Machine Learning
Papers
Fresh from arXiv, typeset properly. Highlighted words trade on the market.
All of arXiv
Thursday 27 August
The recently introduced block-sparse featurizer (BSF; Fel et al., 2026) is similar to a sparse autoencoder (SAE), but its atomic unit is a small subspace (a block of directions) rather than a single direction. It is designed for features that live on low-dimensional manifolds, which are especially frequent in vision. This work studies the BSF's strengths and weaknesses, finding how it still somewhat suffers from classic SAE failure modes, like feature splitting and composition. We propose several architectural changes to the BSF, including a Tournament Top-K selection rule that significantly reduces feature splitting, and we also extend the block paradigm to the crosscoder.
Computation and Language
Trajectory-Level Speculative Decoding for Diffusion Language Models
Diffusion-based language models (dLLMs) enable parallel token generation through iterative denoising, but existing decoding strategies collapse to single-token generation under low confidence, severely limiting throughput. Unlike autoregressive models where speculative decoding operates on token sequences in a fixed left-to-right order, dLLMs require speculating over denoising trajectories-sequences of multi-token updates with explicit positions and unmasking orders. We develop a trajectory-level speculative framework that constructs draft denoising trajectories via confidence-stratified tree exploration and verifies them through blockwise parallel evaluation with bidirectional attention masking. Our method further introduces inter-block speculation, exploiting diffusion models' bidirectional structure to perform cross-block lookahead. We formally characterize when this approach is exact and identify trajectory drift as the fundamental cost of increased parallelism. Building on Fast-dLLM's dual-cache infrastructure, our framework reduces denoising iterations by 30-40% and increases tokens-per-step from 2.6 to 4.3, achieving 7-14x speedup over vanilla dLLMs and 1.3x over Fast-dLLM with less than 1% accuracy change across reasoning and code benchmarks.
Read the full abstract
Machine Learning
DAMP: Decay-Aware Mixed-Precision Recurrent-State Quantization
Complex reasoning and agentic applications increasingly rely on long-context inference, where growing KV caches increase both memory usage and decoding overhead. Hybrid models reduce these costs by combining Softmax Attention with Gated DeltaNet (GDN) or Kimi Delta Attention (KDA), which maintain fixed-size recurrent states. These states are commonly stored in FP32 and consume substantial GPU memory, while their updates are limited by memory bandwidth. Quantization can reduce both storage footprint and memory traffic, but we find that uniform INT8 and FP8 degrade complex reasoning accuracy, while INT4 and NVFP4 collapse it to near zero. To our knowledge, this is the first study of post-training recurrent-state quantization for GDN and KDA. Our analysis reveals that outliers in GDN and KDA states are concentrated in particular key channels and value dimensions. Learned decay influences how much quantization error is retained. We find that largely the same GDN heads and KDA key channels exhibit slow decay across tasks. Based on these insights, we propose DAMP, which jointly considers quantization error and decay-based error retention to select high-risk key channels offline. Under a fixed storage budget, it retains these channels in FP16 and stores the remainder in INT8. We evaluate DAMP on Qwen3.6-35B, Kimi-Linear-48B and Kimi-K3 across six reasoning and code generation benchmarks. At 9.9 bits per state value, DAMP maintains average accuracy close to FP32. In SGLang, DAMP reduces recurrent-state storage by 69.1%, accelerates the recurrent-state update kernel by up to 2.59x , and lowers full-model time per output token by up to 19.0%.
Read the full abstract
Machine Learning
Quantization-Triggered Backdoors in Language Models: Cross-Quantizer Transferability and the Validation–Deployment Gap
Post-training quantization is often treated as a semantically neutral optimization for edge deployment of Large Language Models. When a full-precision source checkpoint is evaluated and quantization is applied downstream without equivalent re-evaluation, this workflow creates a structural validation–deployment gap: because quantization is a many-to-one mapping over parameter space, source-precision certification does not guarantee behavioral equivalence in the deployed configuration. We formalize this gap through Quantization Behavioral Equivalence Classes (QBECs) and prove that QBEC membership does not imply behavioral equivalence, providing a theoretical basis for quantization-triggered backdoor attacks. Building on a three-stage adversarial fine-tuning framework, we embed latent malicious payloads into models that satisfy the source-precision checks used in our evaluation, yet activate targeted adversarial behavior upon INT8 or 4-bit compression. We evaluate this threat in two operationally motivated scenarios, tactical machine translation and political content analysis, extending prior work from decoder-only causal LMs to multilingual encoder-decoder sequence-to-sequence models. Results show that backdoored translation models move from zero measured friend–foe corruption at repaired FP16 to up to 85.02% inversion after quantization, and that a paired stance classifier measures an ideological shift of up to $\Delta\mathrm{Bias}=0.33$ upon compression. A cross-quantizer transferability analysis further shows that attack persistence varies across quantization schemes and model architectures, rather than being determined by nominal bit-width alone. These findings demonstrate that source-precision auditing alone does not rule out quantization-triggered behavior and that the final deployed configuration must be included in behavioral certification for trustworthy edge AI.
Read the full abstract
Cryptography and Security
eBPF-Based Cybersecurity Mechanisms: A Systematic Literature Review
Extended Berkeley Packet Filter (eBPF) has emerged as a kernel-level framework enabling dynamic security enforcement in modern operating systems. While eBPF's cybersecurity potential has attracted significant attention, existing work remains fragmented across domains, evaluation methodologies, and deployment contexts. This systematic literature review applies PRISMA methodology to identify, categorize, and synthesize peer-reviewed research on eBPF-based cybersecurity mechanisms. Following a structured screening of 3,735 records from six databases, 54 primary studies (2018-2026) were analyzed and organized into a seven-domain taxonomy: DDoS mitigation, intrusion detection, IoT security, container security, microservice protection, networking, and security tools. Analysis reveals eBPF enables low-overhead security enforcement (median 2.4% CPU overhead [1.1-8.6%], ranging from nanosecond-scale costs for infrequent hooks to 10-20% for kernel hot paths) with high detection accuracy (94-99%). It particularly excels in kernel-level monitoring, real-time packet processing, and cloud-native workload protection. However, significant challenges persist: verifier-imposed constraints limit algorithm complexity, 85.1% (46/54) of studies require low-level programming expertise, kernel version fragmentation hinders portability, and 96.2% (52/54) of research fails to address eBPF's own vulnerabilities. This review identifies critical research gaps in multi-tenant isolation, adversarial machine learning (ML) robustness, production validation, and standardized evaluation frameworks. By consolidating fragmented knowledge and highlighting architectural trade-offs between safety and expressiveness, this work provides a foundation for next-generation eBPF security systems and actionable directions for kernel programmability research.
Read the full abstract
Computation and Language
How Do Linear Probes Emerge? A Circuit-Tracing Framework with Concept-Targeted Attribution
Transcoder attribution graphs are usually trained to explain why a model assigns high probability to a particular next token. We introduce Concept-Targeted Attribution (CTA), which instead trains attribution graphs with respect to a linear probe direction. CTA therefore yields probe-specific circuits that explain why an internal concept representation arises in a prompt, independently of whether it is expressed in the generated token. Using Cross-Layer Transcoders, we show that these probe-targeted graphs contain predictive structure: graph-level features predict probe accuracy across four widely studied concept categories ($\rho = 0.91$, $R^2 = 0.84$), while local features identify the sparse components driving per-prompt classification. This connects probe performance to interpretable circuit structure, allowing us to ask not only whether a probe works, but which internal computations make it work. Causal ablations further show that probe-targeted and logit-targeted graphs capture functionally distinct mechanisms. Removing probe-relevant features reduces internal concept scores while largely preserving generated tokens, whereas removing logit-relevant features changes the generated token in 92% to 100% of cases with near-zero effect on probe scores. CTA provides a framework for moving from behavioral probe accuracy to mechanistic explanations of probe performance, enabling more detailed audits of internal concept representations, including safety-critical ones. Our code is available at https://github.com/vedantpalit/concept-targeted-attribution
Read the full abstract
Physics and Society
The Mechanics of Democratic Dominance: A System Dynamics Paradigm for Dynamic Consent Engineering
Traditional frameworks of political communication operate under linear, event-driven assumptions that treat voter persuasion as a static, transactional function. This paper introduces Dynamic Consent Engineering (DCE), a novel interdisciplinary paradigm that synthesizes Edward Bernays' foundational principles of public relations with the S-E-E-D (Snowball, Equilibrium, Elasticity, Dominance) framework of system dynamics. By expanding Bernaysian operational constraints into a four-dimensional resource matrix (incorporating algorithmic media infrastructure alongside manpower, mindpower, and capital), we mathematically formalize how democratic institutions construct, optimize, and sustain political dominance. We map Bernays' classic eight-step engineering workflow directly onto non-linear feedback loops, information time lags, and demographic carrying capacities. Through rigorous modeling of systemic feedback structures, we isolate the dynamic root causes of political elasticity, policy resistance plateaus, and threshold-triggered trust collapses. Finally, we establish a four-step diagnostic pipeline and validate the framework via historical reference mode verification, demonstrating that long-term political stability depends not on ephemeral rhetoric, but on the structural synchronization of policy execution and dynamic feedback calibration.
Read the full abstract
Artificial Intelligence
WM-R1: Training GUI Agents to Reason and leverage World Models with Reinforcement Learning
GUI agents trained with reinforcement learning (RL) have showcased strong environment learning capabilities on mobile platforms. However, RL typically demands extensive real-environment interactions, leading to high resource costs and instability, especially in GUI scenarios. To address these, we propose WM-R1, the first reinforcement learning framework that trains mobile GUI agents with world models instead of real environments. Specifically, world models serve as the source of state transitions during all rollouts, replacing the real Android environment within the training loop. WM-R1 also embeds world models directly into the thinking process, enabling agents to reason about the consequences of candidate actions before committing to the final action. Crucially, WM-R1 eliminates the need for real-environment interaction, supports massively parallelized and step-level granularized trajectory generation grounded in world models, and introduces a multi-dimensional rule-based reward that jointly optimizes task success, trajectory efficiency, and world model utilization. For efficient training, we curate a high-quality dataset of 2000 challenging tasks. Experiments on Android mobile benchmarks demonstrate that WM-R1-trained agents significantly outperform GRPO-only baselines and inference-time simulation methods. Code is available at https://github.com/genalyu/WM-R1 .
Read the full abstract
Machine Learning
Learning Complementary Routes for Scouting Persimmon Orchards during Shoot Growth with CoverShare
At the study orchard, scouting during persimmon shoot growth focuses on leaf color change, leaf area, and internode length. Several routes may share an access lane, yet each can offer a view of row segments that the others miss. We introduce CoverShare to learn complementary routes under a fixed action budget. Within Policy Gradient for Parallel State Entropy maximization (PGPSE), CoverShare redistributes auxiliary rewards according to the state coverage lost when each policy is omitted and the number of policies visiting each state. It preserves the pooled entropy objective and total auxiliary reward. We simulated the annotated road network of a 6.7-ha site with 107 observable persimmon scouting units across five zones. At 40 actions per route, six distinct policies covered an expected 17.84 units perinspection, compared with 8.28 for six independent runs of one policy, averaged across eight training seeds. With the same policies trained on 40-action routes, extending each inspection route to 80 and 120 actions raised coverage to 22.52 and 28.38 units. Across seven public discrete-state tasks, CoverShare increased state support by a mean of 16.29% over the team-entropy baseline. Controlled ablations identified leave-one-policy-out coverage as the main contributor to these gains.
Read the full abstract
Artificial Intelligence
Thinking Costs Tokens: When More Structure is Worth the Price
Adding inference structure to a language model lets it search, verify, and revise, but these actions consume the very budget they are supposed to use well. In this paper, we investigate whether there exists a token-budget threshold, below which the overhead of planning and verification hurts performance and above which it helps. We evaluate two systems on FinQA and TAT-QA financial reasoning tasks, using GPT-5.4 mini across 14 budget tiers ranging from 250 to 42,000 output-equivalent tokens. The first system is a monolith, which is a single LLM call. The second is a verified search architecture that adds planning, label-blind checking, and repair capabilities. We run 1,000 cases for a total of 28,000 completed cells. Both systems score 0% at the two lowest tiers, where neither can fit a complete prompt. At 1,000 tokens, the monolith reaches 18% accuracy while verified search scores near 0%, since the planning overhead leaves no room for an answer. From 1,500 tokens onward, verified search surpasses the monolith and maintains a consistent advantage, reaching approximately 44% at the highest tiers while the monolith reaches approximately 40%. The crossover occurs between 1,000 and 1,500 output-equivalent tokens, confirmed by a strict intersection-union test ($p \le 0.001$ at both endpoints).
Read the full abstract
Computation and Language
A Survey on Rubric-Guided Reinforcement Learning for Language Models
Reinforcement learning from human feedback (RLHF) has become the dominant paradigm for aligning large language models (LLMs) with human preferences. However, traditional RLHF relies on scalar reward signals that lack interpretability and fail to capture the multifaceted nature of response quality. Rubric-guided reinforcement learning addresses these limitations by introducing structured, interpretable evaluation criteria, or rubrics, as the backbone of reward design, feedback generation, and policy optimization. In this survey, we introduce a Bayesian framework that defines constitutions as prior distributions $P(R)$ over evaluation criteria and rubrics as conditional instantiations $R_x \sim P(R|x)$. Under this unified view, we present a taxonomy of rubric-guided RL along the prior-posterior axis, covering constitutional AI, instance-specific rubrics, process-level supervision, self-evolving rubrics, and their agentic and multimodal extensions. Furthermore, as rubrics are natural-language artifacts, we present a linguistic analysis of how granularity trade-offs, semantic drift, and linguistic reward hacking impact alignment reliability, identifying key open problems for future research.
Read the full abstract
Cryptography and Security
Circuit Discovery Helps Detect LLM Jailbreaking: A Mechanistic Interpretability Study
Despite extensive safety alignment, large language models (LLMs) remain vulnerable to jailbreak attacks that bypass safeguards to elicit harmful content. While prior work attributes this vulnerability to safety training limitations, the internal mechanisms by which LLMs process adversarial prompts remain poorly understood. We present a mechanistic analysis of the jailbreaking behavior in a large-scale, safety-aligned LLM, focusing on LLaMA-2-7B-chat-hf. Leveraging edge attribution patching and subnetwork probing, we systematically identify computational circuits responsible for generating affirmative responses to jailbreak prompts. Ablating these circuits during the first token prediction can reduce attack success rates by up to 80%, demonstrating its critical role in safety bypass. Our analysis uncovers key attention heads and MLP pathways that mediate adversarial prompt exploitation, revealing how important tokens propagate through these components to override safety constraints. These findings advance the understanding of adversarial vulnerabilities in aligned LLMs and pave the way for targeted, interpretable defense mechanisms based on mechanistic interpretability.
Read the full abstract
Quantum Physics
Quantized Low-Rank Quantum State Tomography: Hyperbolic Quantization and Riemannian Least-Squares Recovery
We study low-rank quantum state tomography from finite-bit Pauli batch responses. To avoid bias introduced by generic quantization, we propose HyperQuant, a mean-preserving hyperbolic quantizer adapted to the second-moment scale of Pauli responses. We establish minimax distortion guarantees and show that exact mean preservation enables direct rank-constrained least-squares recovery without altering the population target. We derive nonasymptotic recovery guarantees and an explicit bit–shot tradeoff under which finite-bit responses retain the error order of unquantized batch averages using fewer response bits. For efficient computation, we develop QuantRGD, a Riemannian gradient method with provable linear convergence to the corresponding statistical neighborhood under explicit resource conditions. Numerical experiments validate the predicted quantization, recovery, and convergence behavior.
Read the full abstract
Software Engineering
Image Augmentation as Test Generation for Deep Learning-Based Image Retrieval Systems
Ensuring the reliability of deep learning-based image retrieval systems is a software engineering challenge. This paper presents a dual contribution: (1) a literature review of augmentation and generation techniques which resulted in the identification of 50 techniques which we organized into a ten-category taxonomy, and (2) a large-scale empirical study that evaluates these techniques as test generators for embedding-based image retrieval systems. Augmented images are embedded using Amazon Titan and OpenCLIP, and evaluated across four analytical dimensions: (1) embedding-space similarity, (2) embedding uncertainty measured via four estimators, (3) semantic realism scored by LLaVA, and (4) retrieval failure rate. Experiments are performed on three datasets: CIFAR-10, ImageNet-1K, and a dataset from an industrial partner (March Networks). Across all evaluated datasets and embedding models, and under the single severity level tested for each technique, weather simulation and SaSPA are the image augmentation/generation techniques that produce the highest embedding uncertainty and failure rates while maintaining a favorable balance between performance stability, visual realism, and augmentation effectiveness. The results we discuss are configuration-specific and may shift under milder or stronger perturbation settings. In contrast, GAN-based augmentation techniques are among the lowest in realism, indicating the presence of synthetic artifacts and perceptual inconsistencies that reduce their suitability to produce realistic test inputs. Overall, our findings provide practical guidelines for selecting augmentation techniques that maximize test diversity while preserving realistic image characteristics, thereby enabling the construction of comprehensive and effective test suites for image retrieval systems while reducing the cost of manual data labeling through the use of metamorphic testing.
Read the full abstract
Computation and Language
INSPIRE: An Internalize-Then-Improve Approach for Example-Driven Mathematical Reasoning
Mathematical reasoning has seen rapid progress in large language models (LLMs), yet existing methods optimize predominantly for final-answer correctness, raising the question whether models truly internalize mathematical concepts or merely memorize solution patterns. In human mathematics education, example-based reasoning such as constructing counterexamples to test theorem boundaries reflects deep conceptual understanding, but remains underdeveloped in current LLMs. Enhancing this capability through preference optimization presents two key challenges: (1) the model's limited example-based reasoning ability makes constructing effective preference pairs inherently difficult; and (2) capability acquisition is progressive, as the model must first learn to adopt this strategy before learning to apply it correctly. Therefore we propose INSPIRE, an Internalize-Then-Improve approach combining Reference-Guided Student Internalization (RGSI), which produces high-quality preference candidates under the policy model's own distribution, with a stage-wise rubric preference training strategy that decomposes learning into method-oriented and correctness-oriented stages. Experiments across multiple model scales and families demonstrate consistent improvements, even surpassing larger open-source models, while evaluations on out-of-distribution benchmarks confirm no degradation in general mathematical reasoning ability.
Read the full abstract
Machine Learning (Statistics)
Optimal Transport for Network Comparison: A Unified Review with New Spectral Bounds and Machine Learning Applications
Network comparison using optimal transport is a growing area of research in network science. Unlike standard graph metrics, optimal transport computes both network dissimilarity and a transport plan that explains how one graph morphs into another. In this paper, we review how optimal transport compares undirected, unweighted simple graphs using three primary distances: the Wasserstein, Gromov-Wasserstein, and Bures-Wasserstein distances. We examine the closed form of the Wasserstein distance in one dimension via node feature probability distributions, and show how the transport plans of the Wasserstein and Gromov-Wasserstein distances visualize how mass is shifted to transform one network into another. Beyond reviewing existing transport-based approaches, we establish new spectral lower and upper bounds for the Bures-Wasserstein distance and characterize the tightness of the lower bound under eigenbasis perturbations. Finally, we evaluate these distances using a synthetic network dataset for clustering and a real-world temporal network.
Read the full abstract
General Physics
Operations, Maintenance, and Industrial Scaling of MW-Class Orbital Data Centers
Megawatt-class orbital data centers require continuous maintenance, replacement, inventory, and service capacity in addition to spacecraft power/thermal systems. We formulate an analytical lifecycle framework for permanent/transient failures, modular orbital replacement units, robotic servicing, spare inventory, scheduled technology refresh, correlated faults, cybersecurity, optional human support. The model combines nonhomogeneous component hazards, capacity-weighted availability, multiclass robotic-service capacity, Poisson base-stock inventory, replacement-flow accounting, human-support break-even relations. For a 1 MW cluster with 10 active 100 kW nodes, 1 reserve node, ~200 5 kW compute cartridges, low, nominal, high deployed-mass allocations span ~50-75 kg/kW. Assumptions yield 70.2 random or life-limited interventions and 323-349 planned refresh operations/(MW-year), for a total of 393-419 standardized operations/(MW-year). Analysis gives a first-generation logistics of 5.3-9.0 t/(MW-year), with a nominal case of ~ 6.6 t/(MW year), 560-700 productive robot-hors/(MW-year). Planned refresh exceeds random replacement under the stated component populations, hazards, 3-15-year intervals. At ~400 standardized operations/(MW-year), the post-internal-recovery exception probability is <$10^{-3}$, with an objective near $10^{-4}$ at large scale; terminal non-recovery $p_U$ requires a smaller mission-level allocation. The target catastrophic-loss hazard for a 100 kW node is 0.01-0.03 1/yr. Parametric workload and cost cases place contingency visits at 10s of MWs, periodic campaigns at 10-100s of MWs, dedicated personnel at several 100 MWs to GWs. The reference first deployment is uncrewed, autonomously fault-managed, robotically maintainable, supported by specific inventory based on a 6-month replenishment horizon, compatible with later human access without permanent habitation.
Read the full abstract
Quantum Physics
Quantum Information at Infinity
We introduce the Quantum Information Space at Infinity (Quinfinity) $\mathcal{Q}_\infty$ as the inverse limit of the symbolic Quantum $N$-Spaces $\mathcal{Q}_N$, identified with the complete local real algebra of formal power series under the adic topology. The pure state pro-variety is formalized as the inverse limit of $N$-level pure state varieties, driven by the algebraic colimit of a direct system of real radical ideals. This real architecture internalizes the imaginary unit, collapsing the metric landscapes onto an invariant Riemannian isometry, which identifies the pro-variety directly with the classical Fisher-Rao statistical manifold. Within this framework, the Heisenberg uncertainty principle undergoes a structural regularization at infinity being intrinsically manifested as a localized truncation obstruction governed by adic algebraic derivations. Concurrently, the continuous Liouville-von Neumann equation operates as an internal derivation tangent to the pro-variety, while the open Gorini-Kossakowski-Sudarshan-Lindblad asymptotic master equation is intrinsically obtained via a non-associative symmetric Jordan product. This dissipative flow acts as a contracting radial vector field that dampens higher-order jet configurations, driving the state trajectories down the hierarchical tree toward the absolute zero element as a non-singular universal attractor.
Read the full abstract
Robotics
Beyond Relative Geometry: Metric-Aware Geometry Perception for Robotics
Recent embodied models increasingly leverage geometric representations to improve spatial reasoning and robotic manipulation. However, existing reconstruction methods only reconstruct relative geometry with arbitrary scales, causing predicted object dimensions and spatial distances to vary across scenes, viewpoints, and input configurations. This inconsistency prevents geometric perception from being directly aligned with robotic actions defined on the real-world scale. To address this limitation, we propose Metric-Aware Geometry Perception (MAGP), an end-to-end, plug-and-play framework for metric geometry reconstruction that can be seamlessly integrated into robotic policies. At its core, Metric Scale Equivariant Augmentation encourages the model to reconstruct metric geometry from camera parameters and depth observations, ensuring that the reconstructed geometry follows the metric scale specified by observations. Flexible Metric Conditioning further enables MAGP to support arbitrary view counts and combinations of camera and depth inputs, improving robustness to heterogeneous robotic sensing configurations. Together, these designs produce geometrically consistent reconstructions with stable object dimensions and spatial distances across scenes and sensing conditions. Experiments on ETH3D, MegaDepth, and ScanNet++ demonstrate that MAGP maintains strong relative geometry accuracy while reducing the absolute error by over an order of magnitude, from 2.01m to 0.07m. When integrated into multiple robotic policies, MAGP consistently improves performance on LIBERO, RoboTwin, and zero-shot LIBERO-Plus, with gains of up to 6.26% on RoboTwin. These results demonstrate the effectiveness and generalizability of metric geometry for robotic manipulation.
Read the full abstract
Cryptography and Security
ROPE: Routed Origin Policy Enforcement against Indirect Prompt Injection
Tool-using LLM agents execute consequential operations on behalf of users. Because the LLM driving an agent processes trusted user instructions and untrusted content in the same context, agents are vulnerable to indirect prompt injection attacks that induce unintended actions by mixing malicious instructions into benign data. Defenses based on prompting or content detection are easily evaded by paraphrasing attacks, while system-level defenses that constrain tool calls using inferred user intent, or restrict which sources can supply values to sensitive tools, greatly compromise utility. We present ROPE (Routed Origin Policy Enforcement), a defense that greatly improves the utility-security tradeoff by leveraging system roots of trust to create and enforce per-task policies on sensitive tool parameters. Our key insight is that authentication and integrity protection provide a basis for identifying the source of content independently of instructions embedded in it. Before a task unfolds, an LLM router reads only trusted task input to set a provenance rule for each sensitive tool parameter, accounting for the nature of the task and the values and sources the user specified. As the agent works, an origin labeler records where each piece of content came from, using the user's request, platform metadata, and protected record fields as trust anchors. Before any sensitive tool executes, a deterministic verifier requires each of its origin-guarded values to match a value in a source its rule accepts. We show that ROPE provably guarantees parameter value provenance to sensitive tool calls and is immune to paraphrasing attacks. Against eleven defenses on AgentDojo and AgentDyn under static, adaptive, and long-horizon attacks, ROPE exhibits near-perfect security while compromising far less utility than the best baseline defenses. Code and data are available at https://github.com/xhOwenMa/ROPE.
Read the full abstract
Cosmology and Nongalactic Astrophysics
The Atacama Cosmology Telescope: Constraints on the anisotropic screening and birefringence effects with DR6
While we have learned much from recent surveys of the mm-wave sky, many secondary anisotropies of the CMB remain under-explored. We consider two of these here: the anisotropic screening induced by Thomson scattering of CMB photons off free electrons, and cosmic birefringence, the rotation of CMB polarization due to parity-violating physics beyond the standard model. To measure these effects, we apply state-of-the-art estimators for statistical anisotropy to the ACT Data Release 6 CMB data and construct noisy maps and power spectra of both effects. For anisotropic screening, we find no significant detection with the baseline data, and place an upper limit on its power spectrum that is two orders of magnitude above predictions from current reionization models. We obtain evidence for some foreground contamination at one of our frequencies that is consistent with expectations from extragalactic simulations. For our anisotropic birefringence analysis, we also find no detection and place an upper limit on the signal that is consistent with previous analyses. We report a corresponding upper limit on the Chern-Simons coupling term $g_{a \gamma} < 4.9 \times 10^{-2} / H_\mathrm{I}$, with $H_\mathrm{I}$ the inflationary Hubble scale. These analyses pave the way toward obtaining improved constraints with the next generation of ground-based CMB surveys.
Read the full abstract
Quantum Physics
Spectral Fingerprints of Gauge Theories on a Quantum Computer
Maximally mixed state spectral sampling is an unbiased quantum algorithm that allows for extraction of a finite-resolution spectral distribution from a Hamiltonian over potentially the entire allowed range of energies. We show how it may be focused on any desired area of the spectrum in order to learn about the full fingerprint of the model of interest: from its ground state phenomena such as quantum criticality, obtained from the lowest lying energies, to its thermalization behavior, obtained from the mid-spectrum. We demonstrate this technique specifically on a $(1+1)d$ non-Abelian $SO(3)$ gauge theory, providing a comprehensive analysis of the steps necessary for performing this algorithm, as well as what is possible in the near-term with superconducting quantum hardware, performing simulations with circuits that are $78$ two-qubit gates deep. We show how this algorithm is able to take advantage of emerging dynamical circuit capabilities in near-term hardware to roughly halve the number required qubits, as well as how quantum readout error mitigation is trivial for this method. Along the way, we propose a novel strategy for compiling the controlled-time evolutions needed for spectral sampling by means of Pauli-frame optimizations. We illustrate two physical applications of quantum spectral sampling – disordered many-body transitions, and mid-spectrum densities of states – and what postprocessing steps they require beyond the Fourier outputs of the algorithm.
Read the full abstract
Computer Vision
UrbanGround: From Local Perception to Spatial Agency in a Real-Scale City
Multimodal large language models (MLLMs) can interpret a street view, but reliable urban action depends on whether such local evidence remains useful after the agent starts to move. In this paper, we investigate how far current MLLM agents can turn local urban perception into reliable action in a real-scale city. We propose UrbanGround, an urban sandbox built from Hong Kong's territory-wide 3D geospatial data. It combines the city's geographic structure with continuous, collision-constrained control through a shared evaluation interface. Agents use first-person observations and an interactive map to select actions across tasks ranging from local question answering to long-horizon navigation. Our analysis follows the growth of the spatial problem through three research questions. We first test whether an agent can gather and interpret local visual evidence to answer spatial questions. Then we ask whether these abilities support navigation as destinations become farther away and less explicit. Finally, we examine whether the resulting behavior survives changes in route availability and pedestrian motion. MLLM agents usually show useful atomic abilities in visual recognition and short-range spatial reasoning, while orientation and pedestrian-aware movement remain unreliable. Their central failure emerges over extended exploration, where local abilities do not compose into sustained goal-directed behavior and errors accumulate without effective correction. We hope UrbanGround will support broader study of how far MLLM agents can explore reliably in open-ended urban environments.
Read the full abstract
Computation and Language
CritICL: Inference-Time Weak-to-Strong Generalization from Small Language Model Failure Modes
Recent advances in inference-time scaling have significantly improved the reasoning performance of large language models (LLMs). However, these methods typically rely on repeated generation or external verification. To address this limitation, we introduce CritICL, a novel inference-time framework that improves reasoning while maintaining high efficiency. Our key insight is that LLM failure modes exhibit structured patterns across model scales within the same family. Instead of treating failures as undesirable outputs, CritICL leverages them as a source of guidance. Specifically, we utilize failure modes derived from weaker models and incorporate them into inference through critique-based in-context examples. We propose two variants: CritICL-dynamic, which adaptively predicts input-specific failure modes and retrieves critiques, and CritICL-static, which uses a global failure mode profile to provide stable guidance. Experimental results show that CritICL consistently outperforms standard in-context learning and achieves performance competitive with or superior to test-time scaling methods, while requiring significantly fewer generations and lower token cost. Code available at: https://github.com/umwyf/CRITICL
Read the full abstract
Artificial Intelligence
WikiSkill: Compiling Agent Experience into Persistent Knowledge for Skill Evolution
Agent skills package specialized knowledge and workflows into reusable resources that extend AI agent capabilities. Recent work automatically discovers such skills from agent experience, which enables agents to progressively adapt through interaction. However, the insights that guide skill development typically remain scattered across optimization histories, limiting their systematic reuse across iterations. We introduce WikiSkill, a framework that co-evolves agent skills with a persistent knowledge base (wiki). At a high level, WikiSkill separates raw execution experience, accumulated knowledge, and executable skills, while continuously consolidating experience into the wiki, which subsequent skill updates can build on. Across diverse benchmarks and models, WikiSkill consistently outperforms state-of-the-art skill-evolution methods and improves over no-skill baselines in most model-benchmark settings. We find that skill evolution complements model scaling: larger models generally benefit more from evolved skills, while smaller models with skills can outperform substantially larger models without them. We also find that evolved skills transfer effectively across models and model families, and skills evolved by other models can outperform self-evolved skills. Finally, our ablation studies confirm that persistent knowledge accumulation in the wiki is critical for effective skill evolution. These results demonstrate the benefits of systematically accumulating and refining agent experience for developing reusable and transferable skills.
Read the full abstract
Strongly Correlated Electrons
Metallogenic quantum criticality: Fermi surface nucleation at transitions between gapped phases
Continuous quantum phase transitions in interacting many-body systems exhibit universal phenomena that are largely independent of microscopic details. An organizing principle distilled from canonical examples such as the magnetic transition in Ising models is that the low-energy physics of transitions between gapped phases is governed by scale-invariant quantum field theories with gapless degrees of freedom at isolated points in momentum space. In this paper, we show that this principle can fail in a radical way: an entire gapless Fermi surface can emerge at a quantum critical point separating two fully gapped insulators. These exotic phase transitions are dubbed metallogenic quantum critical points (MGQCP). We construct a controlled family of effective field theories for MGQCPs between distinct fractional Chern insulators and propose a periodically modulated quantum Hall bilayer as a minimal microscopic setting. The critical theory exhibits an unusual mixture of physical signatures: while transport and compressibility follow scaling laws compatible with a conformal field theory, the electron spectral function develops a power-law non-analyticity at the Fermi momentum, characteristic of non-Fermi-liquid metals. We discuss how the concept of MGQCP may shed new light on critical phenomena in moire materials with tunable Chern bands. More generally, our results point to a broader class of quantum critical phenomena in which extended manifolds of gapless excitations emerge in momentum space despite being absent in both neighboring phases.
Read the full abstract
Mesoscale and Nanoscale Physics
Distinguishing Quantum Capacitance Signatures of a Topological Majorana Wire from a Normal Wire Segment
Majorana zero modes (MZMs) are spatially separated, near-zero-energy excitations expected at the ends of a topological superconducting (TS) wire. A quantum-dot interferometer can be used to probe the quantum capacitance of the TS wire, and the location and magnitude of the capacitance resonances provide information about the MZMs, while their magnetic flux dependence probes coherent coupling to the two Majorana modes. It has been recently shown that, a gapless (i.e., $\Delta=0$) wire segment can also exhibit flux-dependent quantum capacitance oscillations through Aharonov-Bohm interference and, with suitable tuning, can reproduce a Majorana-like response. Here we show that the two mechanisms can be distinguished experimentally. In the gapless normal wire segment, the two parity-dependent signals originate from separate energy resonances corresponding to the lack of generic zero-energy states. By contrast, in the topological wire, the pair of low energy levels with even and odd parity are nearly degenerate in energy, and therefore, the two parity branches remain within the same broader resonance region in the quantum-dot potential, and persist under independent variations of the dot potential and wire chemical potential. Our results show that the experimentally observed quantum capacitance response is distinguished not by a particular flux trace at one optimized point in parameter space, but by its persistence over a finite range of independently controlled parameters, including the quantum-dot potential. This parameter space stability provides a direct means of ruling out the gapless normal wire segment as the origin of the observed Majorana-like response.
Read the full abstract
Programming Languages
Tacet: A Language and Type System for Automatic Statistical Validity Accounting
Empirical comparisons between systems are a standard form of evidence in computer science research, but few are checked for statistical validity: most are never framed as statistical tests at all. Existing multiple-comparison procedures could control the resulting error, but need inputs (what an analysis examined, and how its observations are arranged) that are not recoverable from a list of p-values. We introduce Tacet, a language in which an analysis declares what it generated, states what it expects to find, and is refused any claim it cannot afford or cannot properly test. Its core calculus T pairs a free estimation sublanguage, carrying a reported footprint and a purity bit that records whether any outcome was consulted in building a value, with a priced claim sublanguage, carrying a wealth transformer, connected only by a mechanism that prices a comparison. A sample selected by reading outcomes sets the purity bit and is recorded as having examined everything it read, permanently, so it can never be granted a one-sided or confirmatory price, without the system ever asking whether the analyst intended to cherry-pick. Whether a comparison is paired or clustered is computed statically from the artifact schema, from declared functional dependencies between key fields alone and before any data is read, and a mechanism that assumes that structure away is refused rather than priced. Because the wealth transformer is antitone in the realized p-value, affordability can be checked before the analysis runs too, turning pre-registration into a typing rule. We prove the metatheory machine-checked in Lean 4 with no admitted gaps, and demonstrate the approach on a reference implementation and two case studies on published artifacts, the SWE-bench Verified leaderboard and BIG-Bench Hard.
Read the full abstract
Algebraic Topology
Koszul duality and Morita categories
We prove an $(\infty,2)$-categorical version of Koszul duality for operads and cooperads in spectra by showing that there is an equivalence between a Morita category of operads and bimodules and a dual Morita category of cooperads and bicomodules. This result subsumes various forms of operadic Koszul duality present in the literature.
Read the full abstract
Software Engineering
SWE-Prime: Fewer Trajectories, Better Performance
To improve large language models' ability to resolve real-world software issues, prior work has focused on constructing large-scale agent trajectory datasets and performing supervised fine-tuning (SFT) on successful trajectories. However, task success does not guarantee high-quality supervision: successful trajectories may still contain ineffective, redundant, or risky steps. Directly using such trajectories for SFT can introduce noisy supervision and encourage models to imitate undesirable problem-solving behaviors. Therefore, we propose SWE-Prime, a multi-granularity, two-stage SFT data selection method that progressively filters training data at the trajectory and segment levels. Specifically, the first stage performs trajectory-level screening based on process quality, result quality, and data representativeness, selecting a high-quality and representative subset of successful trajectories. The second stage performs segment-level selection by grouping consecutive steps into semantic segments and assessing each segment based on its contribution to the final solution, learnability, and potential risks. During SFT, all segments remain in the sequence to preserve context, while only selected segments contribute to the loss computation. Experiments on SWE-Bench Pro and SWE-Bench Verified show that training on the 10% trajectory subset selected by SWE-Prime outperforms training on the full resolved dataset, yielding relative performance gains of up to 12.2% and 24.2%, respectively.