Cryptography and Security
Papers
Fresh from arXiv, typeset properly. Highlighted words trade on the market.
All of arXiv
Friday 28 August
Retrieval-augmented generation (RAG) augments LLMs with external documents, but public or user-editable sources expose RAG systems to data poisoning: attackers can inject malicious documents to steer outputs toward targeted answers. Existing poisoning attacks often rely on query inclusion, inserting the target query into poisoned documents to improve retrieval; however, this creates lexical and embedding-space artifacts that make them easy to filter. We propose CamoDocs, a poisoning attack that avoids direct query inclusion by camouflaging adversarial documents among benign content. CamoDocs chunks synthesized benign and adversarial drafts, replaces selected tokens in benign chunks with dispersion tokens that spread poisoned-document embeddings, and applies coherence filtering to limit readability degradation. Across seven RAG defenses, three open-weight LLMs, and three benchmarks, CamoDocs achieves strong average ASR while avoiding query-overlap artifacts exploited by simple query detection. It also remains effective against proprietary models, achieving average ASRs of 61.80% on GPT-5.4-mini and 55.09% on Claude-Haiku-4.5. Finally, we show that erasure-heavy clustering defenses such as TrustRAG can reduce ASR, but only with substantial utility drops on retrieval-dependent benchmarks such as NeoQA. Code is available at https://github.com/jaewonalive/CamoDocs.
Astrophysics of Galaxies
Dynamical Modeling of Kinematically Decoupled Galaxies: Structural Components
We present the results of dynamical modeling of two galaxies with inconsistent kinematics – PGC 35706 and LEDA 2220522 – using data from the MaNGA integral-field spectroscopic survey and deep photometry from the DESI Legacy Imaging Surveys. The Schwarzschild orbit-superposition method in the axisymmetric approximation was employed to identify structural components based on the orbital circularity parameter $\lambda_z$. Three components have been identified in the galaxies: a spheroidal (bulge/halo) component and co-rotating and counter-rotating stellar disks. The spheroidal component dominates by mass in both galaxies: about 60% in PGC 35706 and 51% in LEDA 2220522; the contributions of the co-rotating and counter-rotating disks are 29% and 9% in PGC 35706 and 30% and 18% in LEDA 2220522, respectively. The kinematics of the ionized gas in both galaxies is consistent with the rotation of the counter-rotating disk, indicating that it formed from externally accreted gas with opposite angular momentum. The results support a scenario in which counter-rotating disks form through gas accretion from the intergalactic medium or through minor mergers with gas-rich satellites. This work demonstrates the effectiveness of the Schwarzschild method for studying the complex kinematic structure of galaxies and their formation history.
Read the full abstract
Hardware Architecture
Neuromorphic architectures as numerical solvers for computational neuroscience
Neuromorphic computing is closely associated with spiking neuronal networks. However, an alternative class of so-called "rate-based" models arising from computational neuroscience and machine learning forgoes spiking interactions and instead relies on continuous coupling between neurons. Existing neuromorphic implementations designed around spike-based interactions are not well-suited for emulating such models. Here view the distributed simulation of these models as message-passing algorithms on parallel hardware. Leveraging prior art in numerical algorithms and distributed simulation, we outline steps that enable the design of efficient digital neuromorphic accelerators for non-spiking neuronal models. In particular, we show that multi-bit packets, rather than spikes, are the most efficient communication strategy in packet-switched networks and that compared to basic numerical integration methods, higher-order differential equation solvers decrease both computation and communication costs while achieving lower numerical error, but that these benefits are ultimately limited by arithmetic precision. Using our proposed design principles, we convert an existing neuromorphic architecture into a distributed numerical solver - a spikeless neuromorphic system - for continuously-coupled neuronal models. We thereby demonstrate that our theoretical considerations indeed translate into practical advantages, namely reduced energy consumption and delay.
Read the full abstract
Computer Vision
GraspHOI: Full-Body 3D Human-Object Reconstruction with Finger-Level Grasps from a Single In-the-Wild Image
Existing monocular full-body 3D human-object interaction (HOI) methods do not combine explicit finger-level grasp optimization with category-agnostic object reconstruction. Despite plausible body-object configurations, their fingers may float from or penetrate objects instead of forming a grasp. We present GraspHOI, the first framework that reconstructs a full-body 3D HOI from a single image while explicitly optimizing finger articulation against the reconstructed object. GraspHOI recovers object geometry directly, without predefined meshes or a fixed category vocabulary. It reconstructs the body, hands, and object separately, aligning them in metric camera space via depth-based registration and image-space alignment. Occlusion-aware palmar correspondences seat the object against the grasping hand, and contact-aware optimization refines arm and finger articulation to form surface contact without excessive penetration. Across four benchmarks and six baselines, GraspHOI improves relative human-object placement, hand accuracy, and contact plausibility. Full pipeline code will be released.
Read the full abstract
Formal Languages and Automata
Separating Words with Automata in the Half-adversarial Case
We consider the problem of separating words with deterministic finite automata (DFA) (Goralčík and Koubek, 1986). This problem asks: given two distinct words $u,v$ of length at most $n$, what is the size of the smallest DFA that accepts one and rejects the other? The best upper bound on the worst-case over all pairs of words of length at most $n$ is $\tilde{O}(n^{1/3})$ states (Chase, 2021), while the best lower bound is $\Omega(\log n)$. In this work, we consider the half-random, half-adversarial case: we show that if $u$ is a uniformly random binary word of length $n$, then with high probability, for any word $v$ not equal to $u$, there is a DFA with $O(\log^{7/3} n \mathrm{poly}\log\log n)$ states that separates $u$ and $v$. Our results are based on a novel analysis that exploits the structural sparsity of random words: we show how to apply block-wise compaction with small deterministic transducers to reduce the separation problem to the case of words with short run-length encodings.
Read the full abstract
Artificial Intelligence
MAP: A Benchmark on Multimodal Accessibility Planning for Real World Places
We introduce MAP, the first benchmark to evaluate multimodal AI systems as assistants for users with accessibility requirements when planning visits to places in the real world. In our evaluation, systems are presented with requests to verify or recommend a point of interest meeting an accessibility requirement. MAP contains two novel assessments: Claim verification for accessibility planning assesses if information on places and stated accessibility features is supported and identifies places that satisfy requested accessibility features. Visual evidence retrieval for accessibility planning checks if a multimodal AI system can select visual evidence for the requested place and accessibility feature. Our methodology supports comparison of AI systems in a setting where place information and accessibility information can change over time by evaluating systems and refreshing ground truth data at scheduled times. The benchmark is based on automatic rating and human rating for a proportion of responses.
Read the full abstract
Computer Vision
Semantic Head Specialization Guides Hybrid ViT Attention for Multimodal LLMs
Hybrid attention dominates frontier LLMs, yet Vision Transformers (ViTs) in multimodal LLMs lack a satisfactory hybrid design, with no consensus on why certain attention patterns work better. To fill this gap, we study ViT attention heads and find they differentiate into object- and background-specialist roles, a pattern most pronounced under full attention; we call this Semantic Head Specialization (SHS). We propose SHS-Index to quantify this specialization, show that it distinguishes full-attention from chunk-window ViTs, and find that it strongly tracks downstream benchmark performance. We then identify three structural factors that shape SHS—window interaction, token serialization, and local softmax allocation—and use them as design principles for hybrid attention. Guided by these factors, we design Ariadne Attention, a hybrid that matches full attention on 22 image and video tasks at 6.5x less attention compute. Our findings establish head specialization as a measurable property for diagnosing and designing principled hybrid ViT attention at the multimodal-LLM scale.
Read the full abstract
Computation and Language
When Linguistic and Internal Confidence Diverge in Large Language Models
Users often ask large language models (LLMs) to report how confident they are, but it is unclear whether such linguistic confidence tracks the model's internal confidence. We study this question across 8 classification tasks, 2 generation tasks and 30 models from three families. For classification, we compare linguistic confidence with logits-based confidence along three axes: association, magnitude agreement and calibration. For generation, we test whether linguistic confidence tracks semantic-entropy-based uncertainty. The axes frequently diverge. Instance-level association is weak on average, although it improves on easier items and for stronger base models. Instruction-tuned models often report higher confidence and sometimes show higher association, but they also have larger confidence gaps and worse calibration. Prompt design mostly changes the distribution of reported confidence. Attitude cues inflate confidence without improving alignment, while score exemplars can preserve rank-order signal when they avoid collapsed confidence values. Regression analyses show that distributional properties of confidence scores explain much of the observed alignment pattern, with model metadata playing a smaller role after controls. These results support a lossy-channel view of linguistic confidence. A more dispersed verbal confidence distribution can carry useful rank information, but it does not make the scores calibrated. Linguistic confidence should therefore be evaluated with multi-axis diagnostics before being used in downstream reliability pipelines.
Read the full abstract
Mesoscale and Nanoscale Physics
Electrically Tunable Two-Component Exciton Condensate in a Coulomb-Coupled Graphene Trilayer
Multicomponent condensates possess internal phase degrees of freedom unavailable to a single-component condensate, yet their components are rarely controllable in solids. Here we realize a graphene trilayer with negligible interlayer tunnelling in which the layer-specific carrier densities are continuously tuned by electrostatic gating. Quantum-capacitance measurements demonstrate that charge-incompressible quantum Hall states at total filling factors 1 and 2 persist across the full range of layer-filling configurations and continuously connect the three bilayer exciton-condensate limits. This persistence provides evidence for a trilayer excitonic state. Static Hartree-Fock and time-dependent Hartree-Fock calculations yield two independent finite phase-stiffness eigenmodes and two linearly dispersing Goldstone modes, respectively, when all three layers are partially filled, whereas only one phase-stiffness eigenmode and one linear Goldstone mode remain when one layer is unfilled. The stiffness eigenmodes rotate continuously between the two adjacent-layer exciton bases as charge is transferred among the layers, revealing electrical control of the condensate-mode composition. Together, the experimental and theoretical results support the identification of a two-component exciton condensate with a continuously tunable internal structure.
Read the full abstract
Soft Condensed Matter
Nonholonomic collective flows: velocity–orientation locking in a continuum with microstructure
We develop a continuum theory for a fluid of elongated particles that advance along their own axes. The same kinematics is ideally shared by flocks of sheep, self-propelled rods, vehicular traffic, and turning flocks of birds. Within the framework of continua with vectorial microstructure, in which each point also carries an orientation, we impose the no-side-slip (skate) condition $\vv=u\,\nn$ as an ideal, non-integrable internal constraint. We derive the pure equations of motion and the equation governing the constraint reaction from the principle of virtual power. Using a constitutive closure provided by the Ericksen-Leslie theory of nematic liquid crystals, we show how the constraint also shapes collective effects. It turns parabolic orientational diffusion into hyperbolic orientation-density waves and forbids any steady simple shear. It fixes the empirical Toner-Tu convective coefficient to the flow-alignment ratio measured in colloidal rollers, thereby providing a mechanical foundation for angular sound in micropolar active hydrodynamics. It also recovers the inertial spin model of bird flocks, with the addition of a banking force and a turn-density coupling. Finally, it recovers the classical macroscopic models of one-dimensional traffic flow, while describing steering, lateral tyre forces, and road geometry in two dimensions.
Read the full abstract
Quantum Physics
Quantum Federated Learning Based on Bures–Uhlmann Geometry for Heterogeneous Noisy Clients
Quantum federated learning enables collaborative model training across quantum devices without sharing raw data, and it faces the data and hardware heterogeneity inherent to noisy quantum devices. Utilizing the quantum geometric tensor is a natural remedy, yet pure-state approaches and diagonal approximations discard the correlations that encode parameter incompatibility. To address this, we extend the parameter-space geometry to the mixed states that noisy clients actually prepare. The real part of the resulting mixed-state geometric tensor is the Bures metric, which measures how fast the physical state changes under parameter variation, and the imaginary part is the mean Uhlmann curvature, which quantifies the incompatibility of estimating multiple parameters simultaneously. Accordingly, we employ the Bures metric as a local preconditioner and use the mean Uhlmann curvature to develop an achievable-precision aggregation rule that dynamically down-weights unreliable clients. Furthermore, we establish theoretical guarantees by proving a convergence theorem and a variance-dominance proposition. Empirical evaluations on a trapped-ion quantum emulator demonstrate that the proposed method maintains high accuracy across diverse device-heterogeneity conditions and outperforms standard federated averaging, whose accuracy degrades under strong noise.
Read the full abstract
Computation and Language
PersonaForge: Realistic Multi-Turn User Simulation for Agentic Systems
Large language models are increasingly used as agentic workflow executors, yet existing training data and benchmarks largely assume informationally complete, single-turn queries. Our analysis of 16K real-world sessions shows that 75.9% of interactions are multi-turn, revealing a substantial gap between how users interact with agents and how such systems are trained and evaluated. We introduce PersonaForge, a user simulation framework for synthesizing realistic multi-turn user–agent interactions. PersonaForge combines a four-dimensional persona space, SOUL-driven behavioral control calibrated to real-user statistics, and Reverse Deep Construction grounded in authentic seed queries. Using PersonaForge, we construct a 6.3K-record training dataset and PersonaForge-Bench, a manually annotated 138-task benchmark spanning over 20 professional domains with four-dimensional scoring. Experiments on Qwen3.5-27B show that PersonaForge training improves the composite score by +4.1%, with gains across all four dimensions and the largest improvements in Task Completion (+6.0%) and Response Quality (+6.8%). Further analyses show that PersonaForge-trained agents use fewer turns and tool calls, suggesting improved interaction efficiency, while ablations confirm the contribution of SOUL components and adaptive simulation. Together, PersonaForge and PersonaForge-Bench establish a foundation for training and evaluating agents under realistic multi-turn user interaction.
Read the full abstract
High Energy Physics (Experiment)
Search for scalar leptoquarks produced via muon-quark scattering in proton-proton collisions at $\sqrt{s}$ = 13 TeV
The first search for TeV-scale scalar leptoquarks (LQs) produced in muon-quark ($\mu$q) interactions is presented. It is based on proton-proton collision data recorded at a center-of-mass energy of 13 TeV with the CMS detector at the LHC, corresponding to an integrated luminosity of 138 fb$^{-1}$. Quantum fluctuations inside the proton generate charged-lepton components, enabling the study of lepton-induced processes. In particular, the interaction of a muon from one colliding proton with a quark from the other proton enables the study of resonant production of a single LQ and its subsequent decay. The final state includes one or two muons and a jet that can come from the hadronization of either a light quark (u) or a b quark. The primary observable is the invariant mass of the muon-jet system, whose distribution peaks at the LQ mass. The data are found to be in agreement with the background predictions. Upper limits are set on the product of the LQ production cross section and its decay branching fraction to the $\mu$q final state. The results exclude broad regions in the parameter plane of the LQ mass and the LQ-$\mu$q coupling, improving upon previous CMS searches by extending the coverage for LQs with masses between 1.5 and 3.6 TeV, and couplings between 0.2 and 0.6 for the LQ(u$\mu$) scenario; and masses above 1.8 TeV, and couplings above 1 for the LQ(b$\mu$) scenario.
Read the full abstract
Mathematical Physics
Critical Points of Line Restrictions of Signed Point-Charge Potentials
Let $\alpha>0$ and restrict a finite inverse-power potential with arbitrary real coefficients to a line. Combine terms having the same projected centre and squared height $\left(a, b^2\right)$, delete classes whose coefficient sum is zero, and let $m$ be the number of remaining classes. We prove the following dichotomy. If $m=0$, exact cancellation occurs if and only if every class sum vanishes, and every ordinary point of the original domain is critical. If $m \geq 1$, there are at most $2 m-1$ critical points: when all effective heights are positive the zeros are counted with analytic multiplicity, while in the presence of effective sources on the line the assertion is one global distinct-point bound. This proves the signed line conjecture of Gabrielov-Novikov-Shapiro, valid throughout their range and in fact for every $\alpha>0$. The structural input is a projective paired Haar theorem: for finite $\beta>1$, the full $2 m$-dimensional space $\sum L_j / Q_j^\beta$, with pairwise nonproportional positive-definite binary quadratics and arbitrary real linear numerators, has at most $2 m-1$ projective zeros counted with multiplicity. For positive charges, the substitution $p=2 \alpha$ proves Conjecture 3 of Edelsbrunner-Fillmore-Oliveira throughout its stated range $p \geq 1$ and extends the same conclusion to every $p>0$. For every $n$, an explicit positive $n$-charge configuration attains $2 n-1$ simple critical points.
Read the full abstract
Machine Learning (Statistics)
Localizing Global Discrepancies: Marginal Contributions and Contextual Anomaly Detection
Global goodness-of-fit and discrepancy statistics can establish that a sample departs from a reference distribution without identifying which observations drive the departure. We develop a framework for this localization problem by assigning to each observation its conditional or marginal contribution across random statistical contexts. This connects resampling diagnostics and data valuation to projection theory and event-level anomaly detection. For symmetric statistics, fixed-size replacement is exactly equivalent to centered conditional localization. For U-statistics, the addition score equals the first Hoeffding/Hájek contribution; for smooth distributional functionals it is related at leading order to the influence function; and for unbiased known-background MMD it reduces exactly to the MMD witness. This viewpoint also yields more efficient estimators. Matched-context subtraction removes fluctuations unrelated to the observation, while for pairwise MMD the event-containing terms give a simple localizer. On the LHC Olympics anomaly-detection benchmark, the pair estimator converges to the direct empirical MMD witness with the predicted 1/(Rm^2) scaling, where m is batch size and R the number of batches. At m=1000 and R=5x106 it reaches correlation 0.9993 with essentially identical AUC. We also ask when context contains information beyond an event's own features. In a shared-latent toy model, the full single-event signal and background distributions are identical by construction, forcing isolated-event AUC=0.5. Discriminating information survives only in cross-event dependence induced by the shared latent parameter; the ensemble recovers this information, whereas an independent-latent control does not. This separates two roles of context: efficient localization of a global discrepancy and genuinely additional class information when the alternative contains shared structure.
Read the full abstract
Cosmology and Nongalactic Astrophysics
PIFFLE: Characterizing Foreground Dispersion Measure Contributions Across 4 Decades in Halo Mass in the FRB20230907D Field
We characterize the foreground environment of FRB20230907D, localized to a galaxy at $z=0.464$, which has an observed dispersion measure of ${\rm DM}_{\rm obs}=1031~{\rm pc~cm^{-3}}$. At its redshift, FRB20230907D lies above the Macquart relation, the expected relation between cosmological dispersion measure and the source redshift, indicating a substantial excess DM along this line of sight. We use Subaru/PFS and SDSS spectroscopy, published group catalogs, Rubin/LSST imaging, and eROSITA X-ray data to characterize the foreground structures that may account for this excess. A friends-of-friends search identifies a massive foreground system at $z\simeq0.09$ with $M_{\rm vir}\simeq7.7\times10^{14}~M_\odot$, while low redshift catalogs reveal an additional group at $z\simeq0.02565$. Assuming that the halo gas follows a modified-NFW halo density profile, we estimate observer frame contributions of $205^{+30}_{-25}~{\rm pc~cm^{-3}}$ and $80^{+60}_{-40}~{\rm pc~cm^{-3}}$ from these systems, respectively. After accounting for the remaining foreground and host contributions, we are able to make the first-ever constraints on the dispersion measure of a closely-intersected $M_*\approx 10^8\,M_\odot$ dwarf galaxy: ${\rm DM}_{\rm UGC7596}=58^{+53}_{-40}~{\rm pc~cm^{-3}}$ at $b_\perp \approx 0.03\,R_{\rm vir}$, which is consistent with hydrodynamical predictions for low mass halos. This highlights the enhancement of dense foreground spectroscopy and multi-wavelength data toward the interpretation of FRB dispersion measures.
Read the full abstract
Human-Computer Interaction
AI as Teammate: Rethinking Task Distribution in Medical Training
Integrating Artificial Intelligence (AI), particularly generative AI, into medical training has prompted concerns about learner over-reliance, misuse, and erosion of foundational clinical competencies. We propose a conceptual reframing at the decision level: the problem is not misuse but misclassification - a mechanistic failure of real-time metacognitive evaluation in selecting a subzone-inappropriate AI interaction mode. Drawing on "SCAN" (Substitute, Complement, Aid, Non-Negotiable), a human-centric decision-making framework for generative AI task allocation grounded in Vygotsky's Zone of Proximal Development and metacognition, we advance the emerging social-constructivist conversation around AI in medical education by offering a testable account of AI's role in clinical reasoning development. This framework yields testable predictions for how misclassification can be detected, mitigated, and, more importantly, prevented in the clinical learning environment. Regarding clinical reasoning development, we show how trajectories of skill acquisition (upskilling) and failure (the triad of skill failure: de-skilling, never-skilling, and mis-skilling) operate at the individual task level in ways that fixed-phase, cohort-wide treatments fail to capture. We further identify passive engagement within correctly classified AI-scaffolded tasks as a particularly insidious, detection-resistant pathway to mis-skilling - one requiring subzone re-identification from AI assistance to expert assistance, with human experts serving as epistemic auditors. The paper operationalizes SCAN for clinical curriculum design, supervision, and assessment, and opens an empirical research agenda grounded in cognitive science. This paradigm shift from misuse to misclassification is not semantic: it offers educators a clear perspective on what to look for, what to assess, and what to intervene on.
Read the full abstract
Algebraic Geometry
Local inequalities for $cD$ and $cE$ singularities
We obtain inequalities for isolated $cD_n, cE_6, cE_7$ and $cE_8$ singularities, analogues of the $4\mu^2/(n+1)$-inequality for isolated $cA_n$ singularity in \cite{KOP}. These inequalities are sharp. This enables us to prove the birational rigidity of certain families of Fano $3$-fold weighted hypersurfaces, which contain only terminal quotient singularities and one $cD_n$ singularity.
Read the full abstract
Computer Vision
Real-Time Musculoskeletal Surrogates for Pediatric Cerebral Palsy: a Credibility Pilot
Real-time musculoskeletal (MSK) surrogates could support personalized rehabilitation for children with cerebral palsy (CP), but their credibility depends on subject-wise evaluation, low inference latency, and calibrated uncertainty. We develop a subject-conditioned causal neural surrogate using OpenSim-derived static parameters, temporal joint kinematics, true muscle capacities, and training-only perturbations. On a real pediatric CP gait dataset comprising nine children, we use leave-one-subject-out validation on six development subjects and evaluate a frozen configuration once on three locked test subjects. The surrogate accurately reproduces musculotendon lengths (R-square = 0.92 in development validation and approximately 0.95 on locked subjects; nRMSE < 8%) while requiring only sub-millisecond to few-millisecond neural inference, well below a 100 ms interactive-rehabilitation target. In contrast, direct muscle-force estimation remains unstable at this small, heterogeneous scale: pooled metrics can overstate within-subject, per-muscle accuracy. A Monte Carlo credibility pilot further shows that propagating only +/-5% anthropometry and muscle-capacity variation produces severely overconfident nominal 90% intervals (approximately 4% force coverage and below 1% MT-length coverage). These results establish a leakage-free evaluation and credibility framework for pediatric MSK surrogates, while identifying force modeling and epistemic uncertainty as the central next challenges for clinically credible digital twins.
Read the full abstract
Solar and Stellar Astrophysics
A high-frequency type II radio burst associated with an X2.3 class flare
Radio observations provide a powerful diagnostic of the solar corona, enabling investigations of dynamic phenomena associated with solar flares, coronal mass ejections (CMEs), and shock waves. We present a multiwavelength analysis of a rare high-frequency type II radio burst (starting frequency of $\sim$750 MHz and frequency drift rate of $\sim$0.5 MHz s$^{-1}$) associated with the X2.3-class solar flare that occurred on 6 November 2024. A propagating EUV disturbance was observed shortly after the flare onset in the SDO/AIA field of view, while radio spectrographs recorded the type II burst between 13:46 and 13:56 UT over a frequency range of $\sim$750 to 45 MHz. Radio imaging observations from the Nançay Radioheliograph (NRH) show that the radio sources propagate southward during the event. X-ray spectroscopy from HEL1OS onboard Aditya-L1 and imaging observations from STIX onboard Solar Orbiter reveal signatures of efficient non-thermal electron acceleration associated with the flare. An NLFFF extrapolation identifies a pre-eruptive magnetic flux rope in the source region, while white-light coronagraph observations obtained during the event show no detectable large-scale CME. The speed of the erupting flux rope, derived from stereoscopic EUV observations, is consistent, within measurement uncertainties, with the shock speed inferred from the radio dynamic spectrum. Together, these observations suggest that compact flux rope eruptions, even in the absence of a detectable white-light CME, can generate low-coronal shocks capable of producing high-frequency type II radio emission.
Read the full abstract
History and Overview of Mathematics
The Mathematician Leads: Building a Research Profile in the Age of Frontier AI
I propose that mathematicians who use AI in research maintain a research profile: a record of what the system has learned about working with them, where those lessons came from, and whether they remain useful. I describe how such a profile grew out of my own mathematical work, through corrections, failed approaches, and repeated attempts to keep difficult ideas from drifting toward easier ones. The profile is not a portrait of the mathematician and does not replace mathematical judgment. It belongs under the mathematician's control and must remain open to revision. I invite others to develop profiles of their own and report what they learn.
Read the full abstract
Computers and Society
It Takes Three to Converse: Empirical Observations on How the Developer, the Convener and the Participant Shaped 119 Polis Conversations
Polis is a popular democratic innovation tool that allows asynchronous citizen engagement through atomic statements: short statements that together describe a complex question, inviting the citizen to vote Agree or Disagree on each. This paper uses 119 conversations with 100 or more participants and an extensive data export, drawn from a wider set of 271 collected processes. The paper asks what determines the output of such a process. Three parties shape the result. The developer of the platform has made important design choices that restrict the outcome: the number of groups the platform is able to report (restricted to 2–5) and which statements are prioritized. The convener defines the assignment: the initial statements that set the tone, the policy that accepts or rejects new statements and who can be invited. Finally, the participant works within these boundaries. With access to less than half of the generated statements, they end up responding to more statements when their conversation seems to have an achievable number of statements to complete, than when they are presented with more statements. Due to choices such as warm path clustering, the exported resulting clustering cannot be reproduced based on the voting data. Conveners may want to re-analyse their own conversations once the process is closed, to consider the data in its entirety, and make their own analysis priorities explicit.
Read the full abstract
Plasma Physics
Electromagnetic pressure-gradient-driven instabilities with moderate high mode numbers in tokamak plasmas
Pressure gradient-driven instabilities are investigated in tokamak plasmas using the global gyrokinetic code EUTERPE emphasizing the role of moderate high mode numbers. As the normalized plasma pressure $\beta$ is increased, there is a well-known, sudden transition from ion-temperature-gradient (ITG) instabilities to kinetic ballooning modes (KBM), if the magnetohydrodynamic (MHD) geometry is held fixed. However, if the equilibrium field is recomputed for each value of $\beta$, so that the equilibrium is consistent with the stability calculation, the transition can disappear. In a number of cases, we are only able to find an ITG-KBM transition if inconsistent equilibria are used. In the MHD unstable regime, gyrokinetic simulations and MHD stability calculations show good agreement for moderate ratios of ion temperature gradient to density gradient and small values of the ion gyro-radius. Otherwise, non-MHD contributions are important, such as the diamagnetic stabilization and ion wave–particle resonant effects.
Read the full abstract
Computational Physics
Real-Time Monitoring of MHD Liquid Metal Flows with Shallow Recurrent Decoders
State estimation in magnetohydrodynamic flows is critical for real-time monitoring of liquid metal blankets in tokamak fusion reactors. Due to the multiphysics nature of these phenomena, high-fidelity simulations are computationally prohibitive for real-time applications. This work investigates a data- driven Reduced Order Model framework: the Shallow Recurrent Decoder (SHRED) coupled with Principal Component Analysis, to map sparse temperature measurements to the full thermo-hydraulic system's state. The major contribution of this work lies in the two-parameter analysis of a fully three-dimensional domain representative of the DEMO breeding blanket configuration. Here, the flow is subjected to an external magnetic field varying in direction and intensity and is hindered by two cylinders acting as a water-cooling system, which impose a temperature boundary condition on their surfaces. This double-parametric magnetic variation induces nonlinear transitions in the flow dynamics, ranging from chaotic behavior at low magnetic field intensities to laminarized regimes at high intensities, characterized by the formation of asymmetric side layers at an inclination angle of 30 degrees. SHRED reconstruction maintains a mean relative error of approximately 5% for the temperature, pressure, and velocity fields. This accuracy is maintained across both weak and strong magnetic fields, ranging from 0.075 T to 0.300 T, and for inclination angles from 5 to 30 degrees, reflecting its dominant toroidal component. These errors are only slightly larger than the lower error bound dictated by low-rank truncation. The results establish SHRED as a reliable state estimator for complex and realistic engineering applications involving completely unseen parametric scenarios and validate it as an accurate real-time state estimation technique suitable for online monitoring and control of real facilities.
Read the full abstract
Materials Science
Topological Signatures of Hardness and Structural Order in Network-Forming Materials
Understanding how the topology of network forming materials influences their physical properties remains a longstanding challenge. Here, we investigate the topology of experimentally compatible atomistic models of amorphous silica together with the crystalline polymorphs cristobalite and quartz. By comparing multiple atomistic models of amorphous silica that equally reproduce neutron-scattering data, we show that different microscopic descriptions can imply different topological interpretations of glass stability. To characterize the network beyond conventional geometric descriptors, we introduce the linking valence, which quantifies the average number of topological links per network loop. This topological descriptor separates silica into two distinct classes: the mechanically harder quartz exhibit values more than an order of magnitude larger than those of amorphous silica and cristobalite, despite their common tetrahedral building blocks. Spectral analysis of the loop-linking networks provides a complementary distinction, separating amorphous from crystalline phases and revealing differences in the long-range organization of topological constraints. These results establish topological linking as a new framework for connecting the structure and physical properties of network-forming materials.
Read the full abstract
Software Engineering
Where Does Balance Break? Boundary Discovery for Game Balance Testing under a Finite Simulation Budget
Software testing often relies on assumptions such as reproducible executions and stable correctness criteria. However, many modern software systems exhibit non-deterministic executions and large behavior spaces, making exhaustive exploration impractical and single-run judgments unreliable. These characteristics make it difficult to identify where acceptable behavior ends and problematic behavior begins. Competitive multiplayer games represent a challenging instance of such systems, where balance must be maintained so that no single strategy dominates. Even small parameter changes can trigger abrupt balance disruption, yet detecting such failures requires repeated simulations under non-deterministic outcomes and high-dimensional parameter spaces. In this paper, we formulate game balance regression testing as a boundary-discovery problem under a finite simulation budget. The objective is to efficiently identify inputs near the boundary that separates balanced and unbalanced regions. To address this problem, we propose BBExplorer, which combines multi-directional candidate generation, budget-aware two-stage screening, and adaptive step-size shrinkage for boundary refinement. Experimental results on two games with different levels of complexity show that the approach is strong in low-dimensional settings and remains effective in higher-dimensional ones. It also exhibits stable boundary behavior across unseen random seeds and threshold settings. These results indicate that BBExplorer is effective for practical balance regression testing and, more broadly, for boundary-oriented testing in non-deterministic, budget-constrained systems.
Read the full abstract
Artificial Intelligence
EvoUndo: Recoverability-Constrained Self-Evolution for LLM Agent Harnesses
LLM agents increasingly modify their own prompts, tools, middleware, resources, and execution harnesses at runtime. Such self-evolution can improve capability, but a successful mutation may leave persistent effects that cannot be safely reversed in states different from the one in which it was created. We introduce EvoUndo, a framework for representing, synthesizing, diagnosing, and independently verifying recoverability of model-generated self-modifications across counterfactual states. Across 600 unseen one-shot self-evolution tasks, we identify 197 capability-improving mutations that fail recoverability verification. Under the original recovery representation, conventional repair strategies recover 0/197 of these natural failures. Deterministic oracle analysis recovers 48/197 under the original recovery language L0, while the extended recovery calculus increases empirical oracle recovery to 191/197. A protocol-locked 2x2 grounding-by-expressivity intervention then separates two bottlenecks: exact state-address grounding increases successful recovery from 0/48 to 38/48 (79.2%) when the original language is sufficient, while extending the recovery language enables recovery on 142/143 (99.3%) failures in the oracle-defined S1 stratum. On the primary gpt-oss-120b backbone, adding exact-address diagnostics to the richer language reduces recovery to 133/143 (93.0%); a Qwen3.8-27B replication preserves the grounding and expressivity effects but not this negative interaction, indicating that the latter is model-dependent. These results indicate that reliable agent self-evolution requires co-designing verification, state grounding, witness semantics, and recovery-language expressivity rather than relying on iterative prompting alone.
Read the full abstract
Cryptography and Security
Optimal Adversarial Testing: Extracting Honest Test Results from Dishonest Test Takers
In applications, it is often required to test objects or people to determine their qualities in terms of certain metrics. However, besides being naturally noisy, the test results can be corrupted by adversarial behaviors of objects or people being tested (test takers). For example, dishonest test takers can cheat in the exams to distort the test results. With the development of AI technologies, such distortions driven by cheating using AI technologies are becoming more commonplace and severe. In this paper, we propose optimal testing strategies which can still recover needed test results even if there are cheaters polluting the results. The proposed testing strategies will optimally re-test selected group of test takers using different testing security measures. We determine the optimal testing strategies using a dynamic programming method.
Read the full abstract
Artificial Intelligence
GRACE:Gradient-guided Coreset Selection for LLM Unlearning
Machine Unlearning methods for Large Language Models typically assume pre-specified forget and retain sets. In realistic settings, however, requests may provide only a few examples of undesired behavior, requiring forget and retain sets to be inferred from heterogeneous corpora. We study this data-selection problem and propose GRACE , a gradient-guided coreset selection method that constructs both forget and retain sets for LLM unlearning. GRACE first computes a forget direction from seed examples that elicit the undesired behavior, then selects a compact forget coreset whose gradients approximate this direction using non-negative orthogonal matching pursuit. To preserve model utility, it selects retain examples after projecting out the forget direction and applying clustered orthogonal matching pursuit in the remaining gradient space. Across two target domains, two model families, and four unlearning algorithms, GRACE improves model utility while maintaining comparable forget quality, with particularly consistent gains over prior gradient-based selection methods.
Read the full abstract
Artificial Intelligence
Propagating construction-time knowledge quality into medical question answering: A framework grounded in clinical guidelines
Large language models have facilitated knowledge graph (KG) construction from clinical guidelines, but extracted triples vary in structural validity and evidential support. Meanwhile, graph-augmented question answering (QA) systems typically optimize query relevance during retrieval, with limited reuse of quality information produced during KG construction. This creates a disconnect between construction-time quality control and inference-time evidence use. We investigate whether construction-time triple quality can serve as a persistent signal for downstream evidence selection and presentation. We propose a quality-aware framework that models structural conformance (SchemaConf) and evidential support (EvidScore) as complementary dimensions and fuses them into a per-triple quality signal, Q(t). Rather than using quality solely for filtering, the framework retains Q(t) and derived quality tiers as graph attributes and propagates them into quality-weighted subgraph retrieval and tier-conditioned evidence prompting, while preserving passage-level provenance. Experiments on Chinese diabetes clinical guidelines show that the utility of the quality signal is distribution dependent. Under cross-version and cross-model shift, the fused Q(t) provides stronger triple-quality discrimination than either component alone (AUC 0.748 vs. 0.703 for EvidScore and 0.645 for SchemaConf). In guideline-grounded QA, propagating construction-time quality reduces required-knowledge omission from 16.3% to 5.3% and conflicting outputs from 16.3% to 2.7%, with an evidence-grounded precision of 81.6% and near-zero invalid citations. Blinded clinician ratings favor the full framework over no retrieval (4.68 vs. 4.21 on a five-point scale) and approach the oracle condition (4.80), while cross-generator experiments show consistent trends.