arXiv:2606.26651v1 Announce Type: new Abstract: We study the online resource allocation problem, where a seller sequentially receives independent requests for $m$ types of resources with limited supplies from $n$ heterogeneous agents arriving in an unknown order. Each request from an agent can be fulfilled in different ways, with resource consumption in $[0,1]^m$, and generates different values for the agent. The objective of the seller is to maximize the social welfare, which is the sum of the values obtained from each agent. Recently, Ghuge, Singla, and Wang [GSW STOC'25] studied the learnability of the online resource allocation problem with heterogeneous agents and proposed a learnable pricing algorithm using only a single sample. However, their core algorithm is a dynamic pricing algorithm, which may introduce fairness concerns, as different agents face different prices. Furthermore, the algorithm crucially needs to know the arrival order of the agents in advance. To address these issues, in this paper, we study the learnability of anonymous pricing algorithms for online resource allocation using samples and queries to agents' value distributions. First, we show that a polynomial number of samples suffices to learn the classic dual pricing algorithm. Second, we show that a polynomial number of pricing queries suffices to learn a near-optimal anonymous pricing algorithm, in which the item pricing vector faced by each agent is drawn from the same predetermined distribution.
Science Journals
arXiv:2508.02753v5 Announce Type: replace Abstract: Time Series Forecasting (TSF) faces persistent challenges in modeling intricate temporal dependencies across different scales. Despite recent advances leveraging different decomposition operations and novel architectures based on CNN, MLP or Transformer, existing methods still struggle with static decomposition strategies, fragmented dependency modeling, and inflexible fusion mechanisms, limiting their ability to model intricate temporal dependencies. To explicitly solve the mentioned three problems respectively, we propose a novel Dynamic Multi-Scale Coordination Framework (DMSC) with Multi-Scale Patch Decomposition block (EMPD), Triad Interaction Block (TIB) and Adaptive Scale Routing MoE block (ASR-MoE). Specifically, EMPD is designed as a built-in component to dynamically segment sequences into hierarchical patches with exponentially scaled granularities, eliminating predefined scale constraints through input-adaptive patch adjustment. TIB then jointly models intra-patch, inter-patch, and cross-variable dependencies within each layer's decomposed representations. EMPD and TIB are jointly integrated into layers forming a multi-layer progressive cascade architecture, where coarse-grained representations from earlier layers adaptively guide fine-grained feature extraction in subsequent layers via gated pathways. And ASR-MoE dynamically fuses multi-scale predictions by leveraging specialized global and local experts with temporal-aware weighting. Comprehensive experiments on thirteen real-world benchmarks demonstrate that DMSC consistently maintains state-of-the-art (SOTA) performance and superior computational efficiency for TSF tasks. Code is available at https://github.com/1327679995/DMSC.
arXiv:2606.27030v1 Announce Type: new Abstract: Polariton interferometry through optical near-field microscopy has become a powerful tool in nanophotonics, enabling direct spatial access to the propagation characteristics of strongly confined, evanescent polariton modes. Scattering-type near-field optical microscopy has matured as the prime tool for such studies, yet mostly the out-of-plane components of the optical near fields are probed, owing to the elongated geometry of the nanotip. Here, we demonstrate a complementary far-field nonlinear microscopy approach which allows to selectively probe in-plane polariton field components. Accessing the full vector field is interesting when studying complex mode patterns such as hyperbolic polaritons or skyrmions, where the in-plane field components are typically only inferred from the out-of-plane component but not measured directly. To this end, we use nonlinear infrared-visible wide-field sum-frequency generation microscopy, where the short visible wavelength of the nonlinear signal provides the high spatial resolution to access evanescent modes in the infrared. The symmetry selection rules of the nonlinear process further enable polarization-selective imaging of both in-plane polariton field components through spatial interferometry. The concept is demonstrated experimentally using surface phonon polaritons at the AlN-air interface launched by a gold antenna. A simple, semi-analytical model reproduces the peculiar propagation patterns. Hyperspectral imaging with a tunable narrowband laser further gives access to the polariton dispersion. The wide-field methodology holds high promise for in-depth and high-throughput studies of infrared nanophotonic structures.
arXiv:2606.27032v1 Announce Type: new Abstract: Energy trading decisions depend not only on current market prices, but also on expected future market conditions, and operational constraints. This makes the state representation given to a reinforcement learning agent an important design choice. We study this in HydroDam, a pumped-storage arbitrage environment, using a fixed Double DQN agent. The environment, action space, reward function, network, and training protocol are kept fixed; only the market features are changed. We compare absolute price/calendar features, relative features that compare current prices with recent market history, forecast features, and all combinations of these three feature families. Policies are trained and selected using 2007--2011 Belgian day-ahead prices and evaluated on two test settings: a later same-market test set from 2012--2025 and 39 other ENTSO-E market zones. Absolute features only reaches 28.8% on the test set and a median 5.7% across zones. Relative-only and forecast-only states also stay below a rolling price-score heuristic in the cross-zone median. Combining feature families is much stronger: absolute + relative reaches 49.9% on the test set and a 39.8% cross-zone median, while absolute + relative + forecast reaches 55.6% and 47.5%. These results suggest that state representation is not a minor preprocessing choice in storage-trading RL, but a central part of the policy design: robust transfer requires combining price scale, recent relative price context, and short-horizon forecast information, rather than relying on any single feature family.
arXiv:2606.27043v1 Announce Type: new Abstract: Open quantum systems generally do not perfectly preserve phase coherence: coupling to uncontrolled environments requires a density-matrix description based on the Liouvillian framework beyond pure-state wave evolution. Realizing and probing such dynamics in a programmable platform is therefore essential for connecting coherent physics to realistic dissipative settings. Here we implement a tunable open-system quantum walk in a photonic mesh lattice, where controlled phase noise produces adjustable dephasing and non-reciprocal gain-loss imbalance provides an independently tunable non-Hermitian drive. This allows us to continuously interpolate between coherent quantum walks and incoherent classical walks, and to observe how directional transport evolves in the Liouvillian regime. Using non-Hermitian skin dynamics as a probe, we measure the center-of-mass drift over both the coherence and non-Hermiticity parameters, revealing a crossover from coherence-enhanced to decoherence-enhanced transport in quantitative agreement with quantum-channel simulations. We further program spatial and temporal interfaces to demonstrate interface accumulation and a long-time drift governed by the instantaneous channel. Our results establish a controllable photonic platform for simulating open quantum dynamics and show that decoherence can actively reshape non-Hermitian transport.
arXiv:2606.27089v1 Announce Type: new Abstract: Modern image generation model rapidly grows their sizes to meet high-fidelity image synthesis. However, they gradually become unaffordable for their enormous parameter consumption and computation budget that lead to massive resources requirement and gpu memory footprint. In this paper, we propose TMP, the first Tree-structured Mixed-policy Pruning framework that generalizes prevalent image tasks (T2I and TI2I) and architectures (Mixture-of-Experts (MoE) and Diffusion transformer (DiT)). It could be applied to the step-distilled models and contribute as the last stage. We perform experiments upon current open-sourced SOTA HunyuanImage-3.0 instruct and a popular efficient model Z-Image turbo. The proposed pruning framework manages to compress HunyuanImage 3.0 from 80B to 20B parameters at 75% reduction ratio, sacrificing limited generation quality. We also optimize to enable the inference of the pruned 20B version of HunyuanImage 3.0 on a single 24GB 4090 GPU by engineering skills. The inference script and model weight have been integrated into the existing HunyuanImage3.0 open-source github and huggingface repository. Besides, we prove the efficacy of TMP by compressing Z-Image turbo from 6B to 4B (33% reduction) with negligible degradation.
arXiv:2606.27091v1 Announce Type: new Abstract: LLMs fine-tuned for security classification are usually evaluated on held-out examples from the same distribution as their training data. We show that this can miss vulnerabilities introduced by fine-tuning itself: models can learn token-level indicator semantics that preserve canonical accuracy while failing under behavior-preserving transformations such as PowerShell alias substitution, command reconstruction, string construction, execution indirection, and case mutation. We study Foundation-Sec-8B-Instruct and its base model, Llama-3.1-8B-Instruct, on matched PowerShell classification cohorts. Causal interventions localize the classification circuit to a late-attention route inherited from Llama rather than created by fine-tuning. Fine-tuning concentrates and semantically specializes this inherited structure, improving baseline behavior while creating transformation-sensitive attack surfaces. A three-tier evasion benchmark finds Foundation-Sec misses on iwr substitution, Invoke-Expression reconstruction, and case-mutated Invoke-Expression/IEX variants that Llama does not share. We also derive a pre-deployment monitoring method: a linear probe at the classification boundary and an indicator-token sign test identify command families where canonical indicators change role after fine-tuning. These signals prioritize red-team variant generation using only canonical inputs, showing that security fine-tuning can improve task accuracy while expanding the evasion surface. These results caution against treating small task-specific fine-tunes as straightforwardly safer security classifiers: specialization can convert inherited model structure into brittle indicator rules that preserve held-out accuracy while expanding the evasion surface. Robust AI-enabled security will require specifying the full transformation space of the task and monitoring semantic drift through fine-tuning.
arXiv:2606.27140v1 Announce Type: new Abstract: We develop the fTNN, a deterministic tensor neural network subspace method for problems involving the fractional Laplacian on bounded domains, taking the fractional Poisson equation and time-dependent fractional advection-diffusion equation as typical representatives. The work employs a geometry-adapted integration split featuring a spatially dependent near-field radius, which decomposes the fractional Laplacian into three contributions: a singular near field, a regular interior far field, and an analytical exterior far field. Then the singular radial integrals are treated by Gauss-Jacobi quadrature, the regular radial integrals by Gauss quadrature, and the angular variables by deterministic angular quadrature, yielding a fully deterministic integration framework of the fractional Laplacian operator. To accurately resolve low-regularity solutions and the associated loss functional, we construct boundary-singularity-aware trial functions enriched with explicit boundary features, and propose two strategies for automatically selecting the leading exponent and evaluating the loss function from the singularity structure induced by the fractional operator, or jointly by the fractional operator and the source term. For time-dependent fractional PDEs, we design a spatiotemporally separable neural network that factorizes the time-space residual into a sum of low-dimensional temporal and spatial integrals, and we integrate this representation with an alternating neural network subspace optimization strategy for efficient training. Numerical experiments show that the proposed framework attains high accuracy on the tested benchmarks and improves substantially over existing fPINN and Monte Carlo baselines, particularly for problems with strong boundary singularities and long-time simulations.
arXiv:2606.27146v1 Announce Type: new Abstract: Long-horizon robotic manipulation is highly sensitive to physically infeasible transitions, contact-induced disturbances, and the lack of effective self-correction during execution. Although Vision-Language-Action (VLA) models provide strong task grounding through multimodal learning, they typically generate actions in a feed-forward manner without explicitly checking physical feasibility or diagnosing execution errors online. We present PhysReflect-VLA, a plug-and-play execution-time reliability framework that augments VLA policies with physical feasibility evaluation and structured self-reflection in a closed-loop control pipeline. A Feasibility Operator evaluates whether candidate actions induce dynamically consistent state transitions; an Action Explanation Operator verifies transition coherence; and an LLM-based Reflection Module analyzes state discrepancies to generate corrective guidance for subsequent actions. A two-stage training procedure stabilizes feasibility modeling and integrates reflection into the control loop. Experiments on multi-stage, contact-rich real-world manipulation tasks show consistent improvements in stage-wise stability and overall task success compared with representative VLA baselines with an average gain of 5.4\%. Ablation results further indicate that feasibility checking and reflection-based correction both contribute to improved execution robustness. These results highlight the importance of embedding physical consistency checks and online self-reflection for reliable long-horizon robotic manipulation.
arXiv:2606.27160v1 Announce Type: new Abstract: We follow up our study of basis set truncation errors for atoms in magnetic fields [{\AA}str\"om and Lehtola, J. Phys. Chem. A, 2023, 127, 10872]. Our previous study employed an approximate real-valued model. In this work, we implement a scheme to allow the use of complex basis functions and the true, complex Hamiltonian with linear molecules in a parallel magnetic field within the usual real-basis machinery of quantum chemistry. Our method performs additional unitary transformations before and after a conventional Fock build, thus allowing the reuse of existing software methods and algorithms. We apply our approach to calculations on low-lying configurations of the atoms $Z \leq 18$ in static magnetic fields up to 0.6 $B_0$. The calculations employ the uncontracted aug-cc-pVTZ and the benchmarking quality AHGBSP3-9 Gaussian-type orbital basis sets. We compare total energies obtained with real and complex orbitals using these basis sets to fully numerical ones at the complete basis set limit. We see that the states of the real-valued Hamiltonian are superpositions of the true eigenstates that are correctly captured by the complex calculations. Our results show that the complex basis machinery is necessary for targeting states with the correct symmetry for the studied range of magnetic field strengths. The novel tool is key for future work where we aim to optimize basis sets for finite-field calculations.
arXiv:2606.27161v1 Announce Type: new Abstract: Multimodal large language models (MLLMs) have achieved strong multimodal reasoning capabilities, but their efficiency is limited by the large number of visual tokens, which introduces substantial computational overhead. Visual token pruning offers a natural solution, yet existing methods are imperfect: attention-based criteria tend to retain redundant tokens, while diversity-based criteria are often agnostic to user instructions. Even methods that combine multiple criteria still lack a principled formulation of the intrinsic objective of token pruning. In this paper, we revisit visual token pruning from a first-principles perspective and formulate it as constructing Token Optimal Preservation Sets. Through a top-down information-theoretic analysis, we identify three fundamental principles for effective token selection: Task Relevance, Information Coverage, and Semantic Diversity. Based on these principles, we propose TOPS, a training-free and model-agnostic pruning module that can be applied to various MLLMs. Extensive experiments on 7 MLLM backbones and 14 benchmarks demonstrate that TOPS outperforms prior methods under diverse pruning settings. Notably, on LLaVA-NeXT, TOPS removes 77.8% of visual tokens while preserving 100.0% and 100.6% performance on its 7B and 13B models, respectively, suggesting that pruning redundant visual tokens can sometimes mitigate hallucination and inspire future lightweight MLLM design.
arXiv:2606.26677v1 Announce Type: new Abstract: Langton's ant is a simple two-dimensional cellular automaton whose long-term behavior exhibits remarkable complexity. While it is known that the ant eventually escapes any finite connected region of the grid, the quantitative aspects of this escape remain poorly understood. In this paper, we study the escaping time of Langton's ant, defined as the maximum number of steps the ant can perform within a finite connected domain before leaving it. We establish general upper bounds on the escaping time as a function of the domain size, and derive improved bounds for rectangular domains. In particular, we obtain a factorial upper bound for square domains via an inductive decomposition argument. We also obtain linear upper bounds for rectangular domains of height two and three via a column-by-column analysis. More generally, for rectangular domains with a fixed height, we establish a polynomial upper bound in the number of columns. These results are complemented by exact values computed through an optimized simulation algorithm that exploits the geometric symmetries of the grid and employs a backtracking branching strategy to avoid exhaustive search over all color configurations. We also provide lower-bound constructions, proving that the linear upper bounds for rectangular domains of heights two and three are asymptotically optimal.
arXiv:2606.27280v1 Announce Type: new Abstract: We present a patch descriptor retrieval method that returns the exact nearest neighbour -- provably identical to exhaustive full-vector search -- while evaluating only a small fraction of the database, and does so deterministically: the same (database, query) pair always produces the same result, independent of run order, thread count, or hardware. This contrasts with approximate nearest-neighbour (ANN) approaches such as HNSW and IVF-PQ, which trade exactness for speed and may return different results across runs. The enabling mechanism is Hierarchical Normalization (HN): a normalisation scheme that splits the pre-normalisation feature vector into a K-dim major component (norm sqrt(1-alpha)) and a (128-K)-dim minor component (norm sqrt(alpha)). Since the minor inner product is bounded by alpha (Cauchy-Schwarz on the prescribed norms), the major similarity plus alpha is an admissible upper bound on the full similarity: the search scans the K-dim major component for all entries, then applies full 128-dim evaluation only to entries that cannot be pruned -- a provably exact branch-and-bound scan. We train HN-modified HardNet on the notredame split of the UBC patch dataset and evaluate on trevi and halfdome. With a cache-optimised Structure-of-Arrays layout and K=8, alpha=1/32, the search achieves 13.7x (trevi) / 12.7x (halfdome) speed-up over brute-force 128-dim search, with only 0.4% of entries requiring full evaluation. At K=16, alpha=1/8, FPR@95 rises from 0.0062 to 0.0064 on trevi at 7.2x speed-up, with 98.8% of entries bypassing full evaluation.
arXiv:2606.27281v1 Announce Type: new Abstract: Small language models can run locally on consumer hardware, but structured reasoning often pushes users toward repeated sampling or larger models. This article studies a bounded neuro-symbolic alternative for local inference: a model translates a problem into typed finite-domain rules and constraints, a symbolic layer checks traceability and consistency, and a deterministic solver performs the reasoning step. The resulting Verifiable Formalization and Repair pipeline (VFR-LLM) tests when symbolic verification can replace repeated sampling without weakening accuracy. We evaluate it through LM Studio on Apple Silicon, using Qwen3-4B-2507 as the primary model, with Phi-4-mini-reasoning and Gemma-3n-E4B as robustness checks. On 120 generated pure-precedence problems, Qwen VFR-LLM achieves 0.983 accuracy, versus 0.700 for serial self-consistency using one model call instead of five. On a 120-instance BBH-derived extended logical-deduction subset, it reaches 0.933 versus 0.283. The advantage persists against a stronger cost-aware adaptive self-consistency baseline, which lowers sampling cost but not the single-call accuracy gap. Gemma reproduces the same model-dependent boundaries and Phi is negative on typed constraints. The contribution is bounded: finite-domain logic can replace repeated local sampling for some structured tasks, saving model calls and serial latency, with stratum-dependent token savings.
arXiv:2606.27284v1 Announce Type: new Abstract: Gay dating applications have become critical platforms for sexual minority men to seek relationships and community, yet they also expose users to deceptive interactions that remain underexplored in HCI and CSCW research. This study examines how gay male users in China experience, identify, and respond to deception on dating applications. Through semi-structured interviews with 22 participants across platforms including Blued, Aloha, Fanka, and Soul, we make three contributions. First, we identify a typology of deceptive practices extending beyond profile misrepresentation to encompass relational, emotional, financial, and commercial forms of deception. Second, we document the layered, probabilistic verification strategies users develop through long-term platform use, showing that trust assessment operates as a multi-signal, provisional process rather than a binary judgment. Third, we demonstrate that risk recognition is a collaborative practice shaped by the circulation of experience, the abstraction of recurrent tactics, and the codification of shared rules within the community.
arXiv:2606.27286v1 Announce Type: new Abstract: Mechanistic epidemiological models are widely used to support infectious disease forecasting and public-health decision making. Bayesian calibration of such models is commonly performed using Markov chain Monte Carlo (MCMC), which can become computationally expensive for high-dimensional nonlinear systems and repeated near-real-time analyses. Here, we investigate simulation-based inference (SBI) using neural posterior estimation as a scalable alternative for Bayesian calibration of a mechanistic SECIR epidemiological model using COVID-19 intensive care unit (ICU) occupancy data from Germany during 2020. We compared SBI and MCMC across multiple epidemic phases using both 31-day inference windows and a substantially more challenging 201-day reconstruction problem involving multiple transmission change points. Posterior agreement was evaluated quantitatively using Wasserstein distances and Kullback-Leibler divergences together with posterior predictive checks. Across the 31-day windows, SBI recovered posterior distributions in strong agreement with MCMC while accurately reproducing observed ICU trajectories. In the 201-day setting, SBI preserved the dominant posterior structure despite increased uncertainty. SBI, by combining CPU and GPU resources, substantially reduced computational runtime compared with MCMC, which was restricted to running on CPUs. Whereas MCMC required approximately 1000 seconds for the 31-day inference problems, SBI achieved comparable posterior and predictive performance in approximately 60-70 seconds on a single GPU. For the 201-day inference problem, SBI required an average of 157 seconds, while the MCMC runs took over 19,000 seconds. Our results demonstrate that SBI provides a rapid and computationally efficient framework for Bayesian calibration of mechanistic epidemiological models, supporting repeated near-real-time inference and rapid outbreak analysis.
arXiv:2606.26683v1 Announce Type: new Abstract: Finite automata with translucent input letters are a recent model of discontinuous input processing. Basically, classical finite automata are equipped with a translucency function that defines, depending on the state, the set of translucent input symbols. While processing the input, translucent symbols are skipped and only visible symbols are read and processed. It is distinguished between deterministic and nondeterministic models and, in addition, between returning and non-returning models. In the former case, the automaton restarts from the left end of the input after having consumed some visible symbol, whereas in the latter case the automaton restarts from the left end of the input when the right endmarker symbol is seen. Returning finite automata with translucent letters have been introduced by Nagy and Otto and its non-returning variant has been introduced by Mraz and Otto. Many results concerning the computational capacity, relations between deterministic and nondeterministic models, and relations between returning and non-returning models are known. Moreover, some results on closure properties and decidability questions have been obtained as well. However, some questions have still been open since many years. In this paper, we will give answers to some of these open questions. In particular, we show the non-closure under concatenation, Kleene star, reversal, and inverse homomorphism for the non-returning deterministic as well as nondeterministic model. We also obtain non-closure under inverse homomorphism for the returning deterministic and nondeterministic model. Finally, we investigate the emptiness problem for non-returning finite automata with translucent input letters and show the decidability of the problem in case of deterministic as well as nondeterministic automata.
arXiv:2606.26684v1 Announce Type: new Abstract: Parallel Communicating Finite Automata (PCFA) are systems of several finite automata that can communicate by requesting the state of another automaton. As an attempt to make PCFA as defined in [8] and [2] more realistic, the non-forgetting model is introduced where automata retain their own states while querying. As in previous publications, several variants of these automata systems are considered. The computational capacity of the non-forgetting model is investigated and compared to ''forgetting'' systems and multi-head finite automata. It is shown that most variants of non-forgetting PCFA are as powerful as multi-head automata in the deterministic and nondeterministic cases where the number of automata equals the number of heads. The only exception is the special case of deterministic centralized systems in non-returning mode. Here, a strict inclusion is proved. In the course of this proving, a proof from [2] is completed. With that, this paper answers some questions for nfPCFA that are open for the classical ''forgetting'' case.
arXiv:2606.26685v1 Announce Type: new Abstract: In the past literature, families of two-way finite automata and pushdown automata having limited state complexity (i.e., the total number of inner states) and stack-state complexity (i.e., the total number of inner states multiplied by the total number of strings "pushable" to a stack), have been studied in direct connection to (mainstream) space-bounded complexity classes equipped with Karp-Lipton style advice of limited size when all inputs given to the automata have bounded length. Here, we acknowledge two major factors -- size and ceiling -- of such families, which have a significant impact on the complexity of finite and pushdown automata families, where the "size" refers to (stack-)state complexity and the "ceiling" refers to an input's length bound. In this line of study, we further explore those effects caused by different sizes and ceilings.
arXiv:2606.27325v1 Announce Type: new Abstract: Recent advances in action-conditioned world models show promising progress in modeling complex interactions and forecasting future states under diverse action sequences. While these models are often driven by stronger visual representations and model capacity, action conditioning itself remains underexplored. Most existing approaches compress the entire action sequence into a single representation, which works well for low-DoF control but becomes less reliable in high-DoF scenarios. We observe that high-DoF dexterous actions are inherently heterogeneous, spanning multiple orders of magnitude, where large-scale motions coexist with subtle but important signals. When uniformly aggregated, optimization exhibits an imbalance across action components, which hinders the modeling of fine-grained effects and affects action fidelity. We therefore propose DexAC-WM, which treats action conditioning as a structured process rather than global compression. DexAC preserves dimension-level semantics via action tokenization and aligns action signals with visual dynamics through local refinement and global modulation. To address the limited high-level semantic grounding in existing world models, we further introduce a semantic branch that provides rich object-scene priors, which enables world model to capture dynamic visual details while supporting high-DoF action-conditioned video prediction. Experiments on EgoDex and EgoVerse show that combining the semantic branch with DexAC significantly improves FID, FVD, and PCK, demonstrating gains in visual-temporal realism and action-following consistency. We further verify that DexAC extends to other backbones, showing the scalability of our structured action-conditioning design. These results suggest that scaling world models to high-DoF control requires both structured action modeling and semantic grounding.
arXiv:2606.27326v1 Announce Type: new Abstract: Modern generative world models render increasingly realistic action-controllable futures, yet they frequently hallucinate: rollouts remain visually fluent while drifting from the ground-truth dynamics. We hypothesize that hallucination concentrates in low-coverage regions of the state-action space, where lightweight data-centric signals can both detect it and guide mitigation. To test this, we introduce MMBench2, a 427-hour, 210-task dataset for visual world modeling with ground-truth actions, rewards, and live simulators, and train a 350M-parameter world model on it. We identify three distinct hallucination modes: perceptual, action-marginalized, and scene-diverging -- each anchored to a different stage of the pipeline, and develop three signals that accurately predict where the model will fail. To close coverage gaps at training time, we develop a coverage-aware sampling technique; to close them online, our hallucination predictors serve as curiosity rewards for targeted data collection, yielding a data-efficient finetuning recipe that adapts the pretrained world model to entirely unseen environments with as few as 50 real environment trajectories. Overall, our findings reveal that hallucination in world models is inherently a data coverage issue, and that the same signals used to detect it can also be used for mitigation. An interactive web version of our paper is available at https://www.nicklashansen.com/mmbench2
arXiv:2606.26700v1 Announce Type: new Abstract: Motion feasibility prediction plays a central role in robotics, particularly in task and motion planning and manipulation. A major bottleneck for this problem in cluttered environments is that infeasible planning attempts by Sampling-based motion planners (SBMPs) can incur substantial computational cost. Also existing approaches for infeasibility certification are limited to low-dimensional configuration spaces and often assume simplified geometric environments represented by primitive objects with known parameters. We study the complementary problem of learning motion feasibility prediction directly from raw RGB-D observations for a 7-DOF manipulator operating in realistic cluttered scenes. We introduce the first large-scale benchmark for this setting, comprising 2.7M grasp feasibility labels over 88 scanned objects and 190 cluttered tabletop scenes. We benchmark three representative classifier families spanning MLP- based, volumetric-CNN, and point-cloud-based Transformer architectures under matched training conditions. Our best model, GRASPFC-PTX (a point-cloud transformer), achieves an AUROC of 0.996 on Novel objects while providing predictions significantly faster than SBMPs.
arXiv:2606.27354v1 Announce Type: new Abstract: Neural surrogate models offer fast approximate mappings from PDE parameters to solutions, but they typically treat solving as a purely statistical task: once trained, they struggle to correct their own constraint violations and extrapolate beyond the training distribution. Recent hybrid methods promote physical correctness by targeting the PDE residual via gradient descent or Gauss--Newton steps, but inherit the compute cost and instability of the underlying classical optimizers. We show, theoretically and empirically, that numerically minimizing the PDE residual can be an unreliable proxy for reconstruction accuracy in ill-conditioned systems, explaining why these methods often do not make accurate predictions despite achieving low residuals. We propose error-conditioned Neural Solvers (ENS), built on a different principle: rather than an optimization target, the PDE residual field is passed as a direct input to the network at each iteration, enabling it to read the spatial structure of its own errors and learn an update policy to iteratively correct its predictions. Across four PDE families, ENS attains the highest prediction accuracy in the large majority of settings, with gains reaching $10\times$ on turbulent Kolmogorov flow, while avoiding the expensive compute cost of hybrid methods. ENS's learned correction policy generalizes under distribution shift, including zero-shot parameter changes and cross-equation transfer, where its relative advantage is largest in the ill-conditioned regimes where residual minimization is least reliable. Project website: https://neuralsolver.github.io/.
arXiv:2606.26227v1 Announce Type: cross Abstract: Multiscale temperature fluctuations are abundant in the intracluster medium (ICM) outside of galaxy cluster cores ($\sim 100~{\rm kpc}$). Their origin is often attributed to turbulent stirring by subhalos or accreting baryons crossing the virial radius. However, their apparent resistance to mixing and thermal conduction in a collisional medium has not been explained. We propose a new mechanism by which steady-state temperature fluctuations can form and persist outside the cluster core. Local thermal instability, or Field instability, is used to explain filamentary condensates in cluster cores but is usually dismissed outside them because thermal conduction should suppress instability. In weakly collisional or collisionless plasmas, however, thermal conduction can be anomalously suppressed by heat-flux-driven plasma instabilities triggered in presence of a local magnetic field, leading to two effects: (i) condensates form in a new parameter regime that overlaps with conditions outside the core, and (ii) condensates reach a steady state as in the hydrodynamic limit. This extends the regime of instability-driven fluctuations to over $\gtrsim50\%$ (depending on hot plasma temperature) of the cluster. We use one-dimensional hydrodynamic simulations of condensates to test our analytical ideas.
arXiv:2606.26238v1 Announce Type: cross Abstract: Odd diffusion is a hallmark of chiral active matter, generating currents transverse to density gradients. Existing theories rely on a linear antisymmetric transport coefficient that exists only in two dimensions, raising the question of whether odd diffusion can occur in isotropic three-dimensional systems. Here we show that such transport is possible through a nonlinear constitutive law. Symmetry considerations reveal that the three-dimensional Levi-Civita tensor permits a leading order isotropic odd current at second order in the density gradient expansion and only in multicomponent systems. The resulting transport generates boundary-driven rotational currents, finite vorticity, and enstrophy despite the absence of external torques or preferred directions. We show how such a constitutive law derives from a microscopic model of particles interacting through nonreciprocal three-body forces using the Dean--Kawasaki coarse-graining procedure. These results establish a minimal framework for odd transport in isotropic three dimensions.