Artificial Intelligence
Papers
Fresh from arXiv, typeset properly. Highlighted words trade on the market.
All of arXiv
Friday 28 August
Extreme low-bit inference offers a route toward smaller models and constrained deployment. Ternary language models restrict weights to $\{-1,0,+1\}$, approaching the limit of $\log_2 3 \approx 1.585$ bits/weight. The practical question for a pretrained model is not simply whether weights can be quantised but which capabilities survive and whether it remains useful for adaptation. We explore this by converting Qwen3.5-0.8B (752M parameters) to ternary weights using 72.4M tokens of quantisation-aware training (QAT). The resulting model, Cloe, is evaluated across 29 benchmarks, representation diagnostics, and downstream fine-tuning. The evidence shows non-uniform degradation. A linear probe recovers 43.76% of MMLU answers from the full-precision teacher's representations but only 26.19% from Cloe (near chance), indicating specialist factual information is lost. However, Cloe retains measurable performance on ten tasks, averaging 77.1% of teacher performance. Crucially, fine-tuning raises Cloe to 89.8% on SST-2 (95.6% of the matched teacher) and reaches 79.4% teacher retention on XSum. We attribute degradation to a combination of quantisation-induced information loss and incomplete recovery due to the limited QAT budget. We also highlight an evaluation pitfall: standard answer-letter scoring failed (Cloe emitted "A" on 98.6% of MMLU questions), necessitating continuation scoring. Ultimately, ternary conversion is unsuitable as a drop-in general replacement yet remains valuable as a compact substrate for task-specific models.
Quantum Physics
Quantum synchronization of spin-1 system: Enhanced synchronization due to additive Lindblad operators
We experimentally investigate quantum synchronization for one of the simplest possible quantum systems, namely an externally driven few-level system with equally spaced energy levels effectively acting as a spin-1 system. Coupling to excited auxiliary states, we realize additive effective Lindblad operators that are associated with non-conventional dissipative pathways, which are shown to enhance, in certain parameter regimes, quantum synchronization. The experimental set-up, which utilizes cold $^{87}$Rb atoms in a MOT, and associated synchronization extraction protocol are benchmarked carefully through dedicated simulations. Convincing agreement is found between experiment and simulations. The dissipation engineering approach established in our work can be readily extended to systems with more energy levels, such as effective spin-$3/2$ or spin-$2$ systems, and has implications for quantum synchronization studies in higher-spin systems as well as for a wide range of quantum science studies and technology applications.
Read the full abstract
Classical Analysis and ODEs
On the measure of the boundary of a Hajłasz-Sobolev $(M^{1,p},M^{1,q})$-extension domain
We prove that the boundary of every bounded $(M^{1,p},M^{1,q})$-extension domain in a $Q$-doubling metric measure space has measure zero whenever $0<q<p<\infty$, with the additional restriction that $p<qQ/(Q-q)$ if $q<Q$.
Artificial Intelligence
Enhancing SAE-based Steering via Neighbor Integrated Feature Selection
Sparse autoencoders (SAEs) disentangle model activations into interpretable features and are widely used for steering large language models. Most existing SAE-based steering methods select features by applying a top- filter based on statistical scores, assuming that higher-scoring features yield stronger steering effects. In this paper, we show that this assumption is often invalid, leading to suboptimal feature selection. Our analysis reveals that effective steering features may be distributed among representationally adjacent, semantically similar groups induced by feature splitting in SAEs. Within such groups, features may exhibit disparate statistical scores despite having comparable steering influence, causing score-based selection to overlook important features. Based on these observations, we propose Neighbor Integrated Feature Selection (NIFS), a plug-and-play strategy that leverages representation similarity to improve feature selection for steering. We evaluate NIFS across multiple SAE-based steering methods and tasks, and demonstrate consistent performance gains over conventional top-$k$ selection.
Read the full abstract
Other Quantitative Biology
Geochemical Hazard Assessment of Martian Regolith for Future Human Exploration
As human missions to Mars move from concept to planning to reality, a systematic quantitative health risk assessment of martian regolith exposure has become critically important. This study presents a comprehensive multi-element, multi-pathway health hazard analysis of martian regolith for a 70 kg adult astronaut on an 18-month surface mission, using bulk silicate Mars geochemical data and Earth Upper Continental Crust reference values as baseline comparators. We computed Average Daily Dose, Hazard Quotient, Hazard Index, Enrichment Factor, Ecological Risk Factor, and Incremental Lifetime Cancer Risk across all three exposure pathways (oral ingestion, inhalation, and dermal contact) for toxic and heavy metals, under four filtration scenarios (0%, 25%, 50%, 95%). Results demonstrate that Cr and Co represent the most critical non-carcinogenic hazards, exceeding the regulatory threshold (HI > 1) even at 50% filtration, while Cr also poses the most significant carcinogenic risk, near the $10^{-4}$ regulatory threshold level. At least 92% regolith filtration efficiency is required to reduce Cr to acceptable non-cancer hazard levels. Nickel exceeds the acceptable cancer risk threshold of $10^{-6}$ at all filtration levels below 95%. No element other than Cr and Co exceeds HI = 1 in the unfiltered scenario, although Fe, Ni and Mn approach concerning levels. These results establish that a minimum 95% High-Efficiency Particulate Air (HEPA)-grade filtration efficiency combined with active chemical sorption is required for acceptable Cr and Co short term exposure and Cr, Ni long term exposure management. This study provides a pathway-specific quantitative risk framework applicable to habitat air quality standards and EVA suit specifications for Mars surface operations
Read the full abstract
Instrumentation and Methods for Astrophysics
X-ray grating spectroscopy as a mission enhancement
We propose to add instruments to any potential future X-ray mission with focussing optics that is considered in NASA's ASTRA framework. Such an instrument is a necessity to study AGN wind outflows and feedback, find the missing baryons, study the intergalactic medium, and analyze abundances and chemical bonds in dust grains throughout the Milky Way. We conclude that those science goals can be achieved with a spectral resolving power > 3000 in the soft X-ray band (about 10-40 Ang) and an effective area a few times larger than current instruments. We describe a possible mission implementation for a soft X-ray grating spectrometer that can be folded in and out or be mounted permanently in the beam. Such an instrument can reach the requirements for a wide variety of host mission properties. A small UV imager and a UV spectrograph can be mounted on the same platform with independent optics. These added instruments vastly enhance the science capabilities of the host mission for a modest cost (100-200 million $) and with weight and power needs that can be easily accommodated in any major mission.
Read the full abstract
Computational Engineering
Pragmatic Information, Computation, and the Efficient Market Hypothesis
The efficient market hypothesis, that prices reflect all available information, imputes a meaning to market moving information, a view of information foreign to standard information theory. Here, after reviewing properties that make a proposed formula for “pragmatic information" a plausible measure of meaning, we consider the role that the receiver's position in the machine hierarchy corresponding to the Chomsky hierarchy play in extracting this meaning. We show that a receiver at a given level in the hierarchy may be unable to extract pragmatic information from a message because it appears random, yet a receiver at a higher level in the hierarchy finds the message perfectly intelligible. Also, the maximum processing rates for messages of different levels of hierarchy serve as a kind of channel capacity, leading to a tradeoff between the amount of pragmatic information extracted and the extraction time. All of the above suggests a recasting of market efficiency in terms of “computational efficiency”, i.e. the question of whether a market appears efficient to a participant with a given computational endowment. Successful trading strategies implementable as finite state machines, the lowest level in the hierarchy, imply departures from “finite state efficiency”. We show, via a stylized example, how pragmatic information can characterize the ensuing approach to finite state efficiency. We also show that actual market dynamics can be more computationally intractable than any finite state machine can process by proving the PSPACE-completeness of the processing of the smart order routing systems of major brokerage firms. We conclude that computational efficiency is the norm and pragmatic information can characterize any departure from it.
Read the full abstract
Computer Vision
A Large-scale Evaluation of Text-guided Models for Facial Editing
Facial appearance editing powers popular applications like FaceApp and Photoshop. Generative Adversarial Networks (GANs) and 3D Morphable Models (3DMMs) have been widely used for facial editing. GANs can perform varied facial edits (e.g., changing hair color, hairstyle), but often produce unstable edits. 3DMMs produce stable edits, but can only alter pose and facial expression. Recently, text-guided diffusion models like Nano Banana have become popular for image editing. Text-guided models are a compelling alternative to GANs and 3DMMs since they can produce both stable and varied image edits. While text-guided models have been widely tested for whole-scene edits (e.g., “make the woman play a guitar”), they have not been comprehensively tested for facial editing. We conducted the first large-scale evaluation ($\sim1$M images evaluated) of six popular text-guided models on a sequential facial editing task. We present Face-Edit-Attributes, the largest collection of $169$ facial editing attributes focused on hair, accessories, and pose edits. We compared model performance using two popular celebrity face datasets: CelebA and CelebSET. Our results show that most models performed hair and accessory edits well, but struggled with editing pose. All models over-edit (e.g., changing hair color when asked only to change the hairstyle). We also evaluated demographic biases in each model. Our results show surprising biases in overediting: almost all models created more overedits for dark-skinned male faces and old faces. The code and data for our results (including our repository of $\sim 1$M images) can be accessed \href{https://github.com/rahul1801/Face-Edit-Bench}{\textcolor{blue}{here}}.
Read the full abstract
Spectral Theory
Complex Analysis in completeness, spectral and scattering problems
These lectures are aimed at junior researchers seeking to enter the field of applications of complex analysis to harmonic analysis, Fourier analysis, spectral and scattering theory for differential operators and related fields. Starting with classical completeness problems posed by Bernstein, Beurling and Malliavin, Wiener, Kolmogorov and Krein, the course progresses to the solutions of the Gap and Type Problems, inverse spectral problems for Schrödinger operators and canonical Hamiltonian systems, and pointwise convergence of the non-linear Fourier transform. Each lecture is supplemented with exercises and references.
Read the full abstract
Software Engineering
UML Class Diagram Evaluation and Repair Strategies based on LLMs
UML class diagrams are a crucial tool for defining the structure of software systems, but designing accurate and comprehensive class diagrams is a challenging task. Traditionally, creating UML models relies on the expertise and experience of professionals. However, with the development of AI technologies, particularly LLMs, new opportunities for software modeling have emerged. Despite this, there has been limited research on the application of LLMs in software modeling, especially in UML class diagram modeling. This study conducts experiments on several typical software system cases. Combining SDMetrics with expert manual review, this paper comprehensively evaluates the practical performance of mainstream LLMs in UML class diagram modeling from multiple dimensions, including size and completeness, relationship correctness, inheritance hierarchy, and design rule compliance. Focusing on typical defects in LLM-generated UML class diagrams, this study reveals that LLMs exhibit uncertainties analogous to human memory. Accordingly, three targeted repair strategies are proposed, including memory reinforcement, external knowledge injection, and detection-guided automated targeted repair. Experimental results obtained from the case studies indicate that (1) compared to expert-crafted class diagrams, LLM-generated UML class diagrams exhibit several issues, such as incomplete identification of key classes, confusion or omissions in relationships, insufficient or absent inheritance relationships, unused classes, and circular dependencies, and (2) after applying the repair methods, all the LLMs show varying degrees of improvement in addressing these issues. The average repair rate for key class identification reaches 85%, the coupling relationship repair rate is 46%, the inheritance relationship repair rate is 69%, while repair rates for unused classes and circular dependencies both reach 100%.
Read the full abstract
Numerical Analysis
Separable Nonnegative Matrix Factorization Using Powered Ratio-of-Norms Regularization
Separable nonnegative matrix factorization (SNMF) has been widely used for low-rank representation and clustering of nonnegative data, owing to its ability to produce part-based and interpretable decompositions. In particular, SNMF is closely related to graph clustering and community detection. To enhance sparsity and identifiability of the learned factors, we propose an $\ell_1^p/\ell_2$-regularized SNMF model based on a powered ratio-of-norms regularizer. The resulting formulation is nonconvex and nonsmooth, which poses significant challenges for optimization. To address this, we develop efficient algorithms based on the difference-of-convex function algorithm (DCA) and the alternating direction method of multipliers (ADMM). The proposed methods decompose the original problem into tractable subproblems, leveraging closed-form proximal operators associated with the powered norm terms. We establish descent and limiting criticality properties for the DCA scheme and convergence under standard assumptions for the ADMM scheme. Extensive numerical experiments on synthetic datasets and hand gesture classification tasks demonstrate that the proposed approach achieves competitive or improved performance in anchor identification and classification accuracy compared with existing SNMF methods, while maintaining competitive computational efficiency.
Read the full abstract
Computer Vision
Blind Stereoscopic Omnidirectional Image Quality Assessment Using Predictive Coding Hierarchy
Stereoscopic omnidirectional images (SOIs) have provided users with newly immersive quality of experience in virtual reality environments. However, developing efficient and accurate perceptual quality assessment metrics for SOIs remains challenging due to many factors such as freely changeable field of views and binocular vision. In this paper, based on the characteristics of the human visual system (HVS), we propose a Predictive Coding Hierarchy-inspired metric (PCH) for blind/no-reference stereoscopic omnidirectional image quality assessment. Motivated by the viewing process of SOIs, the proposed PCH includes a local cyclopean perception module, a global predictive perception module, and a visual quality regressor. First, observers browse different spherical sceneries from viewports, and aggregate the local visual information to infer the perceptual quality of SOIs. Therefore, we extract various viewports, followed by cyclopean conversion and saliency detection to approach the perception and attention of the human brain. After the local aggregation, viewers then infer the global scene in their minds. Based on the binocular mechanism, we fuse left and right views to perform predictive coding hierarchy modelling. Finally, the visual quality regressor is exploited to obtain the ultimate quality score related to both local and global perceptual cues. Extensive experiments demonstrate that the proposed PCH achieves competitive and consistently improved performance compared with state-of-the-art quality assessment methods.
Read the full abstract
Statistical Methodology
Nonparametric Efficient Estimation of Dynamic Treatment Regimes with Competing Risks
Many observational studies evaluate the risks and benefits of time-varying interventions. In these longitudinal settings, the primary outcome of interest is often a clinical event that is precluded by mortality, which acts as a competing risk. Evaluating dynamic treatment regimes (DTRs) is complicated by time-varying confounding, right-censoring, and progressive sample size reduction over time, for which traditional methods such as the g-formula or inverse probability weighting may be misspecified or highly unstable. In this work, we develop the nonparametric efficiency theory and derive the efficient influence function for the cumulative incidence of an event under a DTR in the presence of a competing risk and right-censoring. Building on this result, we propose a sequentially doubly robust estimator that accommodates flexible machine learning for nuisance estimation while retaining root-n consistency and asymptotic normality. We illustrate the utility of our framework by estimating long-term cumulative incidence of clinical fractures under different bisphosphonate "drug holiday" regimes for osteoporosis.
Read the full abstract
Mathematical Physics
The $\mathbb{Z}_2$-Index of a Pair of Pure States and the Topology of Interacting 1D Superconductors
We define a relative $\mathbb{Z}_2$-index $\mathcal{N}(\omega_1,\omega_2)$ for locally-comparable, parity-invariant pure states on a unital $C^*$-algebra. We prove that the index is well-defined, multiplicative, invariant under parity-preserving automorphisms, and locally constant in the norm topology. For the one-dimensional self-dual CAR algebra, we apply this construction to the half-chain automorphism $\sigma$ and define the many-body Majorana number $\mathcal{N}\sigma(\omega):=\mathcal{N}(\omega,\omega\circ\sigma)$ for parity-invariant $\sigma$-local pure states. In the quasi-free Hilbert–Schmidt regime, this index agrees with the usual single-body Majorana number. We then introduce symmetric local automorphism paths and prove that $\mathcal{N}_\sigma$ completely classifies the resulting automorphic-path-components. This automorphic-path equivalence retains the bulk Majorana number but may forget a relative zero-dimensional parity obstruction.
Read the full abstract
Software Engineering
The reach of a verification tool decides its value: A controlled study of verification surface, artifact quality, and cost in AI coding agents
Modern artificial-intelligence coding agents can be equipped with tools for checking their own work e.g. a linter, a boot probe, a shell, a screenshot tool. We call this set the agent's verification surface. This study asks whether increasing only that surface, with everything else held fixed, produces a matching growth in the quality of the software the agent ships. We built a minimal coding agent whose tool list is the single controlled variable and used it to implement 1,116 web applications across six models and eight tool configurations. A condition-blind human graded every application against a frozen rubric, and automatic probes stress-tested the API-observable behaviors. Verification's cheapest benefit arrives first, which is to make sure that the application comes up. Without any tools, about one build in seven fails to launch at all and a single boot probe removes nearly all of these failures at roughly 35 percent of a full shell's token cost, while the full shell multiplies the no-tools cost by 2.35. Screenshots help most where mistakes are visible (e.g. element placement, interaction), though even there the gain over a shell is modest and does not survive correction for multiple statistical comparisons. In cases where failures can only be measured rather than seen, such as keeping scrolling smooth over a 100,000-row list, screenshots add nothing. A verification tool improves the output artifact only where its reach covers the way the application actually fails.
Read the full abstract
Cryptography and Security
Breaking Darknet CAPTCHAs with general purpose LLM
Our work evaluates the effectiveness of automated methods for solving CAPTCHA challenges commonly encountered in darknet environments. These CAPTCHAs are typically designed to operate without JavaScript, resulting in distinct characteristics compared to mainstream CAPTCHA systems. Our study considers three representative challenge types: open-circle localization, rotation-based alignment, and object-selection CAPTCHAs. The experiments reveal a systematic limitation of contemporary MLLMs: while they are generally capable of identifying relevant visual structures, they frequently struggle with precise spatial localization and geometric transformations. These deficiencies can be mitigated either through task reformulation or by augmenting the models with specialized image processing tools. These deficiencies can be mitigated by task reformulation or by equipping the model with specialized image-processing tools. We therefore propose a hybrid framework in which an MLLM serves as a high-level reasoning and orchestration layer while delegating geometric computations to deterministic algorithms via the Model Context Protocol (MCP). The resulting system achieves success rates above 90% across all evaluated CAPTCHA types and demonstrates that combining the complementary strengths of MLLMs and classical computer vision yields a more accurate and efficient solver than either approach alone.
Read the full abstract
Differential Geometry
On the complete maximal maps and their singularities
This article investigates the global structure of maximal surfaces (space-like immersions with zero mean curvature) in Lorentz-Minkowski $3$-space $\mathbb{E}^3_1$, especially focusing on the interplay between genus, the number of singular components–loci, and simple ends. We construct complete maximal maps with arbitrarily many singular components for any genus $p\geq 0$ with simple ends.
Read the full abstract
Classical Analysis and ODEs
Mesh-Degree Rigidity for Positive Chebyshev-Fourier Approximants
Let $d_k>0$ and let $Q_k$ be a polynomial of degree $r_k$ with nonnegative Chebyshev coefficients. We study locally uniform limits of $f_k(z)=C_kQ_k(\cos(d_kz))$, where $C_k>0$. If $f_k\to F$ locally uniformly and $F(0)>0$, the associated positive lattice measures converge weakly, their second moments converge, and their quadratic tails are uniformly integrable. If additionally all zeros of $Q_k$ lie in $[-1,1)$, $d_k\to0$, and $F$ has order below two, then for every $A>0$ with $F(A)F(-A)\ne0$, $\limsup_{k\to\infty} r_kd_k^2\le 4\sum_{\gamma_n>A}\gamma_n^{-2}$, where $\{\pm\gamma_n\}$ are the nonzero real zeros of $F$, counted with multiplicity. If, moreover, $f_k\to F$ locally uniformly with $F(0)>0$, all zeros of $Q_k$ lie in $[-1,1)$, and $F$ has no nonzero real period, has order below two, and is not of finite exponential type, then $d_k\to0$, $r_kd_k\to\infty$, and $r_kd_k^2\to0$. Equivalently, $d_k^{-1}=o(r_k)$ and $r_k=o(d_k^{-2})$. Boundary examples show the role of the hypotheses and the Gaussian boundary at order two.
Read the full abstract
Artificial Intelligence
Efficient Geothermal Well-Control Optimization via Diffusion-Surrogate Reinforcement Learning
Real-time decision-making for enhanced geothermal systems (EGS) is challenging because long-term production periods involve high-dimensional control spaces and a large number of time-consuming high-fidelity hydrothermal simulations. Reinforcement learning provides a natural framework for state-dependent sequential control, but direct policy training with numerical simulators is computationally expensive. To address this issue, we propose a diffusion-surrogate guided reinforcement learning framework for long-horizon EGS well-control optimization. The reservoir temperature and pressure fields are used as system states, while injection rates are selected as control actions. A learned surrogate environment is constructed using conditional diffusion models to predict the evolution of reservoir temperature and pressure fields and a separate reward model to estimate the corresponding economic return. The surrogate environment is then integrated with Proximal Policy Optimization (PPO) for efficient policy training. Experiments on a fractured EGS benchmark show that the diffusion surrogate can accurately reproduce reservoir-state evolution over multiple control stages. The resulting surrogate-assisted PPO policy achieves competitive well-control performance compared with direct simulator-based PPO and existing optimization methods, while substantially reducing the dependence on expensive high-fidelity simulations. These results demonstrate the potential of diffusion-based surrogate environments for efficient reinforcement learning in geothermal well-control optimization.
Read the full abstract
Multiagent Systems
ASTRA - Agentic System for Ticket Resolution and Analysis
Technical operations teams resolve large volumes of incidents by synthesizing fragmented evidence from ticket text, historical cases, system logs, and technical documentation. Existing automation often relies on monolithic generation without explicit evidence modeling or provenance, making outputs difficult to verify when critical signals are sparse across sources. We propose ASTRA, an agentic system for ticket resolution in which a central orchestrator coordinates three specialist information-gathering agents and drives a judge-orchestrator refinement loop to produce evidence-backed troubleshooting reports. TicketSimilarityAgent retrieves relevant historical precedents through dense retrieval and LLM reranking; LogAgent distills hundreds of thousands of log lines into structured, quote-grounded findings using deterministic filtering and constrained LLM analysis; and DomainKnowledgeAgent retrieves relevant technical knowledge via the Model Context Protocol (MCP). Their outputs are transformed into a claim-evidence representation linking each claim to a verbatim source passage, assigning a support level, and preventing cross-attribution. A JudgeAgent scores the report on five criteria, while the OrchestratorAgent converts low scores into targeted follow-up queries for bounded iterative refinement. Evaluated on 987 real-world telecom fault tickets across seven product lines, ASTRA achieves a mean quality score of 4.13/5.0, with 59.9% of reports identifying the fault area at the component-family level or better. Relevance and Clarity scores are 4.88 and 4.94, respectively, while fabricated technical details remain below 3% of error cases. Stratification by fault type reveals that hardware faults remain substantially harder than software or configuration faults (Cohen's d=0.80), pointing to a fundamental limitation of text-based evidence channels for hardware fault diagnosis.
Read the full abstract
General Relativity and Quantum Cosmology
Acceleration and Rotation in Three Dimensions
We present a new family of exact solutions representing accelerating, rotating, point particle and black hole solutions in three-dimensional Einstein gravity. We investigate their geometric properties and the global embedding in three dimensional anti-de Sitter space. Our results extend the catalogue of exact solutions in three-dimensional Einstein gravity and provide a useful setting for the study of black hole thermodynamics, asymptotic symmetries, and holography.
Read the full abstract
Quantitative Methods
FoldKit: A Python library for efficient storage and retrieval of co-folding predictions
AlphaFold 3 (AF3) enables structure prediction of biomolecular complexes through co-folding multiple interacting molecules, making it increasingly useful for de novo protein design and for large-scale studies of protein-protein, protein-peptide, and other biomolecular interactions. However, systematic co-folding experiments can produce large volumes of output data, particularly when multiple random seeds and samples are generated for each input complex. We introduce FoldKit, a Python package for efficient storage and analysis of large-scale AF3 co-folding results. FoldKit converts raw AF3 outputs into a compact, structured representation while preserving the metadata needed for downstream analysis. The FoldKit Python library provides convenient programmatic access to global, single chain, and interface confidence metrics such as pLDDT, pTM, ipTM, ipAE, and ipSAE, as well as an ensemble-level interface for accessing and aggregating these metrics for a single input across multiple seeds and samples. We benchmark FoldKit on three types of AF3 co-folding datasets: (i) a protein design campaign with 2 chains per input, (ii) a TCR-pMHC dataset with 4 chains per input, and (iii) a pooled-AF3 protein-protein interaction dataset with up to 22 chains per input. We find that FoldKit reduces storage requirements by approximately 5-15-fold compared to native AF3 outputs, depending on dataset composition, while maintaining direct programmatic access to individual predictions, ensembles, and confidence metrics. By reducing storage requirements and facilitating programmatic access to relevant outputs, FoldKit facilitates large-scale computational studies of biomolecular interactions. FoldKit is available from PyPI and can be installed using pip.
Read the full abstract
Computer Vision
Beyond Representation Learning: A Systematic Study of Joint-Embedding Predictive Generation for 3D Brain MRI
Joint-embedding predictive architectures (JEPAs) have primarily been developed for self-supervised representation learning. Denoising JEPA (D-JEPA) recently demonstrated strong generative capabilities on natural images, yet the applicability to 3D medical imaging remains unexplored. Building on the D-JEPA framework, we present Med-D-JEPA, a systematic adaptation and evaluation of joint-embedding predictive generation for 3D brain MRI. Med-D-JEPA operates on continuous latent tokens produced by a 3D KL-regularized adversarial variational autoencoder, and combines masked context prediction, representation-level alignment, per-token diffusion, and iterative next-set-of-token sampling. We evaluate unconditional and class-conditional generation quality on BraTS2019 and OASIS-1 datasets; downstream classification utility; and preliminary whole-tumor segmentation on BraTS2020. Across different generation settings, Med-D-JEPA achieves superior or competitive performance compared to several strong baselines on fidelity and diversity metrics. Compared to training with real samples, Med-D-JEPA-based synthetic pretraining improves classification AUC from 0.63 to 0.85 on BraTS2019 and from 0.78 to 0.87 on OASIS-1. In the segmentation study, pretraining on Med-D-JEPA samples improves Dice from 0.74 to 0.80 and reduces HD95 from 13.40 to 9.56 mm. These findings establish joint-embedding predictive generation as a promising direction for 3D medical image synthesis and encourage further research in this direction.
Read the full abstract
High Energy Physics (Phenomenology)
Constrained functional priors for Bayesian inference of hot QCD matter
Bayesian analyses of heavy-ion collisions rely on functional inputs, such as temperature-dependent transport coefficients and the equation of state, whose prior specification remains a significant source of uncertainty. Standard approaches employ low-dimensional parametrizations that can impose artificial correlations and leave the resulting inference sensitive to the assumed functional form. We develop a nonparametric framework for constructing priors over such functions using Gaussian processes, tailored to emulator-based inference in heavy-ion phenomenology. Truncated Karhunen–Loève expansions yield optimized finite-dimensional representations suitable for use as emulator inputs. We investigate how physical constraints can be incorporated into these priors, demonstrating that rejection-based methods produce non-Gaussian measures and induce nontrivial correlations in the expansion coefficients. By contrast, pushforward constructions enforce the constraints while preserving a tractable Gaussian measure in a latent space. These results provide a systematic framework for incorporating physical constraints into functional priors and clarify their effects on the statistical structure of the finite-dimensional representations used in Bayesian inference.
Read the full abstract
Cosmology and Nongalactic Astrophysics
Reconstructing stochastic gravitational-wave signals from flavour deconstruction with LISA
We investigate the reconstruction of a stochastic gravitational-wave background generated by a first-order phase transition in flavour-deconstruction models using the Laser Interferometer Space Antenna (LISA). In these scenarios, the spontaneous breaking of an extended flavour-non-universal gauge symmetry at the TeV scale can both generate the observed hierarchies of Standard Model fermion masses and mixing angles and induce a strong first-order phase transition. We consider representative benchmark points of the model, compute the corresponding thermodynamic transition parameters, and construct the resulting gravitational-wave spectra using a state-of-the-art sound-wave template. We then inject these spectra into simulated LISA data and perform Bayesian inference with SGWBinner, jointly reconstructing the cosmological signal, instrumental noise, and astrophysical foregrounds. We find that the strongest benchmark signal can be successfully reconstructed, whereas weaker signals are substantially degenerate with the unresolved extragalactic compact-binary foreground. We further show that tighter prior information on this foreground, motivated by observations with ground-based detectors, can reduce this degeneracy and improve signal reconstruction. Our results demonstrate that flavour-deconstruction models can produce signals accessible to LISA, while highlighting the importance of astrophysical-foreground modelling for their identification and characterization.
Read the full abstract
Computer Vision
ClearText-Video: A Large-Scale Text-Centric Video Dataset Bridging Video Restoration and Scene-Text Enhancement
Multimodal Large Language Models (MLLMs) have recently made strong progress in visual–linguistic understanding. However, their performance on text-centric video reasoning remains highly sensitive to input quality. Real-world user-provided videos often contain motion blur, compression artifacts, noise, and low-resolution text, which impair reliable text reading and downstream reasoning. Whether MLLMs can robustly read and reason about real-world scene text under diverse quality conditions remains a fundamental open question. We introduce ClearText-Video (CTVid), a large-scale, scene-text-aware benchmark for studying text-centric video understanding under controlled quality variation. CTVid contains 4,639 real-world text-rich egocentric videos, 550K+ frames, 1.6M human-verified scene-text annotations, and 220K+ spatial/temporal question–answer pairs in Chinese and English. For each high-quality video, CTVid provides content-matched Degraded-Quality and Restored-Quality variants, supporting two task families: Text-Centric Video Restoration and Multi-Quality VideoQA. We evaluate 18 representative restoration methods and 16 state-of-the-art MLLMs on CTVid. The results show that visual enhancement does not guarantee textual fidelity or downstream reasoning gains: blur is more damaging than low resolution, restored videos can alter the textual evidence used by MLLMs, and OCR-only pipelines remain far below direct multimodal reasoning. CTVid exposes the gap between video restoration and text-grounded understanding, providing a rigorous foundation for restoration-aware, quality-robust text-centric video systems.
Read the full abstract
Algebraic Geometry
Nonnegative conorms, regular matroids, and the tropical Schottky problem
We prove that a positive-definite quadratic form has nonnegative conorms if and only if it is matroidal, resolving a conjecture of Dutour Sikirić and Kummer. We show that the positive conorm support determines a regular matroid, and the positive conorms are the coefficients of a unimodular integral decomposition of the form. As a result, we show that a positive-definite form lies in the tropical Schottky locus if and only if its conorms are nonnegative and its positive conorm support matroid is cographic. This answers a question of Chua, Kummer, and Sturmfels and provides algorithms for the tropical Schottky decision and recovery problems in every genus.
Read the full abstract
Number Theory
Stable Solutions to Generalizations of Fibonacci Relations of Higher Order
We reformulate and discuss the conditions for the initial values in which a generalized Fibonacci sequence is eventually strictly monotonic. We then extend this exploration to the generalized \(k\)-bonacci sequence. We also give a lower bound on when an eventually strictly monotone generalized \(k\)-bonacci sequence becomes monotone.
Read the full abstract
Optics
Intracavity Dual-Resonance Stimulated Raman Spectroscopy with Cavity Ringdown Readout for Resolving Hydrogen Rotational Raman Transitions
Gas-phase rotational Raman lineshape metrology of hydrogen (H2) is challenging due to its weak Raman scattering cross-sections and intrinsically narrow linewidths. We report the first measurement of the complete Dicke-narrowing evolution of the H2 rotational Raman transitions S0(1) and S0(0), enabled by intracavity dual-resonance stimulated Raman spectroscopy with cavity ringdown readout and kHz-level spectral resolution. Our results establish a benchmark for gas-phase Raman lineshape measurements and provide stringent constraints for regime-based pressure-dependent linewidth models.
Read the full abstract
Human-Computer Interaction
Bringing Data to Life: Designing Data Characters for the Emotional Self
Journaling is a common practice for emotional expression, reflection, and processing. However, as entries accumulate, it can become difficult to interpret and compare their affective content, especially since traditional text-based analyses and visualizations often struggle to convey affective nuance. We introduce Data Characters, a visualization approach that represents affective content in journaling through human-like characters. Using a customizable Data Character as a design probe, we investigate the potential of character-based representations for conveying affective experiences and explore what visual encodings emerge through customization. Preliminary walkthroughs with two participants demonstrate the intuitiveness and feasibility of the approach. This work contributes an exploratory approach to studying how affective experiences can be visually represented and encoded through anthropomorphic forms.