arXiv:2607.03324v2 Announce Type: replace Abstract: Eco-friendly energy management for artificial intelligence data centers (AIDCs) is crucial because of the significant increase in energy consumption-induced carbon emissions from AIDCs resulting from the rapid expansion of AI applications. This paper proposes a hierarchical carbon-aware multi-agent reinforcement learning (CA-MARL) framework for robust and efficient operations of AIDCs under uncertainties while ensuring low-carbon operation of power distribution systems. The framework comprises a workload manager (WM) agent and multiple local AIDC agents trained using a multi-agent transformer method, corresponding to a global AIDC aggregator and a local AIDC operator, respectively. Leveraging AIDC operation data along with nodal carbon intensity (NCI) calculated from the carbon emission flow-integrated distribution system operator problem, the WM agent spatially allocates AI training and inference jobs among all AIDCs. Based on the jobs allocated from the WM agent and NCI information, each AIDC agent schedules economical and eco-friendly operations of the AIDC by performing the following tasks: i) temporal shifting of training jobs, ii) spatial allocation of training graphics processing unit (GPU) blocks and inference GPUs within the AIDC, and iii) control of the supply air temperature of the cooling system. The effectiveness of the proposed framework was assessed using an IEEE 33-node power distribution system.
Science Journals
arXiv:2607.03436v2 Announce Type: replace Abstract: On real open-model pools, 12--36% of the reported router-to-oracle gap is single-draw label noise that no single-commit router can capture, while the majority is genuine, recoverable specialist advantage; this work proves why (a recoverability asymmetry) and releases a protocol to measure it. Routing among large language models (LLMs) trades cost for quality, motivated by the gap between learned routers and a per-instance oracle. But under stochastic decoding that oracle is a single Bernoulli draw, not a reproducible property. We recast the question structurally: the expected oracle decomposes as $O^{\exp}=O^{\mathrm{repro}}+\Delta$, into reproducible single-commit headroom $O^{\mathrm{repro}}$ and a non-negative single-commit selection floor $\Delta$. Our main result is a recoverability asymmetry: this floor is closed by no single-commit router (deterministic or randomized), yet is provably recovered by test-time sampling: best-of-$K$ on the committed model, at the oracle's own budget, dominates the independent-pool single-draw oracle. This cap needs no cross-model independence, pinning "not recoverable" to single-commit selection, not to information. The floor's magnitude is a prospective, conservative localization, not an audit: LLMRouterBench (33 models, 391,645 instances) builds its oracle as a per-query union of single $T=0.2$ draws, so its 20-point gap is by construction a union of stochastic draws; since $O^{\mathrm{repro}}$ is non-identifiable at $k=1$, we re-estimate by fresh $k\ge20$ resampling under one-sided, dependence-corrected bounds. Across three controlled open-model re-generations (arithmetic, competition math, and non-math science), single-draw noise is a substantial minority of the gap, larger on unsaturated benchmarks and approaching half on the hardest queries. We release a multi-sample oracle protocol that routing benchmarks can adopt.
arXiv:2607.04393v2 Announce Type: replace Abstract: Atomic oven generating controllable atomic beam flux plays a fundamental role in quantum gas experiments. Here, we report a new heater design that can heat up an high temperature atomic oven with fast thermal response. The new heater shows a heating rate improved by 7.65 times comparing to that of the conventional resistive heater while the crucible temperature can heated up to 1200K. With this oven, we generated a collimated ytterbium beam with flux exceeding $10^{14} \text{ atoms/s}$ at 823 K. We believe that our design offers a promising solution for shortening experimental dead time and improve the experiment efficiency in cold atom researches.
arXiv:2607.06485v1 Announce Type: new Abstract: Vision-language models (VLMs) are increasingly deployed on infrared (IR) remote sensing imagery in security-critical settings, yet their adversarial robustness remains unexamined. We present AirflowAttack, to our knowledge the first adversarial attack for IR remote-sensing VLMs and the first to weaponize thermal-airflow turbulence as the perturbation prior. A lightweight generator synthesizes a single input-agnostic perturbation regularized toward physically plausible airflow patterns. Optimized on one surrogate CLIP model, it attains a mean zero-shot scene-classification attack success rate (ASR, the fraction of samples whose top-1 class changes) of 48.5% across five diverse CLIP backbones, far exceeding four IR-specific physical baselines (27.7--37.0%). Applied to six state-of-the-art VLMs, it cuts scene-classification accuracy by up to 38.2% relative, yet paradoxically makes some models more confident in their IR analysis, confabulating the perturbation as genuine thermal evidence such as temperature gradients and convection. Ablations show the airflow prior raises physical plausibility at no measurable cost to attack success. Together with a benchmark spanning eleven models and four tasks, these findings expose critical vulnerabilities in the rapidly expanding IR VLM ecosystem.
arXiv:2509.07162v2 Announce Type: replace Abstract: Autonomous multi-finger grasping is a fundamental capability in robotic manipulation. Optimization-based approaches show strong performance, but tend to be sensitive to initialization and are potentially time-consuming. As an alternative, the generator-evaluator-planner framework has been proposed. A generator generates grasp candidates, an evaluator ranks the proposed grasps, and a motion planner plans a trajectory to the highest-ranked grasp. If the planner doesn't find a trajectory, a new trajectory optimization is started with the next-best grasp as the target and so on. However, executing lower-ranked grasps means a lower chance of grasp success, and multiple trajectory optimizations are time-consuming. Alternatively, relaxing the threshold for motion planning accuracy allows for easier computation of a successful trajectory but implies lower accuracy in estimating grasp success likelihood. It's a lose-lose proposition: either spend more time finding a successful trajectory or have a worse estimate of grasp success. We propose a framework that plans trajectories to a set of generated grasp targets, the evaluator estimates the grasp success likelihood at the terminal configuration of the planned trajectories, and the robot executes the trajectory most likely to succeed. Our experiments show our approach improves over the traditional generator-evaluator-planner framework across different objects, generators, and motion planners, and successfully generalizes to novel environments in the real world, including different shelves and table heights. Project Website: https://martinmatak.github.io/fpte/
arXiv:2512.16368v2 Announce Type: replace-cross Abstract: We report on a simple and easy to implement method of feedback cooling trapped ions to temperatures below those achievable using only Doppler cooling. Additionally, the feedback cooling results in significantly shorter cooling times. For selected parameters, we demonstrate cooling to temperatures below $\hbar\Gamma/2 k_\mathrm{B}$. The motion of a single ion is monitored in real-time, allowing for the generation of a feedback signal that is applied to an auxiliary trap electrode. Motion detection is implemented by imaging the fluorescence photons emitted by the ion onto a knife edge and detecting the transmitted light, a method used so far to cool trapped nanoparticles. The intensity modulation of the fluorescence resulting from the ion motion is used to generate and apply the feedback signal and also to determine the ion temperature. While the method benefits from a high rate of detected scattered photons, which can be a challenge, and which we address by using a parabolic mirror for collecting the fluorescence, we expect the method to also be applicable when using lenses with moderate numerical apertures.
arXiv:2607.06363v1 Announce Type: new Abstract: We investigate the effect of a high concentration (32.5 g.L$^{-1}$) of sodium chloride (NaCl) on TTAB (tetradecyltrimethylammonium bromide) vertical soap films also called foam films, pulled out of a bath under controlled humidity conditions. We observe that the film lifetime increases with relative humidity, both in the presence and absence of salt. At any given humidity, the presence of NaCl systematically enhances film stability. Our film thickness measurements show that the thinning dynamics with or without salt are nearly identical down to 100 nm. Down to that thickness, the effect of evaporation can be rationalised by a constant evaporation rate, which becomes non-negligible compared to the drainage rate at film thicknesses below 400 nm. The main effect of salt is the stabilisation of a Newton black film at a thickness of approximately 5~nm, whereas in the absence of salt, the film ruptures upon reaching a critical thickness of about 10 nm.
arXiv:2607.05720v1 Announce Type: new Abstract: Resilient-by-design smart grid control demands frameworks capable of maintaining stability under physical disturbances and communication failures, without reliance on centralized coordination. While Centralized Training Decentralized Execution (CTDE) enables a learning-based control paradigm at the grid edge, individually trained models fail to generalize across unseen fault contingencies and fall short of fully decentralized deployment. Federated learning (FL) restores generalization through collaborative training; however, standard aggregation strategies remain agnostic to the physical heterogeneity of synchronous generators. This work proposes Inertia-Informed Weighted FedAvg (IIWFedAvg), a physics-informed aggregation strategy that embeds generator inertia directly into global model fusion for transient stability control in transmission networks. The proposed framework further integrates interpretable Chebyshev Kolmogorov-Arnold Network (ChebyKAN)-based controllers, augmented with Rate-of-Change-of-Frequency (RoCoF) features to enhance dynamic response awareness. Evaluated on the IEEE 39-bus benchmark under full decentralized deployment, IIWFedAvg achieves a 75% generalization success rate across unseen fault contingencies. It also surpasses the centralized baseline in two out of three stabilized faults, while delivering a 3x improvement in stabilization speed at zero centralized coordination overhead.
arXiv:2603.20182v4 Announce Type: replace Abstract: Although robot-to-robot (R2R) communication improves indoor scene understanding beyond what a single robot can achieve, R2R alone cannot overcome partial observability without substantial exploration overhead or scaling team size. In contrast, many indoor environments already include low-cost Internet of Things (IoT) sensors (e.g., cameras) that provide persistent, building-wide context beyond onboard perception. We therefore introduce IndoorR2X, a benchmark and simulation framework for Large Language Model (LLM)-driven multi-robot task planning with Robot-to-Everything (R2X) perception and communication in indoor environments. IndoorR2X integrates observations from mobile robots and static IoT devices to construct a global semantic state that supports scalable scene understanding, reduces redundant exploration, and enables high-level coordination through LLM-based planning. IndoorR2X provides configurable simulation environments, sensor layouts, robot teams, and task suites to systematically evaluate semantic-level coordination strategies. Extensive experiments across diverse settings demonstrate that IoT-augmented world modeling improves multi-robot efficiency and reliability, and we highlight key insights and failure modes for advancing LLM-based collaboration between robot teams and indoor IoT sensors. Project page: https://fandulu.github.io/IndoorR2X_project_page/.
arXiv:2607.05954v1 Announce Type: cross Abstract: The Ramsey community number $r_k$ is the smallest size at which a network is better described by communities than by none, under a Bayesian detection rule. On the diamond hierarchical lattice we show that $r_k$ is an exact renormalization-group crossing: the block-model sufficient statistics obey a linear map with eigenvalues $\{bs,b\}$, the degree-corrected evidence density flows to $\ln K$ at a community fixed point, and $r_k$ is the generation at which the running evidence clears the detection threshold. Degree correction advances detection by two generations. We derive $r_k(b,s;q)$ in closed form for the whole family. Finally, placing on the lattice the Reichardt--Bornholdt community Hamiltonian -- whose ground state is the partition itself -- we find an exact community-ordered phase: below the ferromagnetic critical temperature the two hubs lock into opposite communities for any resolution $\gamma>0$, a staggered order that persists as $n\to\infty$. Allowing each nested sub-community its own label, the optimal partition is a hierarchy of $q_{\rm opt}\sim\sqrt{n}$ communities, so the number of Potts states that best describes the network grows with the network. This hierarchy orders thermally level by level, through a cascade of first-order transitions whose temperatures fall as $1/\ln q$, so every stable level persists as $n\to\infty$: the emergent partition is detectable, optimal, and thermodynamically ordered.
arXiv:2606.28757v2 Announce Type: replace Abstract: Generative world models hold immense promise as scalable simulators for autonomous systems, particularly for synthesizing rare but safety-critical multi-agent interactions, such as vehicle collisions. However, current evaluation paradigms index heavily on visual fidelity and semantic alignment, leaving a critical blind spot: they cannot reliably quantify whether generated dynamics actually obey the fundamental physical laws required for reliable simulation. Assessing this physical plausibility is inherently difficult due to a lack of physical metrics and the challenge of extracting metric-scale kinematics from uncalibrated video rollouts. To bridge this gap, we introduce CrashTwin, a physics-grounded evaluation framework designed to stress-test the physical trustworthiness of world models. CrashTwin couples a diverse dataset of multi-agent collision scenarios, comprising 25K controllable synthetic and 12K in-the-wild real-world collision sequences with a novel calibration-free reconstruction pipeline, enabling the recovery of 3D physical attributes directly from world model rollouts. We propose a diagnostic suite that systematically evaluates three dimensions: spatio-temporal consistency, momentum and kinetic energy conservation, and world-dynamics integrity. Extensive benchmarking of state-of-the-art models reveals a crucial insight: high perceptual quality frequently masks severe physical violations during complex interactions. By quantitatively exposing these failure modes, CrashTwin provides a vital diagnostic tool for developing physically grounded world models capable of reliable real-world simulation.
arXiv:2604.16157v2 Announce Type: replace-cross Abstract: Atmospheric neutrinos provide a unique avenue to probe theories beyond the Standard Model (BSM) over a wide range of energies and path lengths. The theory of nonstandard interactions (NSI) of neutrinos is one of the important BSM scenarios, which can modify flavor oscillations of atmospheric neutrinos traveling through the Earth. In this work, we use a high-purity $\nu_{\mu}$ charged-current (CC) sample of atmospheric neutrinos from IceCube DeepCore with a livetime of 7.5 years to search for the NSI parameters $\varepsilon_{e\mu}$, $\varepsilon_{e\tau}$, and $\varepsilon_{ee}-\varepsilon_{\mu\mu}$. The $\nu_{\mu}$ CC events mainly come from the $\nu_{\mu}$ survival channel having no significant dependence on $\delta_{CP}$. Therefore, the constraints on $\varepsilon_{e\mu}$ and $\varepsilon_{e\tau}$ obtained using this $\nu_{\mu}$ CC sample are expected to be free from the $\delta_{CP}$-degeneracy. The data sample is found to be in agreement with the standard neutrino interactions. Therefore, we place bounds on these NSI parameters that are consistent with and comparable to existing experimental constraints. These $\delta_{CP}$-free constraints from IceCube DeepCore are complementary to those from the long-baseline neutrino oscillation experiments, where the appearance channel depends on $\delta_{CP}$.
arXiv:2604.22584v2 Announce Type: replace-cross Abstract: For a digraph $D$ and some $X \subseteq V(D)$, the inversion of $X$ is the operation of flipping all arcs both of whose endvertices are in $X$. We initiate the study of establishing arc-connectivity properties by applying inversions of bounded or fixed size. For fixed-size inversions, the feasibility problem is interesting. For all integers $p \geq 2$ and $k \geq 1$, we give a characterization of the digraphs that can be made $k$-arc-strong by applying inversions of size exactly $p$, provided they are sufficiently large. For bounded-size inversions, the feasibility problem is easy, so we focus on minimising the number of inversions. We prove that for all integers $p\geq 3$ and $k \geq 1$ and any $\epsilon>0$, there exists a polynomial-time $(4k-2+\epsilon)$-approximation algorithm for computing the minimum number of inversions of size at most $p$ that make a given digraph $k$-arc-strong. This is in stark contrast to other results on inversion optimization problems. On the other hand, we show that for any $p\geq 3$ and $k \geq 1$ the problem is NP-hard, and, moreover, APX-hard. As a result on parameterized complexity, we show that for any $k \geq 2$, it is $W[1]$-hard with respect to $p$ to decide whether a given digraph can be made $k$-arc-strong by applying a single inversion of size at most $p$. We also prove that for a given multidigraph, it is $W[1]$-hard with respect to $\ell$ to decide whether it can be made 2-arc-strong by applying $\ell$ inversions of size 2.
arXiv:2506.22961v2 Announce Type: replace-cross Abstract: The MPC-in-the-head technique (Ishai et al., STOC 2007) is a celebrated method to build zero-knowledge protocols with desirable theoretical properties and high practical efficiency. This technique has generated a large body of research and has influenced the design of real-world post-quantum cryptographic signatures. In this work, we present a generalization of the MPC-in-the-head paradigm to the quantum setting, where the MPC is running a quantum computation. As an application of our framework, we propose a new approach to build zero-knowledge protocols where security holds even against a verifier that can obtain a superposition of transcripts. This notion was pioneered by Damgard et al., who built a zero-knowledge protocol for NP (in the common reference string model) secure against superposition attacks, by relying on perfectly hiding and unconditionally binding dual-mode commitments. Unfortunately, no such commitments are known from standard cryptographic assumptions. In this work we revisit this problem, and present two new three-round protocols in the common reference string model: (i) A zero-knowledge argument for NP, whose security reduces to the standard learning with errors (LWE) problem. (ii) A zero-knowledge argument for QMA from the same assumption.
arXiv:2607.05680v1 Announce Type: new Abstract: AI systems are increasingly used to provide legal advice, raising questions about whether laypeople accept guidance from algorithms--especially when that advice is legally correct but socially controversial. We report a preregistered survey experiment with 3,348 adults in mainland China examining how people evaluate identical legal advice when it is attributed either to an AI system or to a human lawyer, and when it is accompanied by reasoning or not. Contrary to expectations of algorithm aversion, attribution to an AI system has no net effect on perceived reasonableness. However, mediation analyses reveal opposing psychological pathways underlying this null result. AI-attributed advice is perceived as more objective, which increases perceived reasonableness, but also as less comprehensive and less attentive to special circumstances, which decreases perceived reasonableness. By contrast, providing legal reasoning substantially increases perceived reasonableness regardless of source, largely by enhancing perceptions of objectivity. Qualitative responses corroborate this tension between objectivity and contextual sensitivity in evaluations of legal advice. Together, these findings suggest that public responses to AI legal advisors are shaped not by rigid attitudes toward automation, but by the balancing of competing normative expectations. The results have implications for theories of algorithm aversion and the design of AI recommendation systems in normatively salient domains.
arXiv:2607.05685v1 Announce Type: new Abstract: Large language models are increasingly used as private, always-available conversational systems, but little is known about how people with depressive symptoms use them. Building on CSCW work on disclosure and peer support, we examine ChatGPT as an emerging informal support infrastructure: private, persistent, responsive, and available outside ordinary hours. We analyze 187,093 ChatGPT conversations from 766 participants who completed the PHQ-8, comparing those below the moderate-symptom threshold (score of 10) with those at or above it. Higher-PHQ participants used ChatGPT more for mental-health, interpersonal, loneliness, self-focused, and support-seeking conversations, with pronounced late-night and recurring month-level patterns. Their language contained more first-person singular pronouns and absolutist terms. They more often engaged ChatGPT in high-disclosure contexts, but professional redirection was not higher. Language-based prediction was modest and insufficient for screening (AUROC 0.591). We argue these histories should not be treated as clinical screening data but as evidence LLMs are increasingly used as informal support infrastructure.
arXiv:2607.05718v1 Announce Type: new Abstract: Learning continuous-time representations of dynamical systems from observation data has emerged as a cornerstone of data-driven control and scientific machine learning. However, existing neural differential equations either treat external control inputs heuristically without providing strict structural guarantees, or enforce stability properties under the restrictive assumption of constant or vanishing inputs. This paper proposes the Input-Contraction Neural Differential Model (ICNDM), a novel deep learning framework that seamlessly incorporates time-varying control inputs while ensuring incremental exponential convergence via input-dependent contraction regularization. By leveraging an embedded input encoder and a parameterized metric network, the proposed architecture learns both the non-autonomous neural vector fields and a generalized Riemannian contraction metric simultaneously. We derive sufficient conditions for input-dependent contraction and formally establish an input-to-state contraction property under bounded external excitations. Extensive numerical evaluations on highly nonlinear chaotic oscillators and experimental data from a Permanent Magnet Synchronous Motor (PMSM) drive system demonstrate that ICNDM yields substantial reductions in long-horizon rollout errors and exhibits superior structural robustness against input perturbations compared with state-of-the-art neural differential benchmarks.
arXiv:2607.05910v1 Announce Type: new Abstract: Image guardrails are typically trained and evaluated under a fixed safety policy, implicitly treating safety as an intrinsic property of an image. Real deployments are different: the same image may be allowed in one product, restricted in another, and newly disallowed when a policy boundary changes. We study policy-adaptive image guardrailing, where a model must decide whether an image violates the currently supplied policy and generalize to held-out policy definitions. We introduce PolicyShiftBench, a comprehensive benchmark with 2,000 policy-discriminative instances over 265 images, where each image is paired with 7.55 policy-conditioned prompts on average to test whether models adapt to the active policy rather than relying on image-level safety priors. We then propose PolicyShiftGuard, a compact policy-conditioned guardrail trained with a two-stage training recipe that combines Randomized Policy SFT (RP-SFT) with Boundary-Pair Policy Adaptation (BP-Adapt). BP-Adapt trains matched prompts for the same image and risk category using standard label supervision and a pairwise comparison loss that separates blocking policies from passing policies. Experiments show that existing VLMs and specialized guardrails remain brittle under policy shifts, while PolicyShiftGuard substantially improves policy-sensitive performance. The 7B model achieves SOTA performance of 76.9 Avg. F1 and 72.1 Avg. PSS on PolicyShiftBench, transfers well to UnSafeBench and SafeEditBench, and improves the latency-performance trade-off with a concise output format. Ablations confirm that matched pass/block boundary pairs are essential for stable policy adaptation.
arXiv:2604.23931v2 Announce Type: replace-cross Abstract: Variational quantum circuits (VQCs) are a leading approach to quantum machine learning on near-term devices, yet it remains unclear which circuit architecture yields the best accuracy-parameter trade-off on classical tabular data. We present a systematic empirical comparison of four VQC families -- multi-layer fully-connected (FC-VQC), residual (ResNet-VQC), hybrid quantum-classical transformer (QT), and fully quantum transformer (FQT) -- across five regression and classification benchmarks. Our key findings are: \textbf{(i)}~FC-VQCs achieve 90-96\% of the $R^2$ of attention-based VQCs while using 40-50\% fewer parameters, and consistently outperform equal-capacity MLPs (mean $R^2{=}0.829$ vs.\ MLP$_{720}$'s $0.753$ on Boston Housing, 3-seed average); \textbf{(ii)}~FC-VQC's Type~4 inter-block connectivity provides partial cross-token mixing that approximates the role of attention -- explicit quantum self-attention yields only marginal gains on most datasets while significantly increasing parameter count; \textbf{(iii)}~expressibility saturates at circuit depth~${\approx}\,3$, explaining why shallow VQCs already cover the Hilbert space effectively; \textbf{(iv)}~LayerNorm on the fully quantum transformer improves classification accuracy, suggesting normalization is important when all operations are quantum; \textbf{(v)}~in our noise study on Boston Housing, FQT degrades gracefully under depolarizing noise while QT collapses. All results are validated across three random seeds. These findings provide practical architectural guidance for deploying VQCs on near-term quantum hardware.
arXiv:2604.27694v2 Announce Type: replace-cross Abstract: Renewed attention to the identity of Bitcoin's pseudonymous creator has revived an old worry: that the roughly 1.148 million BTC mined by Satoshi and never moved represent a major tail risk for bitcoin. This paper argues that the worry is overstated. The mechanical downside of selling the position is bounded well below the feared collapse, and the outcomes most consistent with sixteen years of observed behavior are not bearish for bitcoin's effective supply. We analyze the position in two ways. First, we model the case of a purely financial holder. Multiple sale scenarios, checked against both a square-root-law estimate and the historical record of large sales, suggest that bitcoin's current market liquidity could absorb a patient multi-year sale with a cumulative price impact centered around 10 to 13 percent relative to a no-sale case. The same arithmetic also links the downside from a surprise sale to the upside from a confirmed burn: both are bounded by the same effective-supply adjustment, so the doom case and the burn-rally case cannot both be large. Second, we consider the preferences implied by the sixteen-year record. Ideological restraint, privacy, already having enough, and preserving the myth all point toward further dormancy, permanent loss of access, or a deliberate burn. A sale or an act of sabotage remains possible, but the record supports it less strongly. Under both approaches, the mechanical bear case is bounded, and the likeliest outcomes are neutral to mildly positive for bitcoin's effective supply. The argument does not rule out transient overshoot or leverage-driven amplification; it bounds the durable repricing the coins themselves can cause.
arXiv:2605.12410v2 Announce Type: replace-cross Abstract: We propose and analyze a model-based bootstrap for transition kernels in finite controlled Markov chains (CMCs) with possibly nonstationary or history-dependent control policies, a setting that arises naturally in offline reinforcement learning (RL) when the behavior policy generating the data is unknown. We establish distributional consistency of the bootstrap transition estimator in both a single long-chain regime and the episodic offline RL regime. The key technical tools are a novel bootstrap law of large numbers (LLN) for the visitation counts and a novel use of the martingale central limit theorem (CLT) for the bootstrap transition increments. We extend bootstrap distributional consistency to the downstream targets of offline policy evaluation (OPE) and optimal policy recovery (OPR) via the delta method by verifying Hadamard differentiability of the Bellman operators, yielding asymptotically valid confidence intervals for value and $Q$-functions. Experiments on the RiverSwim problem show that the proposed bootstrap confidence intervals (CIs), especially the percentile CIs, outperform the episodic bootstrap and plug-in CLT CIs, and are often close to nominal ($50\%$, $90\%$, $95\%$) coverage, while the baselines are poorly calibrated at small sample sizes and short episode lengths.
arXiv:2607.05694v1 Announce Type: cross Abstract: Logit-based watermarking is a widely used mechanism for identifying LLM generated content, yet its effectiveness is governed by a fundamental trade-off between detectability and semantic distortion. Existing analyses provide limited guidance for principled hyperparameter selection, leaving practical deployments reliant on heuristic tuning. In this work, we develop a power-calibrated statistical framework that establishes explicit quantitative relationships between watermark hyperparameters, detection power, and distortion. This characterization transforms watermark design into a guided optimization problem. Building on these results, we derive practical parameter selection procedures that achieve optimal tradeoffs under constraints. Extensive experiments across multiple language models and datasets validate the theory and demonstrate that the proposed framework consistently identifies Pareto-optimal points.
arXiv:2607.05412v1 Announce Type: new Abstract: STEM education faces challenges in personalization and interdisciplinary integration. AI technology has brought new possibilities, but the mechanisms by which AI reshapes the STEM education ecosystem require systematic investigation. This study employs bibliometric methods to analyze 242 publications from 2015-2025, constructing knowledge maps to reveal the evolutionary trajectory. The findings show that the field has transformed from intelligent tutoring systems to inquiry-based learning and computational thinking cultivation driven by LLMs. AI's key contribution lies in providing intelligent scaffolding that lowers the threshold for understanding knowledge. In this sense, AI is a core driving force promoting its shift from knowledge transmission to capability development.
arXiv:2512.19458v2 Announce Type: replace Abstract: Large Language Models (LLMs) are increasingly embedded in agentic frameworks for scientific discovery. First-principles materials computation imposes a demanding standard for autonomy: successful execution depends on internally consistent inputs, supervision of long-running calculations, and verified outputs. Here we present VASP Agent, a coding-agent-centered system that combines reusable domain skills, deterministic tools, workspace-state inspection, runtime evidence, and scientific guardrails to execute multi-step VASP calculations. The system is evaluated across multiple tasks including structural relaxation, bandgap calculation, equilibrium lattice constant determination, and CO/Pt(111) adsorption. VASP Agent completes all evaluated cases, and its computed numerical results are compared with those obtained using pymatgen and other agentic tools. When large deviations occur, the calculation parameters produced by VASP Agent are more appropriate than those produced by LLM-based workflows. Failure analysis shows that errors that terminate fixed pipelines can be diagnosed and recovered under agentic control.
arXiv:2607.04687v2 Announce Type: replace-cross Abstract: We consider the set of matchings of a graph and a local change operation, called a flip, between them. In the combinatorial setting, the base graphs are either complete graphs or complete bipartite graphs, and in the geometric setting, the graphs are embedded on point sets in the plane, with the requirement that edges must be drawn as straight lines and must not cross. For base graphs with an even number of vertices, we consider perfect matchings, i.e., all vertices are matched, and for base graphs with an odd number of vertices, we consider almost-perfect matchings, i.e., all but one vertex of the graph are matched. A 2-flip between two perfect matchings exchanges two edges, and a 1-flip between two almost-perfect matchings exchanges one edge. The corresponding flip graph has the set of perfect or almost-perfect matchings as vertices, with pairs of them connected by an edge if they differ in a 2-flip or 1-flip, respectively. In this work, we provide a comprehensive picture of Hamiltonicity properties of these flip graphs. We prove that the flip graphs in the combinatorial setting are Hamilton-connected, i.e., they admit a Hamilton path between any two vertices, or, if the flip graphs are bipartite, we prove that they are Hamilton-laceable, i.e., they admit a Hamilton path between any two vertices from different partition classes. In the geometric setting, we prove that any path in them misses exponentially many vertices, in particular, they have no Hamilton paths or cycles. For points in convex position and almost-perfect matchings under 1-flips, we complement this by constructing a cycle in the flip graph that visits almost all vertices.