously ordered at the stochastic level once spatial geometry is taken into account. Additionally, it turns out that the classical continuous-time contact process dominates a variety of related continuous-time models, whereas its most natural discrete-time version, Bernoulli oriented percolation, is not dominant with respect to related discrete-time models.
Read the full abstract
Machine Learning
ProMeta: Few-shot PROTAC-targeted degradation prediction across E3 ligases
Proteolysis-targeting chimeras (PROTACs) have emerged as a transformative therapeutic strategy that selectively degrades historically ”undruggable” targets via the ubiquitin-proteasome system. Despite growing efforts to develop computational predictors of PROTAC degradation activity, existing supervised approaches remain severely challenged by data scarcity and imbalance across E3 ligases, limiting their ability to generalize beyond well-studied ligase contexts. In practice, labeled data are heavily concentrated on a few ligases (e.g., CRBN and VHL), while the majority of E3 ligases remain underexplored yet are critical for expanding the design space of targeted degraders. Developing methods that enable robust cross-ligase generalization with minimal labeled data is therefore essential for improving the practical utility of computational PROTAC discovery. We reformulate PROTAC degradation activity prediction across E3 ligases as a few-shot meta-learning problem and present ProMeta, a prototype-based graph neural network trained through episodic meta-learning on source-E3 tasks and evaluated on held-out target-E3 tasks through support-conditioned inference. ProMeta performs inference without updating the encoder by dynamically estimating class prototypes from minimal target-ligase support samples. On the CRBN-to-VHL benchmark, ProMeta achieves AUROC values of 0.796 under K=2, Q=3 and 0.883 under K=2, Q=5, improving by 19.9% and 6.8%, respectively, over the corresponding supervised GNN baseline. Reverse VHL-to-CRBN transfer under the same protocol yielded AUROC values of 0.702 (K=2, Q=3) and 0.821 (K=2, Q=5), confirming bidirectional applicability while revealing direction and data-regime dependence. Together, these results support ProMeta as a practical framework for cross-ligase few-shot prediction under the evaluated support/query protocols.
Read the full abstract
Computer Vision
StreetDiff: Multi-view Street Scenes Generation via Cross-view Consistent Multi-view Stable Diffusion with Structure Prompts
Multi-view diffusion models have shown strong performance in scenes with strong geometric priors and sparse semantics, such as indoor rooms or simple outdoor environments (e.g., fields, courtyards). However, they often fail to maintain cross-view consistency under camera rotation, especially in structurally complex urban environments. Without explicit modeling of spherical correspondence across views, existing approaches tend to produce object duplication, structural distortion, and layout inconsistency. To address this limitation, we propose StreetDiff, a multi-view diffusion framework that explicitly enforces cross-view alignment during denoising. StreetDiff introduces a Panorama–Perspective Synergy design to decouple global layout reasoning from local detail synthesis, and incorporates a Panorama Alignment Module (PAM) that establishes spherical-projection-based attention constraints across views. By injecting structured alignment constraints without modifying the diffusion backbone, our framework achieves robust cross-view coherence in challenging urban street scene generation tasks. In addition, we construct Street360, a large-scale HDR multi-view urban panorama dataset. Extensive experiments demonstrate that StreetDiff significantly improves structural consistency and visual fidelity compared to prior multi-view diffusion generation methods.
Read the full abstract
Computation and Language
Leveraging Fine-grained Error Correction in Korean Speech Recognition for Consultation Services
Automatic Speech Recognition (ASR) technology is fundamental to customer service automation and large-scale transcription. However, even advanced ASR models exhibit inevitable errors in complex real-world environments such as call center conversations. When privacy restrictions preclude audio access, error correction must rely on text-based post-editing. Existing text-only approaches face significant challenges in low-resource languages, mainly due to a critical scarcity of annotated corpora and tailored correction methodologies. For Korean, this resource gap is particularly pronounced, as existing resources are predominantly designed for ASR training rather than text-based error correction. To address this, we introduce DasanCallDial, the first large-scale Korean benchmark dataset specifically curated for dialogue-level ASR error correction. Derived from genuine call center interactions, it comprises 1,974 dialogues with 115,460 utterances. Leveraging this resource, we propose Detector-Gated Contextual Span Correction (DCSC), a text-only post-editing framework for error-sparse Korean speech recognition transcripts. DCSC combines an encoder-based detector that first performs token-level error detection, followed by a language model-based corrector trained to rectify fine-grained span-level errors. Additionally, we employ dialogue-level context augmentation to enable the model to leverage discourse history for disambiguation. By employing multi-level granularity, our method achieves state-of-the-art performance, effectively overcoming the limitations of general LLMs in low-resource settings.
Read the full abstract
Theoretical Economics
The Response of Consumption to Interest Rates with Borrowing Constraints: An Analytical Approach
I derive an explicit mapping from initial assets, income and the real interest rate to consumption for an income fluctuation problem with a borrowing constraint and CARA utility. I show that there exists a threshold of initial wealth over which the partial equilibrium consumption response to a permanent increase in the real interest rate is positive, consistent with recent empirical evidence. I further show that precautionary savings reinforce the possibility of crowding in and that the possibility of a positive response extends to a more standard CRRA utility whenever the elasticity of intertemporal substitution is strictly less than 1.
Read the full abstract
Computation and Language
When Does Defendant Statement Matter? A Study of Bias and Persuasion in LLM-Simulated Jurors
LLMs have been used to simulate human decision-making in professional settings, yet their behaviors in common-law jury trials remain unexplored. We study when and how a defendant's courtroom statement affects LLM-simulated jurors, focusing on persuasion, ideological bias, and background-based affinity. To support the analysis, we introduce JuryBench, a benchmark containing controversial criminal cases in U.S. criminal law. In each case, a defendant can claim various plausible justifications to support acquittal or reduced liability. We fix the base case and design defendants of different backgrounds, who give courtroom statements with varying emotional appeal or rebuttal. Jurors with diverse ideological profiles across the spectrum are simulated. We examine 20 frontier LLMs, resulting in a total of 432K decisions and rationales, and quantify changes in verdict severity. Our findings show that LLM-jury simulation echoes many human-jury findings. First, emotional persuasion can be detrimental, since jurors may perceive it as evidence of guilt or inconsistency. Next, we show that background fit between jurors and defendants is a stronger and significant factor than other isolated factors, and that jurors are in general harsher toward opposite-background defendants and lenient toward same-background ones. Finally, we find that juror ideology also strongly shapes severity judgments. These findings highlight both the promise and risks of using LLMs to model jury reasoning and call for careful evaluation. The data and code are available at https://github.com/choyingw/JuryBench
Read the full abstract
Instrumentation and Methods for Astrophysics
First Very Long Baseline Interferometry Fringe Detection at 690GHz
We report the first very long baseline interferometry (VLBI) experiment conducted in the 690 GHz atmospheric window. On 2024 November 21, observations with the Atacama Large Millimeter/submillimeter Array, the Atacama Pathfinder EXperiment (APEX), and the James Clerk Maxwell Telescope (JCMT) were carried out using ALMA's newly developed Band 9 phasing capability. Fringes were detected on the ALMA-APEX baseline during a scan of the quasar J0423-0120, with a signal-to-noise ratio of ~12 and useful fringe recovery over solution intervals of order tens of seconds, representing the highest-frequency ground-based VLBI fringe detection reported to date. No fringes were found on the ALMA-JCMT baseline, despite excellent weather conditions, consistent with sensitivity predictions and supporting baseline performance models. The ALMA Phasing System maintained stable phasing at Band 9 for ~1-2 minutes before gradually degrading, indicating limitations under these observing conditions. Our analysis shows that, under excellent weather conditions, 690 GHz VLBI can still support fringe recovery over short solution intervals, despite rapid atmospheric phase fluctuations at these frequencies. This work validated key elements of near-terahertz VLBI operation and establishes a technical foundation for routine observations in the 690 GHz atmospheric window.
Read the full abstract
Artificial Intelligence
Decision Transformer for UAV-Mounted RIS-Assisted Dynamic D2D Communications
This paper studies unmanned aerial vehicle (UAV)-mouted reconfigurable intelligent surface (RIS)-assisted device-to-device (D2D) communication with stochastic link activation. It models UAV motion and attitude, time-varying Rician angles, and angle-dependent RIS reflection. A joint optimization of UAV trajectory, attitude, and RIS phases is formulated to maximize average sum rate under mobility, energy, and hardware constraints. The problem is addressed using deep reinforcement learning and a Decision Transformer trained on expert trajectories from multiple scenarios. Results demonstrate effective cross-scenario generalization, with zero-shot transfer outperforming direct DRL transfer and online fine-tuning achieving competitive performance with fewer interactions.