Peer Revenue Returns subject to peer review

Papers

Fresh from arXiv, typeset properly. Highlighted words trade on the market.

All of arXiv

Tuesday 1 September

Optics

Low-threshold vortex lasing in three-dimensional cavities via flatband bound states in the continuum

Bound states in the continuum (BICs) offer a robust route toward high-Q resonators and topological structured-light generation. While optically pumped BIC lasers have advanced significantly, realizing electrically pumped vortex lasers remains challenging due to conflicting requirements for optical confinement and carrier injection. Here, we demonstrate an electrically pumped vector vortex laser based on a photonic crystal-guiding layer-distributed Bragg reflector (PC-GL-DBR) heterostructure. The photonic crystal slab precisely tailors the symmetry and dispersion of BIC modes, including a flatband BIC state, while an intermediate guiding layer enables optical mode engineering without directly patterning the multiple-quantum-well active region, preserving efficient carrier injection and strong optical confinement. Under optical pumping, the device achieves single-mode lasing with a Q of 9700, a threshold of 0.129 kW/cm$^2$, and a side-mode suppression ratio of 26 dB. Crucially, room-temperature electrically pumped vortex lasing is achieved with a low turn-on voltage of 1.8 V, a threshold current density of 0.13 kA/cm$^2$, and a cavity Q of 8600, successfully preserving the characteristic vector-vortex emission of the topological BIC state. These results establish a practical route toward electrically driven BIC vortex lasers for integrated structured-light sources and topological photonics.
Read the full abstract

All authors Zi-Xin Zhou, Weimin Wang, Huijie Li, Yan-qing Lu, Jun-Long Kou

Probability

Rough differential equations on manifolds via natural bundles

In the article, a novel framework for rough differential equations on finite-dimensional smooth manifolds driven by branched rough paths is developed utilizing the theory of natural bundles. The role of vector fields is played by sections of certain associated fiber bundles. The solutions are defined in a generalized Davie sense via a local approximation in such a way that they are invariant under changes of coordinates. Existence and uniqueness of the solutions is proved and a necessary and sufficient condition for the invariance of a submanifold for the solution is given. The approach allows the treatment of rough differential equations driven by fully branched rough paths on manifolds without directly relying on a shuffle product formula, bracket extension, or the Connes-Kreimer Hopf algebra and without imposing additional structure on the manifold.
Read the full abstract

Differential Geometry

The lower mean curvature bound in Gromov's mean-of-the-mean-curvature conjecture

Gromov conjectured that for a compact Riemannian manifold $X$ with boundary, the total mean curvature $\int_{\partial X} H$ is bounded above by a constant depending only on the intrinsic geometry of $\partial X$ and a lower bound on the scalar curvature of $X$. Previous results towards this conjecture require, in addition, a lower bound on the mean curvature of the boundary. In the present paper, we investigate whether this extra assumption is necessary. In dimension $2$, we show that no lower bound on the geodesic curvature is needed. We estimate the total geodesic curvature of the boundary in terms of its length and a lower bound for the Gauss curvature of the surface. This confirms Gromov's conjecture in $2$~dimensions without any extra assumptions. In contrast, we give examples showing that a lower bound on the mean curvature is genuinely needed in dimensions $n \ge 3$.
Read the full abstract

Computation and Language

PersuaRL: Reinforcement Learning-Driven Multi-Expert Selection for Persuasive Dialogue Generation in Insurance

Large Language Models (LLMs) are revolutionizing digital communication by powering conversational agents deployed across domains such as customer service, digital sales, and insurance. These agents, built on LLMs, can understand user input, retrieve relevant information, and generate coherent responses. However, while they excel at factual communication, they often lack the ability to engage in truly persuasive, context-sensitive dialogue, especially in domains like insurance, where trust and clarity are critical. Building on this need within the insurance domain, our work focuses on improving the persuasiveness of digital agents, aka LLMs. To support this, we introduce InsureDial, a Persuasive Insurance Dialogue dataset, designed to capture the nuances of persuasive communication specific to motor insurance interactions. We introduce PersuaRL, a reinforcement learning-based framework that equips LLM-driven dialogue agents with the ability to adaptively explore, select, and coordinate strategies across multiple expert modules, guided by the evolving dialogue context, to achieve more effective persuasion. We conduct extensive automatic human and qualitative evaluations on two benchmark persuasion dialogue datasets, including our InsureDial. Our evaluations consistently demonstrate that PersuaRL outperforms baseline, generating contextually appropriate and highly persuasive responses.
Read the full abstract

All authors Rohan Kirti, Akash Ghosh, Aryan Vats, Niladri Ghosh, Shipra Shriparn, Roshni Ramnani, Anutosh Maitra, Sriparna Saha

Software Engineering

Athena: Vulnerability-Affected Library Identification via Knowledge Graph Completion

A single vulnerability in a widely used library can cascade through millions of dependent applications, yet more than half of vulnerability database entries contain missing or incorrect affected-library information. Existing automated approaches neglect the relational structure of vulnerability databases, treating identification as an isolated text retrieval problem. In this paper, we propose Athena, the first graph-based approach for vulnerability affected library identification. Athena models vulnerability databases as a knowledge graph and reformulates the identification problem as knowledge graph completion (KGC). It comprises three key modules: a Modeling module that constructs a security knowledge graph integrating CVEs, libraries, CWE weakness types, CPE products, and software ecosystems; a Completion module that applies a modular KGC backbone to predict missing affected libraries for a given CVE via link prediction; and a Re-ranking module that retrieves KGC candidates and rescores them using a fine-tuned LLM augmented with knowledge graph embeddings, jointly leveraging structural and textual information. Our experiments on VulLib demonstrate that Athena significantly outperforms four state-of-the-art baselines, achieving a 32% improvement in Avg. F1 over the best baseline (i.e., VulLibGen). Notably, our KGC backbone with only 110M parameters already surpasses VulLibGen's best configuration at 7B parameters, demonstrating the effectiveness of graph-based modeling; the re-ranking module then provides substantial further gains, consistently outperforming the best baseline across all evaluated LLM backbones.
Read the full abstract

All authors Phong Trinh Duy, Trang Dang Yen, Hung Nguyen-Huu, Bach Le, Quyet-Thang Huynh, Dieu Hoang Vu, David Lo, Thanh Le-Cong

Cryptography and Security

Smart Contracts Claimed Vulnerable by the CVE Database, with Labels and Source Locations

The Common Vulnerabilities and Exposures (CVE) database catalogs vulnerability claims in hard- and software, among them those pertaining to blockchain programs a.k.a. smart contracts. We present CVE-Smart-Contracts, a curated dataset of CVE records up to July 2026 referring to Ethereum smart contracts. The dataset contains the vulnerable artifacts (source code and runtime bytecode), labels according to three taxonomies, and function-level locations. The retrieval of CVE records, collection of additional evidence, validation of the correspondence between records and artifacts, label assignment, and vulnerability localization are automated, leaving 15% to manual analysis. The dataset does not validate the original vulnerability claims, but marks a few records obviously wrong as `refuted'. For the sake of reproducibility, all external inputs are retained, so that rerunning the pipelines results in the same outputs. The dataset comprises 491 records linked to deployed contracts, 26 referring to projects (mostly libraries), 45 without validated artifacts, and six records with refuted claims. The dataset supports empirical security research, in particular the evaluation of code analysis and repair techniques.
Read the full abstract

Cryptography and Security

Reveree: Diagnosing LLM Reverse-Engineering Agents

Reverse engineering (RE) is critical to security tasks such as malware analysis and vulnerability discovery, and large language model (LLM) agents are increasingly able to perform it autonomously. Capture-the-flag (CTF) RE challenges have become the standard proxy for measuring this capability, but evaluation rests on a single criterion: whether the agent captures the flag. This solve rate reveals neither where in the RE process an agent fails nor whether a success reflects analysis of the binary or recall of a public solution. In this paper, we propose Reveree, a diagnostic framework that scores an LLM RE agent's trajectory at three tiers: solve rate, milestone progress through an eight-stage RE schema, and a behavioral profile of its actions. Comprehension stages are scored by an outcome-blinded LLM judge validated against a human expert; all other stages are verified deterministically. Using Reveree, we evaluate nine frontier models and four prompting strategies on 88 picoCTF and NYU-CTF challenges. We find that the base model dominates performance, whereas prompting strategy is a secondary, model-dependent effect. Surprisingly, larger, newer, or costlier models are not reliably stronger. We also find that failures concentrate at the comprehension stages of the RE process, and that extra budget, persistence, or reasoning effort rescues few of them, pointing to a competence limit rather than a resource limit. Regarding memorization, while models reproduce picoCTF flags from challenge descriptions alone, NYU-CTF shows minimal measurable recall, and most solves survive surface perturbation, indicating that genuine analysis coexists with memorization. We release Reveree to the community.
Read the full abstract

All authors Hadjer Benkraouda, Hongyu Cai, Berkay Celik, Gang Wang

Formal Languages and Automata

On smallest synchronizing terms over constant alphabets

We show a subexponential lower bound on the reset threshold of synchronizing deterministic finite tree automata (DTA) over alphabets of just two symbols. This significantly improves the previous one, which was quadratic in the number of states. Our result also narrows the gap towards the lower bound for DTA over alphabets that grow linearly with the number of states, as well as the best known upper bound, both of which are currently exponential.
Read the full abstract

Portfolio Management

Harvesting the Variance Risk Premium in Nuclear and Energy Equities: A Short-Put Portfolio Derisking Strategy

We study whether nuclear and energy-adjacent equity options carry a harvestable variance risk premium. Using CRSP and OptionMetrics data for 2000-2024 on a universe of 49 optioned securities under 41 ticker labels, at-the-money implied volatility exceeds the volatility realized over the following 21 trading days in 18 of 25 years, and 29 of 41 tickers show a positive premium on signal days with t-statistics above 2. A fully mechanical cash-secured short-put book sells one put per name per month near 0.25 delta and 45 days to expiry, and exits on the first touch of take-profit at half premium, stop-loss at twice premium, or a 0.50-delta stop, marked daily at validated quote midpoints. Equal-weight and inverse-volatility weightings earn 8.3% and 7.1% a year at Sharpe ratios of 0.98 and 0.99 with maximum drawdowns of 16% and 13%, versus 15.5% a year, a Sharpe ratio of 0.80, and a 46% drawdown for an equal-weight stock book; in 2008 the put book loses 9.5% against the stock book's 37.2%, and the monthly correlation between the two is 0.62. Performance is reported before transaction costs, on a fixed present-day universe.
Read the full abstract

Computation and Language

LLMPEDIA: Browsing, Verifying, and Comparing the Parametric Encyclopedic Knowledge of LLMs

Flagship language models appear saturated on benchmarks like MMLU (Hendrycks et al., 2021), scoring above 90% - yet benchmarks test only what the experimenter thought to ask, the availability bias of fixed question sets. LLMPEDIA makes this bias measurable and browsable. We recursively materialized ~1.3M articles from three model families' parametric memory (GPT-5-mini, DeepSeek-V3.2, Llama-3.3-70B) without retrieval, then audited a stratified sample of atomic claims against Wikipedia and a curated web stack, coloring every claim supported, refuted, or insufficient (Saeed and Razniewski, 2026). On a uniform random sample the true rate is 68.4% - more than 21 pp below MMLU - with 30.5% of claims insufficient: assertions no benchmark probes and the world's largest encyclopedia cannot adjudicate - long-tail knowledge or plausible hallucination, the evidence cannot tell - extending to free text the coverage gap GPTKB established for triples (Hu et al., 2025). The resulting live, open encyclopedia lets visitors inspect this frontier one claim at a time through five one-click views - link-traversal exploration, claim-level factuality, cross-model and political-persona comparison, and a guided topic drill-down - each page, claim, and verdict at a stable URL. LLMPEDIA is live at https://llmpedia.net
Read the full abstract

Information Theory

Quantum LDPC and High-Rate CSS Codes from Fair-Density Parity-Check Codes

We construct quantum LDPC (qLDPC) and high-rate CSS codes from our recently introduced classical fair-density parity-check (FDPC) codes. To this end, we introduce a structured sparsification of FDPC parity-check matrices, which reduces their check weights while preserving the underlying combinatorial structure and distance guarantees. Combined with the hypergraph-product construction, this yields finite-length qLDPC codes with analytically controlled blocklength, dimension, certified distance, and stabilizer weight. For quantum blocklengths $N<10^5$, the constructions introduced here span guaranteed rates from approximately $0.35\%$ to $25.8\%$ and certified quantum distances from $12$ to $69$, with stabilizer weights between $8$ and $16$. In the large-blocklength regime, allowing the FDPC order and sparsified check weight to scale moderately with blocklength yields a family of high-rate CSS codes, which we refer to as quantum FDPC (qFDPC) codes, with rate $R_Q$ and minimum distance $D$ satisfying \[ R_Q = 1-O\left(\frac{1}{\log\log N}\right), \ \ D=\Omega(N^{1/4}), \] and stabilizer weight $O(\log N\log\log N)$. Finally, the analytically available FDPC weight distribution provides explicit information about the logical operators of the resulting hypergraph-product codes. Over the quantum erasure channel, this structure yields rigorous first-order maximum likelihood (ML) expressions and a higher-order weight-distribution approximation to the ML logical block error probability. This enables estimation of finite-length operating points and the onset of the error-floor regime. To the best of our knowledge, beyond surface-code-type constructions, this is the first finite-rate qLDPC framework to provide both finite-length certified minimum-distance information and an analytical characterization of low-weight logical multiplicities.
Read the full abstract

Emerging Technologies

A Novel Space-Time Coding Architecture for Rydberg Atomic Quantum Receiver-Based Systems

Rydberg atomic quantum receivers (RAQRs) offer high sensitivity and wide tunability, but their magnitude-based readout yields a nonlinear model incompatible with conventional complex-valued multi-input multi-output (MIMO) processing. We propose a low-complexity space-time coding framework for point-to-point RAQR-assisted MIMO links. Data symbols are encoded using real orthogonal designs, while strong-reference heterodyne reception yields an equivalent real-valued linear model. The preserved orthogonality enables matched filter symbol-wise detection without matrix inversion or vector search. An analytical bit error probability expression is derived, proving the proposed scheme achieves the full transmit-receive diversity. Simulations validate the analysis and demonstrate improved performance over spatial multiplexing benchmarks.
Read the full abstract

All authors Asifa Zannat, Milad Abolpour, Dani Korpi, Mikko A. Uusitalo, Mikko Valkama, Ertugrul Basar

Nuclear Theory

QED nuclear medium effects at EicC

We evaluate quantum electrodynamics (QED) nuclear medium effects under experimental conditions at the future Electron-ion collider in China (EicC). We consider neutral-current and charged-current elastic scattering, as well as neutral-current and charged-current deep inelastic scattering. We compute cross-section corrections at first order in the opacity expansion and estimate kinematic modifications due to multiple rescattering within the nucleus. We perform the first calculations of charged-current deep inelastic scattering and polarized scattering. We present results for the anticipated $^{197}_{79}\mathrm{Au},~^{208}_{82}\mathrm{Pb},$ and $^{238}_{92}\mathrm{U}$ nuclei and extrapolate these calculations to the lighter $^{40}_{20}\mathrm{Ca}$ isotope. We find that QED nuclear medium effects are non-negligible when extracting the process-independent non-perturbative structure of nucleons and nuclei at the EicC.
Read the full abstract

Optics

Reaching the optical propagation limit in temporal analog computing

A central goal of optical computing is to perform calculations on the timescale of light propagation. Yet many analog photonic solvers require feedback, storage or field build-up before the answer becomes available, introducing additional latency that limits real-time operation. Here we introduce the concept of two-time modulation for temporal analog computing, in which the material response is independently modulated along two temporal directions, allowing the computational operator to be constructed continuously as the waveform propagates. The solution can therefore form during optical transit, without an additional solution-formation timescale. This form of temporal control allows identical copies of a waveform separated only in time to reach entirely different target outputs within a single spatial channel, and enables nonlocal integral-equation solving in a single passage. The resulting framework brings compact, programmable, real-time analog computation within reach for ultrafast optical information processing.
Read the full abstract

General Relativity and Quantum Cosmology

Josephson interferometry in an Oppenheimer–Snyder-like scale-dependent black-hole spacetime

We develop a covariant framework for Josephson transport, superconducting interference, and shunt noise in the scale-dependent exterior generated by an Oppenheimer–Snyder-like collapse. Static Josephson frequencies and transported currents referred to Killing time acquire one lapse factor, whereas power acquires two. For a junction comoving with the collapsing surface, the coordinate-time phase rate differs from the frequency received at infinity because of null propagation and Doppler effects. We derive the exact radial-null travel-time kernel and show that its local near-extremal logarithmic enhancement crosses over, at fixed emission offset, to pole-controlled extremal behavior. We also obtain the redshifted resistively and capacitively shunted-junction equation and the dc and microwave-driven two-junction interference envelopes. In the negligible-total-inductance limit, static lapse imbalance modifies lobe amplitudes without shifting their centers, whereas microwave-induced translations require dynamical fluxoid closure. In a Hartle–Hawking state, Tolman redshift produces a superconducting exclusion layer and a lapse-independent asymptotic shunt-noise spectrum; its low-frequency limit obeys a parameter-free fluctuation–delay relation. Finally, we derive a shadow–Josephson consistency relation and sensitivity bounds on the running parameter.
Read the full abstract

Social and Information Networks

Don't You Know, Pump it Up! Investigating Cryptocurrency Manipulation in Telegram-Driven Activity

Telegram plays a pivotal role in cryptocurrency communication and has been repeatedly associated with coordinated schemes, such as pump-and-dump manipulation. However, existing studies typically focus on known manipulation chats or a limited set of cryptocurrencies, leaving open the question of how Telegram is leveraged for mass promotional activity (shilling) at scale. Moving beyond these limitations, this work analyzes the interplay between information flows and market activity across public Telegram channels. To this end, we propose a scalable framework that (i) classifies crypto-related messages using a fine-tuned encoder model to filter semantic noise, (ii) detects anomalous spikes in cryptocurrency mentions via adaptive thresholding, and (iii) validates temporal associations between social bursts and market movements using quasi-experimental econometric methods (RDD and DiD). We apply this framework to one year of public Telegram data (14,499 channels and over 20 million messages) aligned with transaction data for more than 17,000 cryptocurrencies. Our analysis identifies 47 events consistent with potential pump-and-dump activity and 73 sustained market reactions, showing that manipulative signals are characterized by extreme temporal synchronization and precede price movements by seconds. Notably, psycholinguistic analysis reveals that pump-and-dump messages are linguistically indistinguishable from organic discussions, highlighting the limits of text-based detection alone. Finally, we estimate the cumulative financial volume of detected pump-and-dump events to exceed $200 million and release a public cryptocurrency dictionary and a fine-tuned classifier to support future research.
Read the full abstract

All authors Filipe Moura, Giordano Paoletti, Carlos H.G Ferreira, Jussara Almeida

Quantum Physics

Quantum time-flip beats adaptive metrology: Asymptotic benefit, activation, and unsimulability

In quantum metrology, adaptive and causal-superposition strategies are proven to be beneficial over parallel schemes for a finite number of channel uses, but their advantages disappear in the asymptotic limit. We show that quantum operations with indefinite time direction, specifically, time-flip (TF)-assisted strategies, referred to as indefinite time-directed metrology (ITDM), can overcome this asymptotic equivalence. Using semidefinite programming, we rigorously demonstrate that TF-assisted protocols can achieve quantum Fisher information (QFI) strictly exceeding the maximum value attainable by parallel, adaptive, and causal-superposition strategies, both for finite and asymptotically many channel uses. Moreover, we identify a class of Pauli noise channels for which ITDM achieves Heisenberg scaling, while all parallel, adaptive, and causal-superposition strategies remain restricted to standard scaling. We call this phenomenon as metrological activation. Interestingly, this activation can be used to exhibit that the quantum time-flip and transposition supermaps cannot be simulated by conventional quantum circuits or causal-superposition strategies using any finite number of channel queries, thereby establishing indefinite time direction as a genuine resource for quantum metrology.
Read the full abstract

Cryptography and Security

A SoK for SoCs: Reading the TI Leaves on AI for Cyber Threat Intelligence Generation and Sharing

Cyber Threat Intelligence (CTI) is essential for defending mission-critical infrastructure, yet the process of transforming raw attack evidence into shareable CTI remains fragmented and understudied. We conduct a literature survey of academic papers, organizing the CTI lifecycle into three stages: Threat Data Collection, CTI Generation and Sharing, and CTI Consumption. The first and third stages are well represented in the literature, whereas only a small number of papers address CTI Generation and Sharing. To learn how this stage is practiced, we survey practitioners across multiple organizations who routinely generate and share CTI. They describe a largely manual process with four recurring challenges: preventing the exposure of sensitive information, extracting indicators from noisy attack data, correlating observed behavior with standardized tactics, techniques, and procedures (TTPs), and translating CTI into the formats that sharing platforms require. Using the insights from the practitioner survey, we divide the CTI Generation and Sharing stage into four steps: Intelligence Extraction, Normalization and Enrichment, Codification, and Distribution. We then conduct pilot studies that probe the feasibility of current Large Language Models (LLMs) for each step. The pilot studies show that LLMs can assist an analyst in each of the four steps. However, the models recover only a fraction of the indicators the evidence contains, struggle to ground every claim in the supplied evidence, and do not judge what keeps shared intelligence useful to its recipients. Each step therefore requires expert supervision. Based on these observations, we derive three research directions for automating the production of shareable intelligence.
Read the full abstract

All authors Saastha Vasan, Hadjer Benkraouda, Jizhou Chen, Leyan Pan, Doguhan Yeke, Shinan Liu, Noah Spahn, Stefano Ortolani, David Evans, Christopher Kruegel, Giovanni Vigna

Representation Theory

Polynomiality for algebras of invariants associated with some Inönü-Wigner contractions in type A

This paper deals with polynomiality of algebras of symmetric invariants (or generated by symmetric semi-invariants) associated with some particular Inönü-Wigner contractions in type A. More precisely we are interested with contractions of some parabolic subalgebras in type A with respect to their decomposition into their Levi factor and their nilpotent radical, the latter becoming an abelian ideal of the contraction. When the parabolic subalgebra $\mathfrak p$ has its Levi factor decomposing into three blocks, which are symmetric with respect to the antidiagonal, we show in this article that the algebra of symmetric invariants associated with the contraction of the canonical truncation of $\mathfrak p$ has a Weistrass section. It follows that such an algebra of invariants is a polynomial algebra, for which we can give the number of algebraically independent homogeneous generators, their weight and degree. The method relies on the construction of an adapted pair for such a contraction. When the Levi factor of $\mathfrak p$ has its central block not of the same size as its two extremal blocks, we also show that the above algebra of symmetric invariants is equal to the algebra generated by symmetric semi-invariants associated with the Inönü-Wigner contraction of $\mathfrak p$.
Read the full abstract

Computer Vision

Monocular Depth Estimation from a Single Image: Progress and Opportunities

Monocular depth estimation has long stood as a fundamental challenge in computer vision, enabling a wide range of applications including 3D reconstruction, robotics, autonomous driving, and augmented reality. This survey traces the field's evolution from early learning-based methods to the emergence of transformative foundation models. We begin by framing the problem, distinguishing between relative and metric depth estimation, and highlighting the key challenges that have shaped a decade of research. We then present common problem formulations and introduce the most widely used datasets, covering indoor, outdoor, and synthetic data. Following this, we review major advances prior to the foundation model era, distilling core insights from influential methods that contributed to improvements in accuracy, efficiency, and robustness. The survey then turns to the recent surge of foundation-model-based approaches, categorizing them into discriminative and generative paradigms and emphasizing the critical roles of large-scale pretraining (e.g., DINOv3) and synthetic data. We compare representative models using both quantitative benchmarks and qualitative examples, and discuss natural extensions to video-based depth estimation. Further, to illustrate real-world impact, we highlight the integration of depth estimation into applications such as visual SLAM, content generation, and robot perception. Finally, we outline open challenges and promising research directions as the field advances further into the era of foundation models.
Read the full abstract

All authors Muxin Liu, Xiaoyang Lyu, Yang-Tian Sun, Yi-Hua Huang, Ziyi Yang, Peng Dai, Xiaojuan Qi

Cryptography and Security

Johnny Still Receives Spam SMS: Assessing the Robustness of SMS Spam Detection

SMS spam detection systems often achieve high accuracy in controlled environments but struggle against adversarial attacks and increasingly sophisticated spam tactics in real-world deployments. In this paper, we evaluate the robustness of SMS anti-spam systems that end users actually rely on, including commercial messaging applications, third-party anti-spam services, and publicly available open-weight models hosted on Hugging Face. We evaluate these systems under both standard and adversarial conditions, considering perceptible and state-of-the-art imperceptible attacks. We include only perturbations that we verify survive real SMS or RCS delivery, rather than lab-only artifacts. Our experiments reveal significant gaps in existing spam detectors' ability to identify adversarially manipulated messages. We further demonstrate that adversarial training alone is insufficient. Using an explicit held-out evaluation protocol, we find that robustness transfers well within a perturbation family but degrades sharply against structurally distinct, encoding-level attacks. To address these weaknesses, we propose a multi-model ensemble that combines adversarial training with spam classifiers diverse in architecture and tokenization. Our results show that this ensemble, particularly when using a minority-voting strategy, substantially improves robustness against both perceptible and imperceptible adversarial attacks while maintaining competitive classification accuracy. We also characterize the resulting precision-recall trade-off and recommend operating points for false-positive-sensitive and recall-critical deployments. These findings highlight the need for comprehensive robustness evaluations and ensemble-based defenses for building more secure SMS spam detection systems in real-world settings.
Read the full abstract

All authors Muhammad Salman, Muhammad Islam, Muhammad Ikram, Mohamed Ali Kaafar

Machine Learning

Pre-carved Niches: The Formation Dynamics of Modular Task Partitions in Early LLM Training

Large language models exhibit a modular internal organization that mirrors well-studied functional networks of the human brain, but how this organization forms during training is unknown: prior work has characterized finished models, not the formation process. We track formation step by step: we train a Pythia-410M model from scratch (two trajectories, bf16 and fp32) and run attribution patching at every step, alongside probes for gradient norms, effective updates, weight norms, and first-order loss decomposition across 14 tasks in four cognitive domains. Three findings. First, the modular map is pre-carved: before any learning, the dominant task pair already overlaps at ~3.6x the attribution substrate (a task-independent baseline), and its layer-0 concentration is an architecture-level constant on this model family. Second, the partition locks in through two sharp jumps whose amplitudes do not track the learning-rate schedule (the second reaching 20.4 sigma quiet-window / 6.2 sigma global), accompanied by gradient-level relative deprivation–winners receive 2.25->2.73x the loser's gradient supply, 9.5-11.5 standard deviations below a random control–that does not propagate to updates or weights. Third, deviation from the substrate appears only in the domain being learned, consistent with the hypothesis that modularity tracks learning. We close by separating the feature-level account we can defend from the mechanistic questions we cannot, and we pre-register the scale-threshold hypothesis behind our ongoing 2.8B experiments.
Read the full abstract

Earth and Planetary Astrophysics

Optical Colors as Compressed Spectral Information for Trans-Neptunian Objects

Broadband optical colors have long been recognized as tracers of the surface spectral properties of trans-Neptunian objects (TNOs), yet their relationship to near-infrared (NIR) spectra has not been quantitatively characterized. We construct an empirical optical-NIR spectral manifold from JWST spectroscopy of 40 TNOs and 8 Neptune Trojans and interpret broadband photometry as compressed representations of this manifold. Using the spectral manifold as a prior, the model conditions on observed broadband optical colors to infer a posterior distribution in the spectral latent space. We then evaluate filter combinations through posterior entropy and Kullback-Leibler information gain and found that the gr colors provide broad localization on the manifold, while additional filters substantially reduce posterior uncertainty. Independently defined optical color populations map coherently onto the major compositional classes of the spectral manifold, providing a physical interpretation of previously identified color groups. The learned manifold also enables the inverse mapping back to optical color space. Sampling and decoding the manifold reproduces the observed topology of optical color space, indicating that these color populations arise naturally as low-dimensional projections of a continuous optical-NIR compositional manifold rather than isolated empirical clusters. The forward and inverse mappings provide a unified interpretation of optical colors as compressed representations of an optical-NIR spectral manifold. As JWST expands the manifold and Rubin maps millions of objects onto it, this framework provides an innovated approach to population-scale compositional studies and the discovery of rare surface types.
Read the full abstract

All authors Hsing-Wen Lin, Larissa Markwardt, Kevin Napier, Fred Adams, Renu Malhotra, David Gerdes

Artificial Intelligence

Jailbreaking Text-to-Image Models Through Cracks: Navigating Heterogeneous Safety Filters via Multi-Agent Debate

Text-to-image (T2I) models remain vulnerable to jailbreak attacks that elicit Not-Safe-For-Work (NSFW) content, despite increasingly being guarded by heterogeneous, multi-layer safety stacks combining text filters, image classifiers, and cross-modal detectors. Existing jailbreak studies either optimize against individual filters or query the complete pipeline with aggregate feedback, making it difficult to identify the active constraint and adapt to conflicts across safety layers. In this paper, we introduce the Detection Surface, a unified geometric framework that characterizes the decision boundaries induced by heterogeneous T2I safety filters and their joint effect on the jailbreak search space. This formulation reveals that successful evasion is governed by a sparse and non-convex region shaped by cross-layer conflicts, where mutations that bypass one filter may increase exposure to another. Motivated by this analysis, we propose CRACK, a multi-agent debate framework for adaptive jailbreak search that decomposes jailbreak search into exploration, diagnosis, and arbitration. CRACK coordinates an Attack Agent, a Defense Agent, and a Judge Agent to iteratively generate prompt mutations, obtain layer-specific diagnostic feedback, and optimize mutation strategies through reward-guided refinement. Through repeated rounds of debate, CRACK adapts its search direction to the evolving cross-layer constraints while preserving the original harmful intent. Extensive experiments across multiple T2I models, datasets, and safety configurations show that CRACK achieves Attack Success Rates (ASR) of up to 99.63% under composite defenses, while requiring fewer queries than existing methods and maintaining semantic fidelity.
Read the full abstract

All authors Kaiyan Wen, Shijie Zhang, Lu Yu, Guangdong Bai

Multiagent Systems

Classic AI Scaffolding for LLM Social Agents

Large language models can produce locally plausible social turns, but fluent next-turn generation is not enough for social simulation. Human encounters such as restaurant lunches and hotel check-ins are bounded social episodes with roles, scripts, material state, obligations, commitments, timing, and closure conditions. We present EpisodeSim, a hybrid LLM-agent architecture that represents classic-AI structures as natural-language control state interpreted by LLM calls. A World Master maintains shared reality, constructs scenes, adjudicates proposed actions, tracks effects and obligations, and controls closure. Experiments with small qualitative ablations on two held-out settings support a design claim: LLM fluency supplies local texture, but coherent social simulation benefits from persistent classic-AI-style scaffolding that organizes behavior over time.
Read the full abstract

Statistics Theory

Nonparametric inference for density-dependent McKean–Vlasov diffusions

The present research is devoted to the nonparametric estimation of a density-dependent drift coefficient in a multivariate McKean–Vlasov diffusion from independent observations at a common time, as well as the stationary density. Under certain assumptions on the (known) potential, we reduce the problem to the one-dimensional one and construct a sieve maximum-likelihood estimator based on sparse ReQU neural networks subject to structural and Hölder constraints. Using the endpoint-adapted graded approximation, we achieve the rate of $\left(b_n\log n/n\right)^{2(\beta+1)/(2\beta+3)}$ for the Kullback-Leibler divergence between the true and estimated stationary densities, with $b_n$ being at most a logarithmic factor. Similarly, it is shown that the constructed estimator for the drift coefficient converges to the true one at the rate of $\left(b_n\log n/n\right)^{\beta/(2\beta+3)}$ in the $L^2$-metric. A matching Assouad lower bound proves minimax optimality of this bound up to logarithmic factors.
Read the full abstract

Commutative Algebra

Symbolic Rees algebras of complementary edge ideals

Let $G$ be a finite simple graph on $[n]$ and let $I_c(G)$ denote its complementary edge ideal in the polynomial ring $S = K[x_1,\dots,x_n]$. We give a combinatorial description, in terms of the structure of $G$, of the minimal generators of the symbolic Rees algebra $\mathcal{R}_s(I_c(G)) = \bigoplus_{k \geq 0} I_c(G)^{(k)} t^k$, and show that this algebra is generated in degree at most $6$. Moreover, we completely determine the minimal generators of $\mathcal{R}_{s}(I_{c}(G))$ in graph-theoretic terms. We then study in more detail the homological invariants of the symbolic powers $I_c(G)^{(k)}$ for the classes of cycle graphs and complete multipartite graphs. For theses families, we study the behavior of the symbolic depth function $k\mapsto\operatorname{depth} S/I_c(G)^{(k)}$, we obtain the limit depth of the symbolic powers and the Waldschmidt constant of $I_c(G)$, and further prove that all the symbolic powers $I_c(G)^{(k)}$ are componentwise linear.
Read the full abstract

Dynamical Systems

Early Warning Signals Can Vanish or Amplify: Dimensionality in Complex Systems

Early warning signals (EWS), such as increasing variance and autocorrelation, are widely used to anticipate critical transitions associated with saddle-node bifurcations. However, real-world systems are often high-dimensional and multiscale, potentially altering the classical behavior of EWS. Here, we investigate how dimensionality, inertia, and red noise influence the detectability of EWS prior to tipping points. We show that when observations are not aligned with the critical direction, stable dynamics in orthogonal directions can mask EWS until very near the bifurcation. We further demonstrate that second-order dynamics with damping modify the autocorrelation structure and may either enhance or reduce signatures of critical slowing down. Finally, we study how colored noise influences EWS. Our results show that the presence and strength of EWS are not universal properties of tipping systems but depend critically on system geometry and stochastic forcing, implying that the absence of detectable EWS does not necessarily rule out an approaching critical transition.
Read the full abstract

High Energy Physics (Phenomenology)

Saturation effects in diffractive deep inelastic scattering

We study the production of $c$ and $b$ quarks in photon-induced processes as a probe of gluon saturation, with a focus on the high-energy region accessible in ultra-peripheral collisions (UPCs) at the LHC. We compare predictions for inclusive and diffractive heavy quark production with and without saturation effects by using the nonlinear Balitsky–Kovchegov and the linear Balitsky–Fadin–Kuraev–Lipatov equations for the evolution of the dipole amplitude. The percentage of coherent diffractive events is predicted to be $~20\%$ in the saturation model, which is approximately twice as large as predictions based on parton distribution functions and shadowing. This indicates that measurements of diffractive heavy quark production in UPCs may provide a sensitive probe for distinguishing linear and nonlinear regimes of quantum chromodynamics.
Read the full abstract

Astrophysics of Galaxies

Sulphur within the extreme environment of the central molecular zone of NGC 253: a chemical modelling approach

Sulphur(S)-bearing species are ubiquitous in Galactic star-forming regions, from dense cold cores to outflows and shocks linked to protostellar activity. A recent observational study investigated the origin of S-bearing species towards the central molecular zone (CMZ) of NGC 253 and showed that Sulphur emission is linked to the presence of forming stars. However, more extensive modelling of these observations are required to determine the exact origin of the Sulphur emission (e.g. shock or thermal evaporation). Using chemical modelling, we examine how S-bearing species behave in the more extreme environment of the starburst galaxy NGC 253, and more generally, how they can help us improve our understanding of the emission linked with the dense star-forming gas in external galaxies. We use the gas-grain time-dependent chemical model UCLCHEM to model static warm clouds and C-type shocks under the physical conditions found in the CMZ of NGC 253. We compare observed abundances and abundance ratios to the modelled output abundances. We found that depending on the model type (shock, post-shock or static cloud), the highest abundance reached by the S-bearing species varies significantly. Hence, we can use their observed abundances of S-bearing species to distinguish between shocked, post-shocked or a quiescent (non-shocked) gas. We also confirm observationally-based conclusions on their emission origins, including in the case of unresolved emission. Comparing GMC-scale observations with chemical modelling is a powerful method to investigate and constrain the origin of molecular emission towards extragalactic star-forming regions. Sulphur-bearing species are useful to distinguish between different types of ISM components (shocked/post-shocked/quiescent gas).
Read the full abstract

All authors M. Bouvier, K. M. Dutkowska, S. Viti, J. G. Mangum, E. Behrens, C. Eibensteiner