Artificial Intelligence
Papers
Fresh from arXiv, typeset properly. Highlighted words trade on the market.
All of arXiv
Thursday 10 September
Learning user-defined concepts as constraint networks has been extensively studied in the constraint acquisition (CA) literature. However, existing approaches typically rely on intensive interactions with a human oracle, making the learning process costly in terms of time and number of queries. In this paper, we propose a neuro-symbolic framework for automatic CA that significantly reduces user involvement by introducing neural Oracle Transformer models which learn to emulate user responses and to generalize conceptual knowledge. Trained on previously available examples, the learned oracle interacts with a dedicated CA engine, FastCA, which systematically refines the oracle's responses into a sound, consistent, and interpretable constraint network. This neuro-symbolic interaction enables the recovery of structured symbolic models from data without prior domain knowledge. Our results demonstrate that this neuro-symbolic interplay effectively aligns data-driven pattern recognition with symbolic reasoning, offering a robust approach to automating model construction in combinatorial domains.
Numerical Analysis
Error bounds for the Sherman-Morrison formula and its modification with improved stability
It is known that the Sherman–Morrison (SM) formula is not numerically stable. In recent work, we introduced SMIR, an algorithm that incorporates iterative refinement to enhance the SM backward error. In this paper we take a different route: adapting an algorithm of Govaerts, originally designed for general bordered linear systems, we develop a modified Sherman–Morrison (MSM) method whose built-in self-correction makes it surprisingly resilient. Whereas standard SM requires the solution of two $n\times n$ linear systems, MSM requires three; by contrast, SMIR requires $2+k$ solves, where $k$ is the number of IR steps and can be substantially larger than three, when IR converges slowly. We then derive backward and forward error bounds for both SM and MSM. The SM backward error bound established here is stronger than the one proved in [Hashemi & Nakatsukasa 2026]: it accounts for every rounding error and holds with no conditions on the size of the capacitance. From these bounds we extract growth factors that are cheap to compute a posteriori and can be used to certify the backward and forward stability of SM and MSM on a given problem. In our experiments, MSM consistently produces backward stable solutions, and is therefore observed to be forward stable as well. A formal proof of the stability — or instability — of MSM remains an open problem.
Read the full abstract
Artificial Intelligence
GTA: Graph Theory Agent and Benchmark for Algorithmic Graph Reasoning with LLMs
Large Language Models (LLMs) are increasingly asked to reason over structured data such as graphs, yet how reliably they can carry out multi-step graph algorithms in language remains unclear. Existing evaluations tend to use simple tasks on small graphs, to score code generation rather than reasoning over the graph itself, or to fix a single input format. We introduce Graph Theory Bench (GT Bench), a benchmark covering 24 classical graph problems in 44 task-structure settings, with over 100,000 examples across four representations: natural language, structured language, adjacency list, and adjacency matrix. Evaluating eight LLMs on GT Bench shows that accuracy is strongly tied to the input representation, that the best representation shifts with graph density, size, and topology as well as with the model, and that this sensitivity persists, attenuated, in the strongest reasoning models. Building on these observations, we propose the Graph Theory Agent (GTA), which pairs a preference-trained representation selector with plan-and-decompose scaffolding around a frozen executor LLM. GTA lifts Phi-4 from 53.5% to 69.1% on the benchmark's easy split and from 33.0% to 41.5% on its hard split, outperforming eight prompting and agent baselines, and transfers without retraining to GraCoRe and NLGraph. Code for benchmark generation and evaluation: https://github.com/xzx34/GTA. The project homepage is available at https://xzx34.github.io/gta/.
Read the full abstract
Machine Learning
Adaptive Chemotherapy Control under Tumor Heterogeneity via Reinforcement Learning
Designing effective chemotherapy regimens is hindered by tumor heterogeneity and drug resistance, which complicate the deployment of patient-specific model-based optimal control across diverse populations. We develop and compare closed-loop deep reinforcement learning (DRL) dosing policies with continuous (TD3) and discrete (DQN) action spaces trained on a high-dimensional heterogeneous tumor model. The DRL policies are benchmarked against a Pontryagin's Maximum Principle (PMP)-derived open-loop benchmark. We assess generalization under parametric heterogeneity using a 100-patient virtual cohort with plus or minus 10 percent uniform perturbations in growth and drug-sensitivity parameters. Across this cohort, TD3 achieves higher average tumor reduction, while DQN yields tighter inter-patient dosing consistency, revealing a clear efficacy-consistency trade-off in this study. Our simulations assume full observation of all tumor subpopulations; translation to sparse and noisy clinical measurements will require partial-observability formulations and/or state estimation. Overall, the results show that simulation-trained DRL can learn state-dependent feedback dosing policies that complement open-loop optimal control benchmarks.
Read the full abstract
Cryptography and Security
A First-Principles Evaluation of Graph-Based Network Intrusion Detection Systems
Graph-based network intrusion detection systems (GIDS) report strong benchmark detection metrics, but those metrics establish little about deployability. We approach the problem from first principles: rather than inheriting the preprocessing, windowing, and thresholding conventions of each published system, we ask what a controlled comparison requires and impose it uniformly. The result is GIDS-Eval, an evaluation framework that decomposes a GIDS into six interchangeable stages and turns those conventions into explicit experimental variables, so reported performance can be attributed to individual stages instead of whole pipelines. We survey nine representative GIDS, reimplement five of them within GIDS-Eval, and evaluate them on four datasets under one matched protocol. We identify nine recurring evaluation gaps and quantify the impact of each: two crafted edges achieve full evasion against three of the eight detector-dataset pairs with anything to hide; the snapshot window alone accounts for a mean 38.3% relative swing in average precision (AP); aligning preprocessing across systems moves AP by up to 61.8 percentage points for a single detector; and none of the 18 detector-dataset pairs we replay can alert as events arrive. We introduce GIDS-Lite, an encoder-free control built in the same framework, which ranks first by AP on two of the four datasets at up to 575$\times$ lower runtime. Architectural complexity is therefore not a consistent driver of detection quality under our matched protocol on current benchmarks, but it does enlarge the runtime, calibration, and attack surfaces operators must defend.
Read the full abstract
Quantum Physics
Sparse-Blossom Decoding in $o(1)$ Time
Matching-based decoding is widely used in quantum error correction, and accelerating it is key to enabling fast and scalable fault-tolerant quantum computation. Minimum-weight perfect matching (MWPM) decoding provides rigorous guarantees for error suppression, while sparse blossom enables its practical implementation at modest problem sizes. However, the runtime of existing sparse-blossom implementations unavoidably increases with problem size, motivating a rigorous parallelization framework that guarantees correctness and a runtime shorter than the syndrome-extraction timescale. Here, we present such a framework and prove that the resulting parallel sparse-blossom algorithm produces the same correction as the original, non-parallel sparse blossom. For the rotated surface code with code distance $d$ and physical error rates below a finite threshold, we prove that the average parallel runtime of decoding for $O(d)$ rounds of syndrome extraction is upper bounded by a quasi-polylogarithmic function of $d$. For a $d$-round decoding window, this implies that the average parallel runtime per round is $o(1)$. We also perform numerical simulation to identify conditions under which the parallel runtime per round decreases with increasing code distance. These results suggest that increasing code distance need not lead to longer parallel decoding times, providing a foundation for scalable parallel matching-based decoding.
Read the full abstract
Computer Vision
Revisiting Multi-Object Tracking Baselines: Hyperparameter Optimization with Multi-Fidelity Greedy Coordinate Search
Multi-object tracking (MOT) is dominated by the tracking-by-detection paradigm, whose methods typically rely on a small set of hyperparameters that are conventionally chosen by hand. Tuning them requires repeated expert-guided experimentation, while the procedures used to select reported values are often not systematically evaluated or fully documented. Hyperparameter optimization (HPO) automates this process, yet it remains rarely used in MOT, and existing studies applying HPO to MOT predate modern deep-detector-based trackers and HOTA evaluation. We systematically apply HPO across two datasets and four tracking-by-detection methods. We also propose Multi-Fidelity Greedy Coordinate Search (MFGCS), which optimizes one hyperparameter at a time by first evaluating candidate values on a small subset of scenes and re-evaluating only promising candidates on the full dataset. Across all eight tracker-dataset combinations, the Tree-structured Parzen Estimator (TPE) and MFGCS outperform both our hand-tuned configurations and the corresponding published results, with improvements of up to 4.38 and 16.05 HOTA points, respectively. MFGCS also reaches a predefined HOTA target faster than TPE in seven of the eight combinations. Within each tracker-dataset pair, all optimizers share the same search space and evaluation pipeline, isolating the effect of the search strategy. We release the code and tuned configurations to enable future work to compare against systematically optimized rather than default or manually tuned baselines.
Read the full abstract
Computation and Language
HypoKG: Evidence-Disciplined Biomedical Hypothesis Generation Beyond Endpoint Knowledge
Large language models (LLMs) can generate biomedical hypotheses, but it remains unclear whether they truly reason from scientific evidence or simply produce convincing-sounding ideas. To study this, we combine three major biological databases: the Kyoto Encyclopedia of Genes and Genomes (KEGG), Rhea, and UniProt, into a unified biochemical knowledge graph and construct a benchmark of 550 paths connecting enzyme sources to rare disease endpoints, yielding 13,200 hypotheses from six LLMs under four conditions varying the biological information each model receives: source enzyme only, full biological path, or source and disease endpoint only. Hypotheses are scored using an expert-derived five-criterion rubric on a 1-5 scale per criterion. We find that models given both the source and disease endpoint often produce the highest-scoring hypotheses, showing that LLMs can generate compelling ideas from minimal information. However, these hypotheses are less grounded in the evidence. In contrast, models given the full biological path generate hypotheses more consistent with known mechanistic relationships. We call this evidence-disciplined reasoning. To confirm this effect, we shuffled intermediate path steps while keeping endpoints fixed. Evidence grounding dropped significantly (delta = -0.793, p < 0.001), confirming models genuinely used path structure during reasoning. Our findings show that knowledge graphs support hypothesis generation in two ways: they identify biological endpoint pairs absent from the literature, and their mechanistic paths guide how LLMs reason between them.
Read the full abstract
Machine Learning
The Rank the Task Demands: A Causal Rank Law for Matrix Memories Trained on Group Composition
Matrix-valued memories make rank the natural budget of a learned representation: the number of independent directions a state spans bounds what it can bind, compose, and track. We report causal evidence, on a group-composition testbed trained under a hard single-state bottleneck with a fixed decoder that cannot launder rank, that gradient descent recruits precisely the rank the task's algebra demands. A companion paper [Larson, 2026a] establishes the analogous recruitment and causal necessity pattern on a $K$-pair associative-binding testbed, where exact recovery provably requires state rank at least $K$; this paper inherits that instrument and extends the rank law from a scalar capacity bound to a representation-theoretic one. We train toward chosen minimal faithful reference representations embedded in larger matrices. On group-composition state tracking over five finite groups spanning the solvable/non-solvable divide, the recruited rank equals the group's minimal faithful real representation dimension $d_{\min}$ (Spearman $\rho = 0.9747$, the design's tie-capped maximum), the dimension-matched solvable/non-solvable pair $S_4$/$A_5$ is statistically equivalent under a pre-registered test, and a pre-registered force-rank test separates a guaranteed similarity ceiling from empirical recovery at the target dimension: one rank below $d_{\min}$, cosine similarity is capped by the target's tied unit spectrum at $\sqrt{(d_{\min}{-}1)/d_{\min}} \le 0.894$, below the $0.9$ threshold in every group by construction, with observed cells at 86-95% (mean 91%) of that ceiling; at $d_{\min}$, not guaranteed a priori, recovery clears the pre-registered anchor-relative bar at four seeds per group in all five groups. Within this testbed, measured effective rank tracks representation dimension; the matched-dimension $S_4$/$A_5$ comparison establishes equivalence within the pre-registered tolerance.
Read the full abstract
Robotics
Pneumatic neurons for soft robots enable inflate-and-fire networks for rhythmic motion
Animals coordinate their movements through distributed neural circuits, but soft robots still typically depend on external, centralized electronics for control. Building soft robots that operate without centralized electronic controllers while remaining responsive to their environment remains a frontier challenge in soft robotics. In this work we introduce a soft-robot control architecture inspired by leaky integrate-and-fire models of biological neural circuits. The Pneumatic neuron (Pneu-ron) is a soft actuator that unifies energy conversion, logic, and actuation in one component. Each module combines a low-boiling-point fluid (LBF), a heater, and a mechanical switch into a self-excitable unit. Boiling the LBF inflates the module and triggers excitation and inhibition of adjacent modules in a process we call "inflate-and-fire". When interconnected into excitatory-inhibitory rings, Pneu-rons generate stable, sequential oscillations whose frequency emerges from the material dynamics and environmental conditions. By harnessing the inflation of Pneu-rons for actuation these networks can drive oscillatory locomotion of soft robots. Pneu-ron networks sustain oscillation under mechanical load and thermal variations, adapting through material physics rather than computation. Dynamical modeling of these networks reveals a dimensionless bifurcation diagram that dictates the network's oscillatory behavior. Encoding logic and actuation into material-level modules presents a new avenue for adaptive, electronics controller-free, soft robots.
Read the full abstract
Strongly Correlated Electrons
Geometric Approach to the High-Throughput Identification of Honeycomb Materials
The honeycomb lattice is an iconic structural motif in condensed matter physics. However, emerging theoretical models require the identification of new honeycomb lattice materials with strict specification on their structural, magnetic, and electronic characteristics. In this work, we introduce a filtering algorithm using simple geometric arguments to detect materials with honeycomb structural motifs. We classify four unique categories of lattices: conventional honeycombs, hyperhoneycombs, staircase or zigzag honeycombs, and fully 3D honeycombs. We also identify several distortion motifs in the conventional honeycomb set: elongated, sheared, and buckled honeycombs. We describe several screening criteria for further analysis of honeycomb systems, specifically the quantification of exfoliability in 2D materials, the identification of magnetic honeycombs, and a density-of-states peak quality metric for identification of electronic instabilities. The database containing the results of the filters and subsequent analysis is a useful tool for identifying novel materials for targeted physical phenomena for experimental characterization.
Read the full abstract
Combinatorics
Elliptic $b$-Hurwitz theory and Jack heat trace
We study a deformation of the central heat trace on $\mathrm U(N)$ obtained by deforming Schur polynomials to Jack polynomials, and prove that it admits an asymptotic expansion to arbitrary order. Its coefficients are governed by elliptic $b$-Hurwitz numbers, which are genus-one counterparts of the $b$-Hurwitz theory of Chapuy and Dołęga \cite{ChapuyDolega22}. We construct the associated elliptic $b$-Hurwitz theory by means of generalized coverings on a torus and identify it with the genus-one closure of the genus-zero simple $b$-Hurwitz theory. At $b=0$ the construction recovers ordinary elliptic Hurwitz theory, while at $b=1$ it gives an automorphism-weighted geometric interpretation of the connected and disconnected twisted elliptic Hurwitz numbers of Hahn–Markwig \cite{HahnMarkwig26}. Our results extend the topological expansion obtained in the classical case in \cite{LemMai25,LM2} and take the form of a coupling between chiral and antichiral elliptic $b$-Hurwitz generating functions.
Read the full abstract
Combinatorics
Nonvanishing higher Specht polynomials and a construction for three row and hook shape Garsia–Procesi modules
Given a polynomial ring quotient $R=\mathbb{C}[x_1,\ldots,x_n]/I$ with an action of the symmetric group induced by permuting the variables, a higher Specht basis is a collection of bases for each irreducible $\mathfrak{S}_n$-module in its decomposition that mimics the behavior of the classical Specht polynomial construction in the lowest degrees. Higher Specht bases have now been constructed for the coinvariant ring, the full polynomial ring, the rings $R_{n,k}$ appearing in the $t=0$ Delta conjecture, the hook shape Garsia-Haiman modules, and the two-row Garsia-Procesi modules, and more. We establish a general theory for determining when a higher Specht polynomial is nonzero, and give a proof of a conjectural higher Specht basis for all Garsia-Procesi modules in the cases of three row shapes and hook shapes.
Read the full abstract
Computation and Language
Automated Detection and Structuring of Social Tipping Point Evidence in Climate related Documents: A Modular AI Framework
The climate literature has grown faster than review teams can read it. That gap matters most for a concept like the environmental social tipping point, the threshold at which a small change triggers rapid, self-reinforcing change in a social system. Evidence of this kind of shift is usually contained in one or two paragraphs within a longer document. As a result, existing text mining tools-which categorize entire documents by topic or highlight isolated claims-leave an expanding set of important evidence without any systematic method for discovery or organization. This paper presents an open and modular transformer-based framework that detects and structures social tipping point evidence at the passage level. The framework joins five components into a single deployable workflow: a DistilBERT boundary splitter for segmentation, an iteratively augmented RoBERTa classifier for detection, a Mistral 7B model that rewrites each detected passage for clarity, a LLaMA 3.2 3B model that rates the passage against five published social tipping point criteria, and a Milvus vector store for semantic retrieval. The system is wrapped in a Streamlit interface backed by MinIO object storage. Evaluated on a 163-passage benchmark labelled by GPT-4.1 and a 51-passage set reviewed by experts, the splitter surpassed three competing methods on a nine-metric composite score (6.137). The tuned RoBERTa model achieved 71.4 percent accuracy with a Cohen's kappa of 0.337 on the full benchmark, and 87.5 percent accuracy with a kappa of 0.742 on passages with labels, outperforming both a climate-focused model and untuned language models.
Read the full abstract
Commutative Algebra
Numerical invariants for three-generated ideals
We study numerical invariants of homogeneous ideals generated by three forms, with particular emphasis on the interaction between the initial degree of the syzygy module, the multiplicity of the codimension-two part of the ideal, and the Bourbaki degree. We extend the Bourbaki degree from gradient ideals of plane curves to arbitrary three-generated homogeneous ideals and obtain sharp numerical bounds in both the general and the equigenerated settings. In the equigenerated case, these bounds extend the du Plessis–Wall picture and lead to the study of numerical, structural, dimensional, and integrable gaps. We show that numerical admissibility does not in general imply realizability and obtain further restrictions from the minimal graded free resolution. As an application, we determine the possible Bourbaki degrees of reduced singular quartic plane curves and describe them in terms of their singularity configurations; the possible values are 0,1,2,3,4,6,7,8.
Read the full abstract
Sound
DriftSE: Speech Enhancement with Generative Drifting
We propose DriftSE, a novel one-step generative framework for speech enhancement formulated as a latent distribution equilibrium problem. During training, the drifting field aligns the generator's pushforward distribution with the clean speech manifold through drifting in a latent domain. During inference, the drifting process is discarded, enabling one-step generation. We establish that its enhancement quality depends fundamentally on the choice of latent representation. Semantic latents preserve phonetic structure but fail to capture physical acoustic cues, whereas acoustic latents reconstruct the physical signal but risk linguistic hallucination. Therefore, we introduce dual-latent drifting, performing parallel drifting in both semantic and acoustic latents to simultaneously preserve phonetic intelligibility and acoustic fidelity. Additionally, we demonstrate that DriftSE enables fully unpaired training by aligning latent distributions rather than exact point-wise targets. Consequently, DriftSE facilitates cross-dataset learning in the absence of paired noisy-clean samples. Moreover, DriftSE exhibits broad architectural flexibility across different generator backbones. Extensive evaluations on additive denoising and convolutive dereverberation demonstrate robust one-step enhancement across both offline and real-time causal settings. Notably, DriftSE achieves state-of-the-art word error rates across all four evaluated datasets while strictly operating at 1 NFE. Code and audio examples are available online.
Read the full abstract
Quantum Physics
Keyless secrecy against bounded adversaries
We introduce a keyless coding/cryptographic primitive that asks for two guarantees at once: the receiver is never fooled into accepting a message other than the one sent, and the adversary learns nothing about the message unless the receiver aborts. Neither the sender nor the receiver holds a secret key, and no computational hardness is assumed. The only restriction is on the adversary's online computation: the receiver aborts if nothing arrives by a fixed deadline, so the adversary must forward something before it, and her map in that window is computed by a circuit of size (or depth) at most $p$; before and after, she is unbounded. For every polynomial $p$ we construct such an efficient quantum scheme, encoding $k$-bit messages into $n = O(k)$ qubits with circuits of size $\mathrm{poly}(n,p)$. No classical scheme achieves this type of everlasting security, at any choice of parameters. Our techniques also resolve open questions about universal tamper detection against families restricted in cardinality rather than in circuit size. We settle a question of Broadbent, Kapshikar and Rochette on relaxed tamper detection. As a corollary, we obtain the first efficient non-malleable code secure against global quantum tampering, with no split-state restriction, whereas the current constructions in the literature require the codeword to be split into non-communicating shares.
Read the full abstract
Systems and Control
Abstract homogeneous chains: a Lyapunov framework for high-order sliding modes in multi-agent systems
This work develops a Lyapunov framework for a broad class of arbitrary-order sliding-mode algorithms in multi-agent systems. We introduce abstract homogeneous chains, a class of nonlinear error systems characterized by common convexity and homogeneity properties. For this class, we establish global finite-time stability for arbitrary order, construct a homogeneous Lyapunov function, and derive a recursive optimization-based gain-proposal procedure. The framework addresses several gaps in existing dynamic average consensus and distributed differentiation results: it provides a recursive numerical optimization-based gain-proposal procedure for EDCHO at arbitrary order, extends REDCHO convergence from local to global, and provides arbitrary-order numerical gain-proposal rules for leader-follower distributed differentiation, previously available only at first order. It also provides a new arbitrary-order observer for multi-leader affine formation tracking with global finite-time convergence.
Read the full abstract
Earth and Planetary Astrophysics
OSIRIS-REx Returned a Pristine Sample of Asteroid Bennu: Takeaways from the Mission's Contamination Control and Knowledge Program
NASA's OSIRIS-REx mission had the objective of delivering a pristine sample from asteroid (101955) Bennu to Earth for scientific analysis–where "pristine" signifies the absence of foreign materials that could affect sample measurements. OSIRIS-REx returned 121.6 g of regolith in September 2023; this study documents the systematic investigation of suspected contaminants encountered during the mission's sample analysis phase. Most suspected contaminants did not originate from the spacecraft or sample curation. Some were introduced during laboratory analyses, reinforcing the importance of procedural blanks and a strategic approach to sharing samples across laboratories with different analytical targets. Several suspected contaminants, such as phosphate and sodium fluoride particles, were ultimately identified as indigenous to Bennu, highlighting the critical role of contamination knowledge in preventing the dismissal of valuable scientific data. We find that the returned sample meets the definition of pristine, except for an isolated 1.24% (by mass) that escaped the sample container and was thereby contaminated with spacecraft particulates. These findings demonstrate the effectiveness of systematically applied contamination science and engineering practices and provide lessons and approaches to help maximize the scientific integrity of future planetary sample return missions.
Read the full abstract
Robotics
Trajectory Bundle Method in SE(3) for Black-Box Fixed-Wing Aircraft Trajectory Optimization
Dynamically feasible trajectory optimization for rigid-body systems is naturally formulated on the special Euclidean group SE(3) but is challenging when dynamics are available only as black-box computations without derivatives. This paper formulates the Trajectory Bundle Method (TBM) for motion planning implicitly on SE(3). Bundles are constructed in the Lie algebra and propagated through nonlinear rigid-body dynamics using exponential and logarithmic maps, enabling derivative-free planning of non-Euclidean trajectories. We show that Euclidean TBM interpolation error is bounded quadratically by bundle diameter and extend this result to SE(3), where the bound additionally depends on a local Lipschitz constant of the Log map. Numerical experiments corroborate these bounds. Finally, we demonstrate SE(3) TBM by optimizing an acrobatic, collision-free fixed-wing maneuver through a rotated aperture without explicit models or derivatives of the vehicle dynamics, aerodynamics, or collision model.
Read the full abstract
Artificial Intelligence
Soft Symbol Grounding for Prototypical Concepts
Neuro-symbolic models are usually trained with supervision only on final labels, leaving the intermediate concepts unobserved. Since many concept assignments are consistent with a given label, training can predict labels correctly while recovering the wrong concepts, a failure known as a reasoning shortcut. Prototypical networks reduce shortcuts by anchoring each concept to a few labeled examples, but existing methods still couple perception and reasoning through a hand-crafted, task-specific differentiable loss that must be redesigned for every task. We introduce Soft-PNet, which removes this loss: it reframes concept grounding as a Metropolis walk over a precomputed cache of feasible symbolic solutions, guided by a prototype distribution built from a single labeled anchor per concept, and trains against one KL objective between the prototype-weighted cache and the network's concept predictions. The objective is identical across tasks and remains applicable when the solution space cannot be enumerated. On MNIST-EvenOdd, Visual Sudoku, and Kand-Logic under scarce supervision, Soft-PNet matches loss-engineered prototypical networks at the concept and label levels and recovers concepts that soft-grounding baselines miss, with no loss engineering and lower training time.
Read the full abstract
Chaotic Dynamics
Variability in a Model of Divergent and Convergent Coupled Ice Streams: Synchronisation, Transients and Chaos
Fast-flowing glaciers, known as ice streams, control the majority of ice sheet mass loss due to thermomechanical coupling at their base, which can induce basal lubrication and highly variable flow. Even quite simple models can show `binge-purge' oscillations. Recent work has highlighted that coupling between divergent ice streams can also give rise to chaotic variability. In this paper, we return to a model of ice streams with a divergent coupling to identify and correct some physical inconsistencies, most notably a global volume conservation error. We propose and validate a revised Divergent Coupling (DC) model that ensures rigorous mass balance. Bifurcation analysis and Poincaré sections reveal previously unidentified complex frequency-locking (Arnold tongues), a period-adding cascade and chaos associated with torus breakdown. We also consider an equivalent model with a Convergent Coupling (CC), where we find no evidence of chaotic variability. These findings suggest the importance of specific topologies on ice stream variability.
Read the full abstract
Robotics
DIA: Denoising Intermediate Advantage for Diffusion Policy Optimization
Diffusion-based robot policies have become widely used in robotic manipulation, where they are typically trained with behavior cloning. However, policies trained purely from demonstrations are limited by the quality and coverage of the available data. Reinforcement learning can further improve the performance of these pretrained policies through interaction. A common approach is to use policy-gradient methods that formulate diffusion-policy fine-tuning as an outer environment MDP together with an inner denoising MDP. However, existing methods typically assign the same environment-level credit to all denoising steps used to construct an action chunk, without distinguishing which intermediate decisions contributed most to the final return. We introduce Denoising Intermediate Advantage (DIA), a policy-gradient method that learns a value function over partially denoised actions and uses it to construct a denoising level advantage for each step of the generative process. DIA combines this inner credit signal with the standard environment-level PPO advantage, providing state-dependent credit throughout the denoising chain. Across Robomimic, FurnitureBench, Franka Kitchen, and D3IL, DIA consistently improves final performance over existing diffusion-policy fine-tuning methods. Beyond final reward, DIA reaches successful states more efficiently and can shift farther from the pretrained behavior distribution, enabling it to discover more effective and efficient task-level strategies and subtask sequences that baseline methods fail to reach.
Read the full abstract
Machine Learning
CRFCAN: A Complex-Valued Cross-Domain Residual Network for Joint Channel and Phase Noise Estimation in Sub-THz OFDM Systems
In sub-terahertz (sub-THz) communications, the coupling of ultra-wide bandwidth and severe phase noise (PN) impairments renders conventional joint channel and PN estimation highly complex and computationally prohibitive. To address this, we propose CRFCAN, a complex-valued residual FFT convolutional attention network designed for joint channel and PN estimation. Unlike existing deep learning schemes that rely on cascaded networks or hybrid frameworks combining neural networks with conventional iterative estimators, CRFCAN performs joint recovery in a truly end-to-end fashion through a physics-inspired cross-domain structure. Specifically, Fast Fourier Transform (FFT) and inverse FFT modules are embedded within residual groups to enable iterative feature interaction across the time and frequency domains, thereby capturing both frequency-selective fading and time-varying phase distortions. In addition, two dedicated residual blocks are introduced for complex feature extraction and multiplicative phase-distortion modeling, respectively. A physics-aware PN output tail with soft normalization is further employed to improve estimation stability while preserving the physical characteristics of the effective PN process. Simulation results demonstrate that CRFCAN significantly outperforms conventional algorithms and state-of-the-art deep learning models in terms of normalized mean square error (NMSE) and bit error rate (BER). Notably, CRFCAN achieves superior performance with single-shot, fixed-complexity inference and generalizes well to unseen PN models without fine-tuning, highlighting its robustness and practicality for sub-THz receivers.
Read the full abstract
Computation and Language
Chopthin-Consensus Power Sampling: A Diversity-Preserving Approach to LLM Decoding
Inference-time power sampling via Sequential Monte Carlo (SMC) can substantially improve large language model (LLM) reasoning without requiring post-training. However, many existing SMC approaches rely on equal-weight resampling, which can aggressively prune low-weight trajectories, discarding potentially correct reasoning paths and degrading the genealogical diversity of the search space. To address this, we introduce Chopthin-Consensus Power Sampling (CCPS). Our method applies the Chopthin resampler to LLM decoding: rather than equalizing weights and forcing unnecessary particle duplication, it enforces an upper bound on the ratio between the largest and smallest weights and carries the unequal weights forward. This targeted intervention preserves a richer set of distinct reasoning paths, keeps the weighted SMC approximation unchanged in conditional expectation, and guarantees a lower bound on the post-resampling effective sample size (ESS). To fully exploit this enriched population, we employ a semantic-majority selection mechanism that merges token-identical final trajectories, clusters semantically equivalent answers, and returns the answer supported by the largest number of distinct trajectories. Evaluating across three open-weight models and five reasoning benchmarks, we show that Chopthin increases oracle coverage in 13 of 15 settings. Combined with semantic-majority selection, CCPS matches or exceeds the final-answer accuracy of the Power-SMC baseline in 14 of 15 settings, delivering absolute gains of up to 10.6 percentage points. These findings demonstrate that diversity-preserving resampling and diversity-aware selection are complementary mechanisms for training-free LLM reasoning. Code is available at github.com/MinooAhmadii/chopthin-consensus-power-sampling.
Read the full abstract
Cryptography and Security
Evaluating Practical Enumeration and Blocking Attacks on the Snowflake Circumvention System
Proxy-based Internet censorship circumvention tools like Snowflake rely on large, dynamic pools of third-party proxies to resist IP-based blocking. We focus on two assumptions underpinning the security of Snowflake: that adversaries cannot easily enumerate proxy IPs, and that blocking those proxies would incur unacceptable collateral damage. In this paper, we test these assumptions by studying practical enumeration and blocking attacks against Snowflake conducted by malicious clients. We combine bounded, ethical real-world measurements with large-scale simulation to evaluate both present-day enumeration and blocking risk and broader attacker capabilities. Over 48 days of real-world measurements from May–June 2025, our attack enumerated over 21,000 unique proxy IP addresses belonging to almost 1,000 autonomous systems. Despite this high number, we find that proxy churn limits the overall effectiveness of enumeration over time, and reduces the impact on clients of individual proxy addresses being blocked. However, at the network level, blocking the top 1% of observed autonomous systems blocks more than 30% of observed Snowflakes while affecting 0% of Tranco Top 100 domains and ~2.5% of Top 1M domains. We discover that the broker's load-aware matching reveals stable, high-capacity proxies to attackers early, especially during periods of elevated demand such as the censorship even in Iran of June 2025, subsequently exposing the networks that contribute disproportionately to system connectivity. In simulation, increasing attacker scale sharply improves both enumeration and blocking success, while higher proxy churn significantly reduces blocking effectiveness. We conclude by discussing and evaluating practical mitigations, some of which have been integrated into Snowflake.
Read the full abstract
Numerical Analysis
Transform before linearizing: robust Newton methods for singular $p$-Laplace and $p$-Stokes equations
For $1<p<2$, the $p$-Laplace equation and its $p$-Stokes generalization are difficult to solve numerically. Newton's method converges rapidly only close to the solution, with iteration counts that grow under mesh refinement and deteriorate as $p\to1$. The more robust Picard iteration converges only linearly. Rather than globalizing or preconditioning Newton's method, we modify the system to which it is applied. We lift the equation by introducing the flux $|\nabla u|^{p-2}\nabla u$ as an auxiliary (or ”lifting”) variable, apply a nonlinear transformation to the resulting constitutive relation, linearize, and eliminate the auxiliary variable by static condensation. Lifting alone leaves the linearization unchanged; it is the preceding transformation that yields the new method. The elimination is algebraic and pointwise at the quadrature points, so the flux variable is never discretized, no inf-sup condition or indefinite system arises, and the cost per iteration is that of a standard Newton step. Together with a pointwise feasibility bound on the lifting variable that keeps the diffusion tensor uniformly positive definite, this yields an iteration that we prove, in finite dimensions and for the $p$-Laplace equation, to converge globally and locally at a quadratic rate; a one-dimensional model problem explains why the lagged flux variable removes the zig-zag behavior of standard Newton for $p$ close to one. Firedrake-based experiments for $p$-Laplace problems in two and three dimensions and for stationary and time-dependent $p$-Stokes flows show iteration counts largely insensitive to $p$ and to mesh refinement, and up to an order of magnitude fewer iterations than standard Newton for $p$ close to one.
Read the full abstract
Earth and Planetary Astrophysics
Steepening Optical Slopes in Exoplanet Transmission Spectra with Charged Hazes
Observations of exoplanetary atmospheres have revealed the ubiquity of aerosols, yet their microphysics remain largely uncertain. Haze charging is a key microphysical process in aerosol-rich Solar System atmospheres, but it has been ignored in exoplanetary haze models. Here we develop a microphysical framework that includes haze charging for exoplanet atmospheres and investigate its effects on transmission spectra. Electrostatic repulsion suppresses coagulation, maintaining smaller particle radii and higher mixing ratios in the upper atmospheres. We find that particle charging steepens the optical spectral slopes when vertical mixing is weak. This behavior arises from the combination of reduced particle sizes and a steepened vertical gradient in haze opacity. Furthermore, we find distinct spectral responses to charging for different haze analogs, owing to the different contributions of scattering and absorption. For absorption-dominated materials such as soot, particle-size reduction has little impact on opacity profiles, leading to only minor charging-induced spectral changes. By contrast, scattering-dominated materials such as tholin and diamond show a stronger spectral signature of particle charging because of the strong dependence of scattering opacity on particle size. While previous studies have invoked haze formation under vigorous vertical mixing as a mechanism for producing super-Rayleigh optical slopes, we demonstrate that highly charged scattering haze can produce such slopes even under weak-to-moderate vertical mixing. Our results highlight haze charging as a previously neglected microphysical process that can affect particle growth and leave observable signatures in transmission spectra, providing a quantitative framework for assessing atmospheric electrical conditions.
Read the full abstract
Distributed Computing
Specifying Paxos for System Builders: Pseudocode Made Executable
This paper presents a precise executable specification—as a faithful mapping from the pseudocode—of Paxos for System Builders, a practical protocol for replication and consensus in distributed systems. Paxos for System Builders has both a robust implementation in C and a clean pseudocode for critical protocol details. This paper shows how the protocol pseudocode can be expressed easily, essentially line-by-line, in a precise high-level language, DistAlgo, for direct execution in distributed systems. Precise specification and direct execution help significantly in understanding the protocol logic and in automatically checking, tracing, and visualizing protocol runs. They also led to discoveries and fixes of small, difficult-to-catch omissions and liveness bugs in the pseudocode though not the C code. The resulting program also has acceptable performance while having similar size as the pseudocode.
Read the full abstract
Probability
Predictable subordination, sharp martingale inequalities and applications
We introduce a new method for obtaining sharp $L^p$ estimates for martingales under predictable analogues of differential subordination. By allowing suitable comparisons between continuous and jump variation adapted to the relevant range of $p$, we retain sharp constants in settings where pathwise differential subordination fails. The key idea is to study together the continuous and jump contributions arising from the appropriate Bellman functions. Although these contributions need not be nonpositive separately, we quantify their defects and show that they compensate at the predictable level. This compensation mechanism is inspired by the author's earlier work. As an application, we substantially improve the explicit dimension-free bounds for the Riesz vectors on the Hamming cube and on $\mathbb{Z}^n$.