Mathematical Physics
On the Cauchy Problem for the Sawada-Kotera Equation: Soliton resolution conjecture, Painlevé transcendents and asymptotic stability
Zheng-Kang Huang and Shou-Fu Tian
We derive long-time asymptotics in different space-time regions for the Sawada-Kotera (SK) equation with admissible data involving continuous spectrum and finite discrete spectrum. Our result proves the
soliton resolution
conjecture, yields the asymptotics governed by the F-XVIII Painlevé transcendent, and establishes the asymptotic stability of \(N\)-
soliton solutions. More importantly, this extends the long-time asymptotic theory to the coexistence of
solitons and radiation, whereas previous results were restricted to the purely continuous spectrum. One key ingredient in our argument is a new meromorphic Riemann-
Hilbert framework that incorporates the discrete spectrum through complete six-point pole orbits while preserving cofactor analyticity and the singular structure at the spectral origin. A second ingredient is the combination of an orbitwise, region-dependent pole reduction mechanism and the \(\bar{\partial}\)-nonlinear steepest descent method, which allows us to select the
solitons contributing in each region and separate them from the continuous radiation. Our framework also resolves the coalescing stationary points at the singular spectral origin through the regular modified SK problem and the Miura map.
Read the full abstract ▾
Algebraic Geometry
Multiplicativity of the perverse filtration in the $\mathrm{SL}_n$ case
Xueqing Wen
We prove that the perverse filtration of the canonical $\mathrm{SL}_n$ Hitchin system is multiplicative when the rank and degree are coprime. Over the
elliptic locus of an even-degree twisted Hitchin system, we compare the
Fourier projectors for compactified Pryms with the endoscopic correspondence. We then use the square-zero property of the variant
cohomology and a
parabolic Higgs moduli reduction to pass to the full canonical
moduli space.
Read the full abstract ▾
Artificial Intelligence
Peer Influence across Heterogeneous AI Models
Frida Nøhr Laustsen, Marie Haahr Petersen, Victoria Popa et al. (+4 more)
When two AI
agents disagree, who persuades whom? As multi-
agent systems increasingly combine
language models of different families and sizes, the answer can determine which judgments survive interaction. Measuring persuasion as the probabilistic shift in an
agent's decision after a single exchange with a dissenting peer, we test seven open-weight models across three
language understanding tasks. We find that persuasion is strong: when models disagree, receivers often abandon their initial judgment after seeing a peer's answer and explanation. Surprisingly, however, neither standalone certainty nor model scale reliably predicts persuasion dynamics. Models producing almost perfectly consistent decisions in isolation can be among the most susceptible to persuasion, and small models can match larger ones as persuaders and resist their influence just as effectively. Furthermore, we show that the size of the shift depends more on the susceptibility of the listener than on the persuasiveness of the speaker. Persuasion patterns are therefore specific to each model pairing, with heterogeneity amplifying persuasion in some combinations and suppressing it in others, allowing a dissenting
agent running a small model to overturn the judgments of a much larger one. These findings show that the behavior of interacting models cannot be inferred from their individual properties but must be evaluated in the combinations in which they will operate.
Read the full abstract ▾
All authors Frida Nøhr Laustsen, Marie Haahr Petersen, Victoria Popa, Ariel Flint, Romualdo Pastor-Satorras, Andrea Baronchelli, Luca Maria Aiello
Superconductivity
Quantum critical superconductivity in a dense fermi dilute bose mixture close to mechanical instability
Vibhu Mishra
We couple a metal to an uncondensed bose gas with repulsive bose-
fermi interactions in two dimensions. Within the random phase approximation, we show that the density fluctuations of the
bosons, which mediate the effective
fermion-
fermion attraction, become long ranged on appropriate tuning of interactions. This leads to parametrically enhanced superconducting temperatures $T_c \propto g$ as opposed to $T_c \propto \exp(-1/g)$ for the BCS case. The dilute bose gas density acts as the primary bottleneck to the enhancement. We discuss the conceptual difficulties of implementing this idea in electron-
exciton mixtures.
Read the full abstract ▾
Machine Learning
LS-AR: Future-Predictive Latent Steering in Autoregressive LLMs
Anubha Gupta and Eduardo Pignatelli
Standard autoregressive (AR) models process high-level task instructions, state history, and transient
tokens within a single shared sequence of
tokens. Consequently, they lack the architectural mechanisms needed to isolate macro-objectives from context noise. To overcome this single-channel limitation, we introduce Latent-Steered Autoregressive (LS-AR), a dual-channel architecture that decouples continuous goal steering from discrete
token decoding via FiLM conditioning. We evaluate a Static Goal Encoder (P_0) for persistent macro-objective retention across long rollouts and a Dynamic State Tracker (P_t) for recurrent latent updates during generation. On long-horizon
retrieval past context limits (H=1024, W=500), LS-AR (Static) achieves 100% target recall where parameter-matched baselines collapse (0%), while increasing throughput by ~35% and cutting peak VRAM by 52.8%. In Blocksworld planning under forced perturbations (k=1), LS-AR (Dynamic) sustains an 89.0% completion rate vs. 71.0% for the baseline, though zero-shot entity scaling (N -> N+1) exposes single-vector capacity limits (0%). Finally, dual-channel authority analysis shows that text goal dropout establishes latent-dominant control, offering structural defence against text
prompt injection while introducing a latent vector attack surface.
Read the full abstract ▾
Machine Learning
ULTRADISCOVERY: Abductive Exploration in an Interconnected, Epistemically Open Universe
Weihan Li, Tianshi Zheng, Yangqiu Song et al. (+2 more)
Scientific discovery often begins when scattered clues call for a new way of describing the world. Such abductive exploration can require constructing the representation in which an explanation is stated, when the world is epistemically open, and composing evidence scattered across contexts, when it is structurally interconnected. Existing
benchmarks rarely separate these two demands or control them independently. We introduce ULTRADISCOVERY, an interactive world of five domains in which an
agent revises an initially successful theory and predicts the outcome of an unseen cross-domain intervention. A $2 \times 2$ design leaves the representation open or discloses it, and leaves the evidence distributed or aligns it, with the latent dynamics fixed. With the representation open,
agents across eleven models often retract the axiom they were taught, and none introduces the unobserved entity or rewrites the variables that a replacement requires. Disclosure triples intervention requests and adds about one of the eighteen findings the world affords, and
alignment adds less. Two vendor-harness systems carry discovery into more domains, and one of them rewrites the variables in Open episodes. No system makes the exact prediction within 200 paid actions. At larger budgets one exact prediction appears with both aids, while every Open episode remains inexact. The results locate the difficulty in the step from accumulating evidence to composing it into a representation that transfers.
Read the full abstract ▾
All authors Weihan Li, Tianshi Zheng, Yangqiu Song, Ginny Y. Wong, Simon See
Analysis of PDEs
Singular Contour Dynamics and Paradifferential Reduction
Xingyu Li, Emeric Roulley and Stefano Scrobogna
We develop a direct paradifferential approach to nonlinear graph integral operators arising in contour dynamics. The main methodological result is an abstract structure-preserving paralinearization
theorem, valid in arbitrary dimension, for a class of homogeneous kernels whose nonlinear dependence on the graph is expressed through normalized finite differences. The
theorem covers both critical principal-value kernels and subcritical locally integrable kernels, gives explicit formulas for the principal symbols, and separates the finite-order symbolic contribution generated by the homogeneous singularity from order-zero far-field terms and arbitrarily smoothing remainders. A structural feature of the reduction is the cancellation of the intermediate symbolic order; at the critical endpoint the construction also preserves the exact skew-adjoint structure of the original singular integral. As a main application, we consider the three-dimensional two-phase free-boundary Euler equations with constant interfacial background vorticity. We first derive an autonomous contour-dynamics equation in canonical surface variables and then apply the abstract
theorem directly to the singular integral operators generated by the Birkhoff–Rott formulation. This yields the complete paradifferential structure of the system without taking the Dirichlet–Neumann reduction as a starting point. After introducing an Alinhac good unknown and diagonalizing the resulting system, we obtain local well-posedness for small Sobolev perturbations in the stable Kelvin–Rayleigh–Taylor regime, uniformly on compact subsets of the stable parameter set.
Read the full abstract ▾
Distributed Computing
PaNGEA: Parallel Node Generation and Exploration Algorithm on GPU
Jean Pauphilet and Yupeng Wu
Primal heuristics for finding high-quality feasible solutions are an important component in mixed-integer optimization (MIO) solvers. Recent advances in GPU-accelerated optimization
algorithms show the potential of GPU acceleration for continuous optimization. In this paper, we introduce the Parallel Node Generation and Exploration
Algorithm (PaNGEA), a GPU-friendly MIO primal heuristic. PaNGEA explores restricted subproblems by combining linear-relaxation solves with a local-search procedure designed for efficient batched execution on GPUs. In addition, instead of relying on a single heuristic for iterative variable fixing, we leverage GPU batching capabilities to generate and explore multiple subproblems in parallel. PaNGEA leverages GPU capabilities in two ways. First, on 283 MIPcc26 and MIPLIB instances, implementing a single-node primal heuristic on the GPU reduces the average gap integral by 8-18% relative to its CPU counterpart. Second, generating and exploring multiple nodes in parallel further reduces the average gap integral by 12-19%.
Read the full abstract ▾
Cryptography and Security
Securing Computer-Use Agents Against Branch Steering Attacks
Giulio Zingrillo, Hanna Foerster, Ilia Shumailov et al. (+2 more)
Modern Computer Use
Agents (CUAs) directly interact with graphical user interfaces and execute third-party web tools, exposing them to indirect
prompt injection across every rendered page and tool response. While the Dual-LLM pattern is the primary system-level architecture offering formal security guarantees - using an isolated Planner LLM (P-LLM) to fix execution paths before processing untrusted inputs via a Quarantined LLM (Q-LLM) - these guarantees break down in graphical environments. Because CUA interaction is inherently dynamic, plans cannot remain data-independent; they must branch based on anticipated runtime web content - covering all possible cases the
agent may encounter. This exposes
agents to branch steering attacks, where an adversary crafts untrusted data to coerce a CUA down a hazardous, pre-approved branch without injecting explicit instructions. We systematically study branch steering attacks and introduce STEER-Bench (101 tasks across 9 domains), showing high attack success against both standard (94.4%) and vanilla Dual-LLM (89.5%) CUAs. We then propose COBRA, an architecture that pairs trusted branching plans with ahead-of-time capability constraints, strictly bounding the parameters and destinations each branch may execute. On STEER-Bench, COBRA reduces attack success to 0% while retaining 97% benign utility.
Read the full abstract ▾
All authors Giulio Zingrillo, Hanna Foerster, Ilia Shumailov, Yiren Zhao, Robert Mullins
Distributed Computing
Coda: Exploiting Admission Flexibility for Coding-Agent Serving
Youhe Jiang, Fangcheng Fu, Binhang Yuan et al. (+5 more)
Coding
agents powered by large
language models (LLMs) repeatedly alternate between model inference and tool calls, creating long-lived sessions with reusable key-value (KV) states and asynchronous request resumptions. Logical readiness, however, does not ensure efficient admission in a shared serving system. Through direct trace analysis and trace-driven replay, we identify two mismatches: reusable KV states reside across storage tiers and incur unequal computational costs for state preparation, while requests with heterogeneous context lengths can decode inefficiently together. Our central insight is that exploiting admission flexibility can improve serving performance while preserving request progress. We present Coda, a coding-
agent serving system that realizes this insight through a readiness-informed admission layer incorporating two mechanisms. Tiered-Aging state admission exploits bounded flexibility in admission order to improve admission efficiency and preserve request progress. Compatibility-Aware execution admission exploits flexibility in
attention grouping within each model iteration to reduce mixed-context interference and improve shared decoding efficiency. For multi-worker configurations, Coda introduces a separate routing layer that considers KV-state residency, worker load, and request-to-worker context-length compatibility to guide efficient request placement. We evaluate Coda across different models and workloads against state-of-the-art coding-
agent serving systems and vLLM. Across the single-worker and multi-worker experiments, Coda improves output-
token and SLO-compliant throughput by 20.3% and 70.5% on average, with peak gains of 29.3% and 140.2%.
Read the full abstract ▾
All authors Youhe Jiang, Fangcheng Fu, Binhang Yuan, Krishna Malladi, Ehsan K. Ardestani, Zhan Shu, Adnan Aziz, Yi Xu
Machine Learning
Zephon: Elastic Determinism for Online, Stateful Foundation Model Data Loading Pipelines
Maximilian Böther, Josh Wills, Ties Robroek et al. (+14 more)
Deterministic data loading is important for foundation model development: model researchers need confidence that differences they observe across costly ablations are caused by the parameter they changed rather than non-determinism in the training data sequence. The data loader must provide elastic determinism, i.e., a deterministic sequence of global training data batches despite changes to the GPU
topology across runs (e.g., due to GPU scarcity), frequent checkpoint-resume cycles, and different data processing execution backends. Achieving this is difficult because modern foundation model data pipelines tokenize, pack, and mix samples online, introducing stateful n-to-m transformations that break sample indexing. Existing data loaders largely assume indexable 1-to-1 pipelines, and the common workaround of offline materialization is expensive and, for some modalities such as
video, infeasible. We present Zephon, a data loader for foundation models that supports online, stateful pipelines while providing elastic determinism and efficient resumption from checkpoints. It partitions the global stream into
topology-independent lanes, serializes ordering decisions while parallelizing stateless work on interchangeable backends, and checkpoints only bounded in-flight state so recovery cost does not grow with training progress. We evaluate Zephon on text and
vision-
language workloads and show that it achieves competitive throughput while providing a combination of guarantees that no existing loader offers for online, stateful pipelines.
Read the full abstract ▾
All authors Maximilian Böther, Josh Wills, Ties Robroek, Sonnet Xu, Paul Burstein, Daniel Zayas, Cody Blakeney, Siddharth Joshi, Haoli Yin, Rishabh Adiga, Haakon Mongstad, Luke Merrick, Pratyush Maini, Ari Morcos, Matthew Leavitt, Ana Klimovic, Bogdan Gaza
Combinatorics
Small Influential Coalitions in $[0,1]^n$ via the Junta Theorem
Ehud Friedgut
Every monotone Boolean function on the continuous cube admits a coalition of $O(n/(\varepsilon\log n))$ coordinates that can force a fixed output (either zero or one) with probability at least $1-\varepsilon$. This old
conjecture of mine was recently proven by Chattopadhyay and Gurumukhani~\cite{CG}. This writeup contains a simplification of their proof, generated by AI after suggesting the use of the junta
theorem of~\cite{F98} and the discretization procedure of ~\cite{F04}.
Read the full abstract ▾
Machine Learning
Light Entropic Optimal Transport on Riemannian Manifolds
Xavier Aramayo-Carrasco, Petr Mokrov and Alexander Korotin
Entropic Optimal Transport (EOT) has become a practical framework for learning
stochastic couplings between complex distributions, with applications in generative modeling and domain adaptation. However, most EOT solvers are designed for Euclidean spaces, while
manifold extensions remain limited and often rely on costly iterative methods, simulated dynamics, or generic neural models that do not fully exploit the underlying geometry. We introduce ManifoldLightOT, a light approach for learning kernel-induced EOT couplings directly on common
manifolds. Using the kernel form of the EOT solution, we construct geometry-specific Gibbs kernels together with compatible potential parameterizations for spheres,
tori, $\mathrm{SO}(3)$, and $\mathrm{SE}(3)$. These choices yield closed-form normalization and directly sampleable conditional distributions. Our formulation naturally extends to products of
manifolds, making it applicable to more complex geometries. The parameters of the potentials are optimized directly from samples using Monte Carlo estimates of the learning objective. Through synthetic and real-world experiments, we show that ManifoldLightOT often outperforms existing
manifold OT methods while retaining direct sampling.
Read the full abstract ▾
Computer Vision
NegT2IBench: When Negation Changes the Picture. A Polarity Benchmark for Text-to-Image Models
Omar Elfatairy, Maria A. Bravo, Jessica Bader et al. (+1 more)
Text-to-image (T2I) models are judged by
benchmarks that measure whether requested content appears, but these
benchmarks largely overlook the complementary ability to satisfy negated constraints, for example, generating "a non-red cup." Measuring negation raises challenges not faced by affirmation-based
benchmarks and requires careful
prompt and evaluation design. We introduce NegT2IBench, a
benchmark of 4,800
prompts covering two attribute types and four relation categories.
Prompts are organized by polarity: the number of positive statements that must hold and negated statements that must not, each ranging from 0 to 2. Varying the two independently separates the effect of negation from the effect of
prompt complexity. Our detector-based scoring is reproducible, auditable, and pinpoints which requirement failed. On 600 images with three-annotator labels, it agrees with humans as closely as
vision-
language judges up to 30x larger, while using only a fraction of their GPU
memory. Across eleven T2I models and 211,200 images, nine score lower on a single negated statement than on a single positive one. Per-statement scoring reveals that the loss is largest for color and near zero for proximity, and that 41.5% of failed statements render exactly what the
prompt forbids. Rendering what a
prompt asks for and withholding what it forbids are distinct capabilities that an aggregate compositional score cannot distinguish. NegT2IBench measures the latter directly, providing a controlled testbed for diagnosing negation failures and developing methods to overcome them.
Read the full abstract ▾
All authors Omar Elfatairy, Maria A. Bravo, Jessica Bader, Zeynep Akata
Analysis of PDEs
Döblin–Fourier cancellation and kinetic Aleksandrov estimates
Amélie Loher, Connor Mooney and Clément Mouhot
We develop a cancellation mechanism for linear second-order kinetic equations in non-divergence form with measurable uniformly
elliptic coefficients. We decompose the dynamics into spatial waves and follow two families of velocity paths along which the waves acquire nearly opposite phases. The
parabolic Krylov–Safonov theory gives a common lower bound for the two velocity marginals. The corresponding contributions cancel up to a small phase error, producing a contraction \textità la Döblin}. Iterating this contraction yields exponential
Fourier decay estimates with enhanced dissipation. We then present two applications. The first contribution is a kinetic Aleksandrov estimate: a maximum principle in which the source is measured in an $L^p$ norm. For rough coefficients $A(t,v)$ independent of position, we obtain the estimate for every $p>2n+1$, where $n$ is the dimension of position and velocity. The
Fourier decay also gives spatial smoothness, and
parabolic regularity gives Hölder continuity in time and velocity. For autonomous coefficients $a(x,v)$ in dimension one, position serves as time away from zero velocity. A Harnack comparison controls returns to small velocity intervals and yields the estimate for every $p>4$. Both thresholds are optimal among those valid for all ellipticity ratios. The second contribution concerns position on the
torus and velocity on the sphere, with measurable uniformly
elliptic diffusion coefficients $A(t,v)$ independent of position. We prove enhanced dissipation with the optimal square-root frequency power and Gevrey regularity in position. For autonomous coefficients $A=A(v)$, we also obtain exponential convergence to
equilibrium in $L^2$ weighted by the invariant velocity measure, and a quantitative spectral gap. To our knowledge, these are the first results on decay established for laws of kinetic equations with rough coefficients.
Read the full abstract ▾
Machine Learning
Smart Sensing for Safer Bridges: From Sensor Signals to AI-Driven Anomaly Detection
Rahul Jaiswal, Joakim Hellum and Halvor Heiberg
Bridges contribute significantly to transportation connectivity and urban development. Therefore, reliable bridge monitoring is crucial for protecting public safety and detecting anomalous behavior in bridge sensor data that may provide early indications of abnormal structural conditions. This paper investigates anomaly detection in real-world bridge sensor data using two different complementary approaches, namely signal processing and the data-driven machine learning model Isolation Forest. The real-time bridge sensor data is collected from an iBridge sensor device installed on a bridge in Norway. The methods are evaluated using anomaly counts, anomaly detection time, processing rate, anomaly rates, visualization, and temporal agreement. Moreover, a controlled anomaly-injection analysis is performed to evaluate the sensitivity of each method. Numerical results demonstrate distinct detection characteristics and computational requirements, highlighting the potential of machine learning, particularly the data-driven Isolation Forest, alongside signal processing for identifying anomalies in bridge sensor measurements.
Read the full abstract ▾
Number Theory
Cohomology of Linearly Constrained Kloosterman Families: Newton Polytopes, Weights, and Boundary Monodromy
Hamed Ebadi
Let $p$ be an odd
prime, $\psi$ a nontrivial additive character of $\mathbb F_p$, $\psi_q=\psi\circ\mathrm{Tr}_{\mathbb F_q/\mathbb F_p}$ and $\lambda\in\mathbb F_p^\times$. For $h\in(\mathbb F_q^\times)^n$ let $\Sigma_n(h)$ be the sum of $\psi_q(u_1^{-1}+\cdots+u_n^{-1})$ over $u\in(\mathbb F_q^\times)^n$ with $h_1u_1+\cdots+h_nu_n=-\lambda$, an additively twisted average of products of Kloosterman sums. For every $n\ge2$, the
cohomology computing $\Sigma_n(h)$ is concentrated in one degree and has dimension $2^n-1$. It has $r(n)=2^n-\binom n{\lfloor n/2\rfloor}$
eigenvalues of weight $n-1$, and its other
eigenvalues are the integers $q^{k-1}$, $1\le k\le\lfloor n/2\rfloor$, with ballot numbers as multiplicities. The part of top weight forms a lisse sheaf $\mathcal W_n$ of rank $r(n)$ on the
torus $\mathbb G_m^n$. We determine its local monodromy at the coordinate boundary and its determinant: $\det\mathcal W_n$ is the Artin–Schreier sheaf of the sum $-2^{n-1}\lambda^{-1}(h_1+\cdots+h_n)$ of the critical values of the phase, twisted by $p^{(n+1)r(n)/2}$. The
geometric monodromy group of $\mathcal W_n$, and the arithmetic one after normalization to weight $0$, are equal to $\{g\in\mathrm{GL}_{r(n)}:\det(g)^p=1\}$. Consequently the normalized sums become equidistributed like traces of Haar-random elements of $\{g\in\mathrm U(r(n)):\det(g)^p=1\}$. The proofs use Laumon's
Fourier transform, middle convolution, the Cayley cubic surface and Larsen's alternative, but no classification of finite groups.
Read the full abstract ▾
Software Engineering
MintEval: Do LLMs Implement the Trading Strategy You Asked For? A Behavioural-Equivalence Benchmark for Natural-Language-to-Strategy Code
Siyu Wang, Yifan Wang and Yuecheng He
Large
language models are moving from producing trading signals to writing the code that executes them. The failure mode of the second role is silent: generated code runs, a backtest plots, yet the risk logic that the trader described is not the logic being executed. Existing code
benchmarks test functional correctness on unit tests and finance
benchmarks test
forecasting; neither measures whether an implementation behaves like the strategy that was asked for. We introduce MintEval, a
benchmark in which reference strategies are generated programmatically from a library of composable building blocks, back-translated into colloquial trader instructions, and re-implemented by the model under test. Generated and reference programs are executed bar by bar on identical
market data and frictions, and compared on their actions rather than on code similarity or profit: alpha is differenced away. MintEval v0 contains 800 tasks on BTCUSDT 15-minute data, stratified by an execution-measured state-span complexity tau that is decoupled from description length. Low-cost models reach a mean ActionMatch of at most 0.544 and reproduce at most 0.087 of tasks exactly; on a stratified subset of 200 tasks a frontier model (Claude Opus 5.5) reaches 0.889 and reproduces 0.575 exactly, yet still fails silently on 0.275 of tasks. Given a menu of building blocks, models identify the strategy almost perfectly, yet 79.2% of the implementations whose specification was read correctly diverge on more than 10% of active bars. The LLM judge of a recent strategy-generation
benchmark, applied verbatim, accepts every one of these silent failures.
Read the full abstract ▾
Artificial Intelligence
RIFAR: Reliability-Informed and Forgetting-Aware Replay for Continual Robot Learning with World-Action Models
Tu Lin, Zheng Lu, Zirui Song et al. (+7 more)
Genuine
embodied agency requires
robots to turn continuous real-world experience into lasting, transferable skills. This demands continual learning that integrates new capabilities while preserving prior knowledge in evolving environments. Experience replay mitigates catastrophic forgetting, but storing complete demonstrations becomes costly as tasks accumulate. World-action models provide an alternative through generative replay. However, visually coherent rollouts may contain actions that cannot realize the predicted transitions, while new-task adaptation can disrupt learned behavior. To address these problems, we propose RIFAR (Reliability-Informed and Forgetting-Aware Replay), a reliability-informed and drift-aware generative replay framework. It reconstructs trajectories from compact historical demonstration prefixes and uses a trainable inverse-dynamics model to assess action-visual consistency. It first adapts to new tasks using current-task demonstrations and high-reliability replay, then evaluates how this adaptation changes action predictions on historical inputs and prioritizes reliable trajectories with greater action drift for further training. Across three LIBERO suites and real-world experiments, RIFAR outperforms the previous state of the art in WAM-based generative replay. On LIBERO-Goal, it achieves an AUC of 90.97 while retaining only 320 historical time steps per task, about 4.9% of those retained by 50-demonstration replay.
Read the full abstract ▾
All authors Tu Lin, Zheng Lu, Zirui Song, Zixian Lin, Zebang Feng, Meijia Chen, Xiaoyu Sun, Wanru Li, Lijie Wang, Xiuying Chen
Computation and Language
An automated pipeline for standardised speech-unit annotation in spontaneous dialogue
Hanlu He, Harald Vilhelm Skat-Rørdam, Ingvi Örnólfsson et al. (+1 more)
Quantifying conversational dynamics requires reliable identification of interactional units and their temporal boundaries, but
speech activity alone does not distinguish conversational turns from listener feedback or within-turn pauses. We present an automated pipeline for extracting turns and backchannels from separate-channel recordings of spontaneous dyadic conversation, designed to provide a consistent first-pass annotation for subsequent human review. The pipeline combines voice activity detection, channel-energy filtering, temporal merging, automatic
speech recognition, and context-based post-processing. We evaluated the pipeline on 99 ten-minute Danish conversations from 33 dyads using segment-level detection reliability and temporal boundary error. Conversations were recorded under both normal and asymmetric listening conditions. In the latter,
speech-shaped noise was delivered to one participant through bone-conduction headphones. Overall detection reliability was F1=0.621, with similar performance for turns F1=0.624 and backchannels F1=0.618. For successfully matched segments, median absolute onset and offset errors were 0.150 and 0.160s for turns and 0.130 and 0.180s for backchannels, respectively. Mean errors were substantially larger for turn boundaries, indicating a smaller number of large boundary mismatches. Performance did not differ significantly across the two experimental listening conditions. In a four-conversation case study, pipeline-human agreement was lower and more variable than human inter-annotator agreement and varied across parameter settings. These results support the pipeline as an automated first pass within a semi-automated annotation workflow, providing a consistent basis for more standardised and reproducible annotation of conversational dynamics.
Read the full abstract ▾
All authors Hanlu He, Harald Vilhelm Skat-Rørdam, Ingvi Örnólfsson, Ivana Konvalinka
Computation and Language
Unmasking Propaganda: A Comparative Analysis of Masked and Causal Language Models
Claudiu Creanga, Ioachim Lihor and Liviu P. Dinu
Propaganda detection is an essential task in natural
language processing (NLP), particularly in the context of manipulative political communications. However, identifying specific propaganda techniques presents a significant challenge due to their often subtle nature and reliance on context, making them difficult to distinguish from legitimate persuasive
language. Propaganda often involves highlighting certain facts while downplaying or ignoring others to create a desired perception. This biased communication aims to influence attitudes, beliefs, or behaviors towards a particular cause or position. This paper explores advances in detecting propaganda techniques through a comparative analysis of modern
language models, using the SemEval-2020 Task 11
dataset. We evaluated both masked
language models (based on XLM-RoBERTa or DeBERTa V3) and causal models (from OpenAI, Google, Mistral, Anthropic and Meta), employing two prompting strategies: base and chain-of-thought prompting. Our results demonstrate improvements over state-of-the-art models, with the best-performing MLM achieving an F1 score of 63.18 in technique classification and the best causal model achieving 63.62. We also observed that certain models excel in specific techniques, such as loaded
language and name-calling, while struggling with others like bandwagon and black-and-white fallacy. These findings suggest that fine-tuning, ensemble modeling, and the use of larger
datasets can further enhance propaganda detection capabilities.
Read the full abstract ▾
Computational Finance
Shapley-based Structural Analysis of Neural Calibration for Stochastic Volatility Models
Shaïn Afzali, Serena Della Corte and Antonis Papapantoleon
Neural network-based approaches have emerged as efficient alternatives to traditional optimization-based procedures for the calibration of
stochastic volatility models. However, existing work has focused primarily on predictive accuracy, with comparatively little
attention devoted to understanding the structure of the learned inverse calibration mappings. In this work, we analyze neural calibration mappings for the Heston and rough Heston models across multilayer perceptron, highway, and softmax-parametrized highway architectures, using complementary Shapley-based methods from explainable AI. Specifically, we consider SHAP and $\nu$SHAP explanations, which capture distinct, complementary notions of feature relevance, corresponding to sensitivity and sufficiency of feature subsets, respectively. Short maturities and smile wings consistently dominate parameter inference, and the dominant attribution structure remains qualitatively stable across architectures despite differences in predictive accuracy and parameter count. Parameter-specific differences between SHAP and $\nu$SHAP further reveal how distinct regions of the implied
volatility surface contribute to parameter recovery and expose substantial redundancy in the calibration input. Building on this redundancy, we show that $\nu$SHAP explanations can guide a significant reduction in input dimensionality for the rough Heston model while matching calibration accuracy relative to the full implied
volatility surface. These findings demonstrate that complementary Shapley-based methods provide structural insight into learned inverse calibration mappings beyond predictive error metrics, and offer a practical route to feature selection in neural calibration problems.
Read the full abstract ▾
Genomics
Generating eukaryotic reference genome assemblies: Earth BioGenome Project quality standards and recommendations
Kerstin Howe, Dolores R. Agius, Tyler Alioto et al. (+45 more)
Reference genome assemblies are foundational resources across the biological sciences as they begin to expose the fundamental building blocks of each species and the molecular toolkits vital for adaptation and survival. High-quality genomes enable insights into
evolution, facilitate ecosystem monitoring and protection and conservation of species, access to new biomaterials and biomedicine, advances in agri- and aquaculture, and support planetary health. The amount of information gathered from a species' genome assembly is directly dependent on its quality. However, different sequencing technologies and software solutions generate a wide range of quality outcomes. Here, the Earth BioGenome Project's Sequencing and Assembly committee, together with a large community of researchers performing eukaryote sequencing and assembly, outlines quality standards for a "reference" assembly and formulates recommendations to ensure these standards are met.
Read the full abstract ▾
All authors Kerstin Howe, Dolores R. Agius, Tyler Alioto, Jean-Marc Aury, Roberto Biello, Mark Blaxter, Matthew W. Brown, Tom Brown, Bernat Burriel-Carranza, Haoyu Cheng, Dario Copetti, Solenne Correard, Fernando Cruz, Paul Davis, Aleksandra Denisova, Giulio Formenti, Pedro A. F. Galante, Caroline Howard, Kamil S. Jaron, Erich D. Jarvis, Sergei Kliver, Sergey Koren, Ilia J. Leitch, Harris Lewin, Heng Li, Kerstin Lindblad-Toh, Jack A. Medico, Matthieu Muffato, Zemin Ning, Javan Okendo, Shujun Ou, Marc Palmada-Flores, Brendan J. Pinto, Aurora Ruiz-Herrera, Abdoallah Sharaf, Ying Sims, Jeramiah J. Smith, Marco Sollitto, Cibele G. Sotero-Caio, Marcela Uliano-Silva, Sibelle T. Vilaça, Robert M. Waterhouse, Melissa A. Wilson, Chentao Yang, Guojie Zhang, Chenxi Zhou, Yang Zhou, Richard Durbin
Group Theory
Uniform Upper Bounds for Conjugacy Separability Growth in Nilpotent Groups
Jonas Deré and Lukas Vandeputte
A group is conjugacy separable if every pair of non-conjugate elements remains non-conjugate in some finite quotient. Virtually polycyclic groups, and hence in particular all finitely generated nilpotent groups, are conjugacy separable. The conjugacy separability growth function measures the complexity of distinguishing non-conjugate elements in finite quotients by giving the smallest order of a finite quotient $Q$ that separates them. Recent work showed that this function admits
polynomial upper and lower bounds for nilpotent groups, but these estimates are neither explicit nor optimal. We prove improved
polynomial upper bounds whose degree is at most linear in the nilpotency class and at most quadratic in the Hirsch length. For $2$-step nilpotent groups, we obtain sharper estimates and show that these estimates are optimal among bounds depending only on the nilpotency class and Hirsch length. This gives a correction of an error in the existing literature. The main tool is a translation of conjugacy separability growth into Lie rings, generalizing previous work in the case of residual finiteness.
Read the full abstract ▾
Cryptography and Security
SecJev: Bringing Security Expertise to System One Decision Models
Zheng Chen, Fei Yu, Haohao Huang et al. (+3 more)
Security workflows need models that turn complex observations and explicit policies into decisions. System One models introduced by Jev return typed predictions and probabilities; security specialization supplies the domain expertise behind those predictions. We introduce SecJev, to our knowledge the first family of Jev-style decision models specialized for security, spanning 0.8B to 9B parameters. Built on Kev's single-pass candidate scorer, SecJev learns Boolean, choice, and ordered decisions from text, telemetry, and observation histories. We develop SecJev-Corpus to unify source-label prediction and explicit-policy evaluation across 14 tasks and eight sources. It covers tool outputs, traffic,
federated updates, consensus, authentication, and vehicle messages. Scene-weighted training adapts the models across these domains while preserving a shared typed decision interface. Security specialization improves every model in the family; SecJev-0.8B outperforms general Kev-9B by 20.51 percentage points in task-macro accuracy. At 0.8B and 4B, SecJev achieves accuracy and latency close to answer-only generative fine-tuning with lower peak allocated GPU
memory. SecJev-4B improves on Kev-4B and security-trained generation in
prompt-injection and traffic transfer tests. We release adapters, decision heads, SecJev-Corpus, and training and inference code.
Read the full abstract ▾
All authors Zheng Chen, Fei Yu, Haohao Huang, Yang Li, Anlong Chen, Lei Chen
Nuclear Experiment
Spin-Density Matrix Elements in the Reaction $\gamma p \to \rho^-\Delta^{++}$ at GlueX
Niklas Herrmann and Farah Afzal for the GlueX Collaboration
The GlueX experiment, located in
Hall D at Jefferson Lab, aims to map the full spectrum of light-
quark mesons with an emphasis on hybrid
mesons, states in which the gluonic field is excited and contributes to the
quantum numbers of the
meson. The search for exotic hybrid
mesons requires a detailed understanding of the underlying production mechanisms. Polarization observables, such as the
spin-density
matrix elements (SDMEs), are an important experimental tool to investigate these mechanisms. The measurement of the unpolarized and polarized SDMEs in the photoproduction of $\rho^-\Delta^{++}$ using GlueX data is reported. In this analysis, for the first time, the correct full formalism is applied in which the upper mesonic and lower baryonic systems are described by a joint
spin-density
matrix. A factorized model yields results consistent with those of the full model, suggesting that this simplified description may also be suitable for other reactions involving
meson-
baryon final states.
Read the full abstract ▾
Machine Learning
Learn Feasibility Once, Optimize All Objectives: Derivative-Free Diffusion Models for Chance-Constrained Programming
Ziwen Liu, Yan Liu, Congying Han et al. (+3 more)
Chance-constrained programs (CCPs) optimize decisions under uncertainty by limiting the probability of constraint violation. Despite advances in traditional and learning-based approaches, optimizing non-convex or non-smooth objectives and adapting to different objectives under fixed chance constraints remain challenging. In this paper, we propose a Derivative-free
Diffusion-based framework that Disentangles constraint modeling from objective optimization, termed \textbf{D$^3$Opt}. We learn the chance-feasible structure once, independently of any particular objective, by training a risk-conditioned
diffusion model solely on constraint-filtered decisions and freezing it as a reusable prior for post-specified objectives. At inference time, we propose an annealed, particle-based Feynman–Kac correction along the frozen reverse
diffusion process to optimize post-specified objectives using only function evaluations. This enables derivative-free optimization of non-convex and non-smooth objectives without objective-specific retraining. We prove that the correction preserves feasibility when this property holds for the frozen prior, and derive an optimization-error bound separating learned-prior coverage, finite-particle approximation, and finite-temperature effects. Experiments on linear Gaussian CCPs, objective-transfer tasks, and chance-constrained economic dispatch demonstrate effective optimization across smooth and non-smooth objectives, including non-convex cases, and objective generalization under fixed chance constraints without retraining.
Read the full abstract ▾
All authors Ziwen Liu, Yan Liu, Congying Han, Tiande Guo, Yao Yan, Weichen Zhao
Number Theory
Telescopic Structure of Numerical Semigroups Generated by Unsigned Stirling Numbers of the First Kind
Takao Komatsu and Kyunghwan Song
Let $$ S_n=\left\langle \genfrac{[}{]}{0pt}{}{n}{1},\ldots,\genfrac{[}{]}{0pt}{}{n}{n-1}\right\rangle $$ be the numerical
semigroup generated by the nontrivial unsigned Stirling numbers of the first kind in the $n$th row. Writing $a=\genfrac{[}{]}{0pt}{}{n}{n-1}=\binom n2$ and $b_j=\genfrac{[}{]}{0pt}{}{n}{n-2j}$, we determine the complete gcd filtration of the reduced generating system. More precisely, if \[ M_j=\frac12\operatorname{lcm}\{m\ge1:\varphi(m)\le2j\}, \] then $$ \gcd(a,b_1,\ldots,b_j)=\frac{a}{\gcd(a,M_j)}. $$ The proof uses
prime-power block
polynomials and exact $p$-adic support at threshold degrees. We then establish a divisor-support property for canonical mixed-radix reductions and combine it with an elementary-symmetric growth estimate to prove that, after inactive generators are removed, the resulting generating sequence is telescopic. Consequently, the Apéry set of $S_n$ is rectangular, yielding an explicit Frobenius formula. The same structure shows that $S_n$ is free and symmetric and gives explicit formulas for its genus and conductor, while its type is one.
Read the full abstract ▾
Algebraic Topology
The zeroth stable homotopy groups of motivic spheres over the integers
Oliver Röndigs
The main result determines the zeroth integral Milnor-Witt stem of the motivic sphere spectrum in the Morel-Voevodsky motivic stable homotopy category of the integers. The component in weight zero is the Grothendieck-Witt ring of nondegenerate symmetric bilinear forms over the integers. Along the way, cellularity of connective Witt theory, as well as an absolute purity result for the {\eta}-inverted motivic sphere spectrum, is established over Dedekind domains of mixed characteristic.
Read the full abstract ▾
Computer Vision
Where to Look Is Not How to Fix: Pre-Denoising Diagnostics and Modality-Dependent Control in Diffusion Composition
Fangzheng Wu and Brian Summa
Understanding compositional failures in text-to-image
diffusion requires identifying both where stress is detectable and how intervention changes the output. We study these questions through a controlled anchor–stress protocol that jointly evaluates text-encoder diagnostics and denoiser interventions. We introduce a text-only Compositional Stress Index (CSI), which separates common from rare compositions across SD1.5, SDXL, and the SD3 text path and provides an upstream diagnostic coordinate. A matched six-
prompt localization study links intervention location to distinct outcomes: residual-minimizing
embedding adapters improve representation fit, while downstream cross-
attention intervention increases color hit rate (CHR) by 0.0272. Across SD1.5 and SDXL denoiser blocks, the largest positive signed diagnostic-accessibility mean occurs at the deep encoder, whereas selective boost has its largest positive mean CHR response at decoder blocks. Selective subtraction and broad ablation reveal further modality- and architecture-dependent responses, including a substantial CHR decrease when SDXL decoder cross-
attention is broadly ablated. We find a diagnosis-control dissociation under our controlled attribute-object composition setting: compositional defects are diagnosable before denoising, but the representation coordinate that exposes risk is not necessarily the coordinate or modality that improves generation.
Read the full abstract ▾