Papers

Existing evaluation methods for healthcare LLMs primarily assess factual correctness,safety, and fluency, while providing limited insight into whether generated interactions reflect culturally situated healthcare reasoning. This limitation is particularly important in maternal health, where care decisions are shaped by social and relational norms. We introduce MH-INDIC, a culturally grounded evaluation framework for maternal-health interactions in urban and semi-urban North Indian contexts that operationalises cultural behaviour through ten dimensions of maternal-health reasoning. Using a 26-item survey administered to 102 pregnant and postpartum women from urban and semi-urban North India, we evaluate ten LLMs. We distinguish population level cultural alignment from profile-level behavioural variation. Although several models approximate the human population-level distribution, all evaluated systems exhibit substantially lower variation across demographic and household profiles than the human cohort, revealing a gap between aggregate alignment and profile-conditioned sensitivity. As a downstream application of MH-INDIC, we use the strongest-aligned proprietary and open-source models to generate culturally conditioned maternal-health dialogues under zero-shot, self-conditioned, and human-grounded prompting. Human-grounded conditioning produces stronger profile alignment and dialogue quality ratings, suggesting that measured cultural profiles can improve the cultural grounding of generated interactions

Recent advances in large language model (LLM) pretraining highlight the role of high-quality training data in improving performance. While model-based filtering has proven effective in selecting high-quality subsets from web-scale corpora, especially for high-resource languages, low-resource languages face challenges due to limited availability of annotated data. This work explores extending quality filtering to over 100 languages by proposing a multilingual adaptation approach that converts an existing English quality classifier into a multilingual variant. Our approach proposes training a small multi-layer perceptron on top of Transformer encoder-only model embeddings, using multilingual text as input and scores obtained from English classifiers applied to machine-translated text as labels. Our 1B, 3B and 8B scale experiments show that our approach maintains the downstream LLM benchmark performance of existing multilingual model-based filtering baselines, without harming regional and cultural knowledge benchmarks. To further evaluate cross-lingual generalization, we compare classifier scores of high-quality synthetic data and web samples, and the correlation of classifier scores with LLM-based ones, revealing that the classifier can learn the scoring criteria of its original English variant, even for languages not included in its training data.

To test conditional independence of $X$ and $Y$ given a text or an image $Z$, one conditions on an embedding $\psi(Z)$ in place of $Z$. The embedded test is valid if $Z$ is independent of $X$ or of $Y$ given $\psi(Z)$, which cannot be confirmed from data, and when this fails, the rejection probability under the null hypothesis can tend to one. We study this failure, and show that focusing on a specific form of dependence relaxes what the embedding must retain. For a residual correlation test inspired by the Generalised Covariance Measure, validity only requires that the parts of $\mathbb{E}[X \mid Z]$ and $\mathbb{E}[Y \mid Z]$ missed by $\mathbb{E}[X \mid \psi(Z)]$ and $\mathbb{E}[Y \mid \psi(Z)]$ are uncorrelated. Otherwise, we treat the discarded information as an omitted variable. Under the null hypothesis, the bias equals the absolute correlation of the missed parts times the geometric mean of two partial $R^2$ values. This identity yields a robust test valid under a declared tolerance for the geometric mean, which, like a sensitivity parameter, is not identified from the data. On synthetic data and text embeddings, the robust test holds its level approximately. On text generated by a language model, under an exact null hypothesis, every embedding, even the generator's own states, biases the embedded test.

End-to-end autonomous driving demands trajectory planners that are both highly accurate and cheap enough for edge deployment. State-of-the-art artificial neural network (ANN) planners meet the accuracy requirement at the cost of heavy dense computation, while spiking neural networks (SNNs)—though promising orders-of-magnitude energy savings through sparse, event-driven arithmetic—still lag far behind in planning accuracy. We present SDPAD, a fully spike-driven end-to-end planning pipeline that closes this gap. SDPAD converts a pre-trained ANN perception stack into integer-spike form via quantized ANN2SNN conversion, lifts multi-view images into the bird's-eye-view (BEV) space with a spike-driven-max (SDM) depth distribution (Spike-3D-Lift), and plans through the Spike-QFormer, a spiking query transformer in which ego, agent, and map queries distilled from the BEV scene are fused by learnable waypoint queries via cross-attention, followed by deformable spike-cross-attention refinement. Every operation is gated by integer spikes and inference is a single feed-forward pass without temporal simulation loops. On the nuScenes open-loop benchmark, SDPAD achieves an average $L_2$ error of 0.40\,m and a collision rate of 0.12%, on par with strong ANN planners while consuming 69.9\,mJ—less than 2% of recent ANN baselines. In closed-loop evaluation on the NAVSIM navtest split, SDPAD reaches 86.3 PDMS, surpassing the previous SNN planner SAD by 4.3 points and matching mainstream ANN planners at a fraction of their energy. To our knowledge, SDPAD is the first fully spike-driven planner evaluated in end-to-end autonomous driving, demonstrating that SNNs can rival dense ANNs in complex driving tasks.

Spectroscopy with undetected photons allows the spectral response of a sample at long wavelengths to be measured through interference of a shorter-wavelength signal field, without directly detecting the light that probes the sample. Previous demonstrations of undetected-photon spectroscopy in the molecular fingerprint region have reached approximately 10.5 microns, leaving longer wavelengths largely unexplored. Here, we extend this approach to 20 microns wavelength, demonstrating infrared undetected-photon spectroscopy using a BaGa2GeSe6 (BGGSe)-based nonlinear interferometer while only the near-infrared (NIR) signal is detected using a silicon detector. Rotation of the BGGSe crystal provides broad wavelength tuning, and Fourier-transform analysis of the signal interferograms yields the corresponding idler spectra and sample transmittance. Broadband absorption spectroscopy is demonstrated using Kapton across a 7.8-20 microns wavelength range, with the measured absorption agreeing with independent infrared measurements. The same system is further employed to detect sulfur hexafluoride (SF6) gas via spectroscopy near 10.55 microns. These results substantially extend the accessible spectral range of spectroscopy with undetected photons while retaining silicon-based NIR detection.

We study the energy-subcritical and energy-critical complex Ginzburg-Landau equations on three-dimensional Riemannian manifolds. We establish the asymptotic behavior of solutions via a purely energy-based approach and subsequently prove a fast/slow dichotomy of the solutions. Moreover, for the slow solution, we show that the solution converges to an explicit asymptotic profile. In particular, we obtain a precise description of both the amplitude and the phase of the leading-order dynamics. As a by-product, the complex Ginzburg-Landau equation approximation yields global weak solutions to the energy-subcritical and energy-critical defocusing nonlinear Schrödinger equations on compact three-dimensional Riemannian manifolds. We also investigate the long time behavior of the solutions to the energy-subcritical and energy-critical complex Ginzburg-Landau equations on a class of complete noncompact three-dimensional Riemannian manifolds and prove decay of global solutions under suitable geometric assumptions.

Accurate trajectory forecasting and well-defined predictive uncertainty are crucial for reliable, safety-critical applications such as autonomous driving. Most trajectory prediction approaches provide point estimates only, while uncertainty-aware approaches typically quantify uncertainty only in the trajectory space. In physics-aware approaches, uncertainty in the predicted motion variables should be explicitly modeled and propagated through the vehicle dynamics. Otherwise, the resulting trajectory-space uncertainty may not fully reflect the variability introduced by the underlying motion prediction. Therefore, in this work, uncertainty-aware extensions of X-TRACK (X-TRACK-DE and X-TRACK-MCD), a physics-aware trajectory prediction framework, are proposed. The proposed framework predicts future vehicle motion variables and models both aleatoric and epistemic uncertainties by propagating motion space uncertainty to trajectory space. Additionally, conformal prediction is applied to the trajectory space predictive covariance to construct uncertainty regions targeting a desired marginal coverage level. Evaluation on the highD dataset shows that X-TRACK-DE improves trajectory prediction accuracy over the deterministic baseline, while both uncertainty-aware variants provide predictive uncertainty that can be conformally calibrated to the desired marginal coverage level.

To adapt billion-parameter Vision Transformers for long-video understanding, recent methods freeze the backbone and train lightweight convolutional modules. While effective for parameter-efficient training, existing adapters do not reduce inference-time computation, leaving scalability with respect to video length largely unaddressed. Token selection can reduce attention cost by pruning redundant tokens, but it breaks the spatial grid structure required by convolutional adapters. This forces an expensive dense reconstruction, nullifying much of the potential speedup. We address this by introducing native sparse 2D convolutions, a primitive that allows these adapters, for the first time, to operate directly and efficiently on dynamically pruned token sets. We integrate this primitive into SV-TAD, an adapter framework for temporal action detection, reducing VideoMAEv2-L computation by up to 64% and achieving 2.2x faster inference, while maintaining state-of-the-art accuracy on THUMOS-14 and ActivityNet-1.3. When scaled to InternVideoNext-L, our approach surpasses the previous state of the art at roughly half its computational cost. Moreover, the sparse formulation naturally supports auxiliary task tokens, which improves fine-grained assembly detection on ATTACH.

Historical pull requests record the design decisions, compatibility constraints, and implementation patterns behind a codebase's current state. Experience relevant to a new task can span related changes whose descriptions emphasize different concerns. We introduce Chronos, a test-time framework that makes this connected history available to large language model (LLM)-based code agents. Chronos distills merged pull requests into structured experience cards and connects them through a typed graph of code-level, developer-intent, and organizational relations. Semantic search identifies entry cards, and weighted multi-hop expansion retrieves connected changes for selective reading. The same memory guides candidate generation and patch selection: a patch-focused change agent and a validation-strategy agent each develop a patch, and an evolution steward consults history to select between them. On SWE-Bench Verified, the full workflow improves SWE-Agent across all six evaluated LLM backbones, raising the mean resolution rate from 69.2% to 72.9% and reaching 79.8% with MiniMax M2.5. With the same backbone, it raises resolution rates from 48.3% to 51.7% on SWE-Bench Pro and from 41.0% to 43.5% on FEA-Bench Lite. Both experience-guided single-agent variants also outperform the base agent. In a human evaluation on 100 tasks with ten cards retrieved per task, graph-grounded retrieval increases the mean number of useful cards from 1.24 to 2.87 over flat semantic retrieval. These results demonstrate the value of PR relations for retrieving useful repository experience and of the evaluated workflows for applying that experience during patch generation and selection.

Autonomous vehicles often suffer from limited perception due to occlusions, blind spots, limited sensor range, and the complex nature of surrounding environments. Multi-agent collaborative perception (CP) addresses these challenges by allowing vehicles to share sensory information and reconstruct the scene cooperatively. However, camera-only perception remains fundamentally limited by the uncertainty of distance-dependent monocular depth estimation. We propose CoCam4D, a Bayesian framework for collaborative perception that explicitly models geometric uncertainty. It uses a VGGT-based feedforward network to generate 3D Gaussian scene representations with associated uncertainty estimates, enabling multiple vehicles or agents to efficiently combine their observations. By sharing compact Gaussian primitives, reliable observations from one agent can reduce the depth uncertainty of another without requiring LiDAR sensors. To support real-world deployment, we introduce Dynamic Object Primitives (DOPs), a compact 35-byte representation designed for efficient C-V2X communication. Extensive experiments show that our proposed method consistently outperforms recent vision-only methods, achieving improvements of 11.48% on OPV2V+ and 10.62% on DAIR-V2X-C, demonstrating the potential of geometrically grounded collaborative perception for LiDAR-free autonomous driving.

For $n\ge1$, we study representations of a simplicial group $G$ in the full $\infty$-subcategory of simplicial vector spaces whose normalized homology is concentrated in degrees $0,\ldots,n-1$, retaining its full mapping spaces. The resulting functor category depends only on the $n$-truncated homotopy type of $\mathbf{B}G$ and admits projectively bifibrant strict diagram and simplicial group-algebra module models. If restriction along a specified map $f:G\to H$ is an equivalence, then $\pi_0(f)$ is a group isomorphism; the proof uses ordinary extension and restriction of scalars. Yet for every $n\ge2$ and prime $q$ invertible in $k$, the nontrivial source $K(\mathbb Z/q,n-1)$ has the same $n$-representation category with vertex evaluation as the trivial group. The group-algebra augmentation is a weak equivalence, so the constant functor is also an equivalence for untruncated representations. Thus $n$-representations depend only on the $n$-truncation of $\mathbf{B}G$, but need not determine that truncation, even together with vertex evaluation.

Sparse Mixture-of-Experts models scale parameter capacity efficiently while maintaining a fixed compute budget per token. However, traditional training paradigms enforce a static choice of top-$k$ experts, which converts a continuous routing distribution into a rigid step function. This constraint introduces a brittle boundary where highly competitive experts are arbitrarily separated into full-supervision and zero-feedback zones based on minor score fluctuations. To address this issue, we propose Elastic Expert Routing, which stochastically samples the active expert budget from a localized discrete distribution centered at $k$. Over multiple training iterations, this mechanism softens the sharp threshold into a gradual probability distribution. Because the sampling neighborhood remains symmetric, this approach matches the expected computational cost of deterministic training, while preserving the inference budget. During supervised fine-tuning, elastic routing improves downstream macro-averages on OLMoE-1B-7B and Qwen3-30B-A3B by $+0.84$ and $+2.02$ points, respectively. In addition, in from-scratch pretraining, it reduces validation LM loss by up to $0.04$ and outperforms the static top-$k$ baseline by $1.6$ points on average across downstream tasks. Comparisons on Qwen3 fine-tuning and from-scratch pretraining further show higher average scores than Noisy top-$k$, expert FFN dropout, and DefaultMoE.

Linearized permutation polynomials over finite fields have important applications in coding theory, cryptography, and combinatorics. In this paper, we give an explicit description of the cycle structure of a general linearized permutation polynomial in terms of the Jordan blocks of its Dickson matrix, thereby providing a Dickson-matrix formulation of a question raised by Mullen and Vaughan in 1988. Building on the cycle structure, we establish a matrix-theoretic framework to characterize their parity. For finite fields of characteristic 2, we determine all odd linearized permutation polynomials explicitly. For fields of odd characteristic, we derive parity criteria in terms of the eigenvalue structure of the associated Dickson matrix. Based on the parity criteria, we then give an algorithm for deciding the parity of a given linearized permutation polynomial.

The memory capabilities of Large Language Models (LLMs) have garnered increasing attention recently. Despite great success achieved, existing retrieval-based memory approaches typically overlook the differences between memories and employ a unified strategy to process all memories, leading to suboptimal performance. Thus, an intuitive question arises: can we categorize memory into different types and select appropriate strategies? However, given the topic-rich, scenario-complex, and boundary-blurred nature of memory scenarios, achieving precise classification of memories is not easy. To address this challenge, we propose a memory multi-class dataset in this paper, termed TriMEM, which provides precise annotations for memory types across diverse scenarios. Building upon this foundation, we propose a novel memory framework, named MemoType, which can adaptively recognize each memory and query type with the learned router model. With the memory and query routing, MemoType can retrieve the memory with corresponding query types and design tailored retrieval strategies, thereby enhancing the retrieval performance. Moreover, we theoretically prove that any single retrieval strategy is subject to a fundamental upper bound on its expected retrieval precision in multi-class corpora, leading to systematic precision degradation. Extensive experiments on three datasets demonstrate that MemoType consistently outperforms existing methods, achieving up to 16.18% improvement in Recall@1.

Adapting a pre-trained 3D Gaussian Splatting (3DGS) road scene to a new time of day requires learning appearance changes from a few anchor images while preserving consistent, real-time rendering. We propose PAM-ToD, a lightweight plug-in that learns color corrections while keeping the pre-trained 3DGS parameters fixed. PAM-ToD scales each Gaussian's existing color to model illumination changes and uses an additive term for additional brightness, such as when street lamps turn on at night. Under a simplified image formation model, unchanged surface albedo can be eliminated from the relation between source and target appearances, allowing us to learn these corrections without separately estimating albedo and illumination. The model corrects colors across the scene while allowing the corrections to vary by location and by Gaussian. To guide learning from a few anchor images, it discourages abrupt spatial changes in these corrections. We also introduce CARLA-ToD, a benchmark with matching geometry, camera poses, and moving-object trajectories across three times of day. A few target-time anchor images are used to train each plug-in, while separate views are used for evaluation. Across the static and dynamic settings, PAM-ToD achieves higher PSNR and lower LPIPS than the baselines, even when the anchor images come from a single synchronized capture across multiple cameras.

Spatio-temporal tracking data has opened new possibilities for detecting complex tactical patterns in football, yet modeling the interactive movements of multiple players remains challenging. This paper proposes a framework combining graph neural networks (GNNs) with a sequential model to classify match phases on a second-by-second basis across a seven-class taxonomy. Match phase classification is tactically meaningful, and the availability of rule-based labels across 203 matches makes it a suitable testbed for a systematic comparison of adjacency constructions and message-passing layers, a question that has received limited attention in existing research. Our selected GNN-LSTM model outperforms all aggregated-feature baselines, including XGBoost and a Long Short-Term Memory (LSTM) network, as the strongest baseline scores 4.6% lower in macro F1. Graph representations using a domain-informed Delaunay triangulation that approximates passing lanes, paired with a custom Spatial Edge-Augmented Convolution (SEAConv) layer that injects edge attributes directly into messages, achieve the best performance by capturing spatial dependencies while limiting uninformative messages from redundant edges. Integrated Gradients attributions indicate the model uses spatial player configurations, particularly horizontal positioning, in combination with possession and ball-status indicators. This work offers an automated solution for fine-grained tactical analysis, reducing the need for manual tagging and providing deeper insight into dynamic team behavior.

In this paper, we argue that Looped Transformers need their own residual connections to prevent performance degradation as the number of iterations grows. We first introduce an interesting phenomenon we observed in Looped Transformers, called latent overthinking: noisy state updates overwrite correct intermediate deductions and even undo completed solutions. This leaves subsequent iterations to recover lost information from an already degraded representation: once an error arises in an earlier loop, often as a result of long-range propagation through the recurrence, later loops find it difficult to correct. In this paper, we introduce InfiLoop, a loop-native residual connection that learns which past computations to retain and how much to accept from each new update. InfiLoop combines content-based weighting with learned temporal decay to maintain a running summary of recurrent states. An exact streaming recurrence keeps its persistent aggregation memory constant as the loop count grows. The resulting adaptive update suppresses unreliable proposals and preserves useful intermediate states. Across extensive reasoning tasks, a 7M-parameter InfiLoop model outperforms existing recursive architectures, reaching 97.9% exact accuracy on Sudoku-Extreme, and 13.6% pass@2 on ARC-AGI-2. Notably, on Sudoku-Extreme, InfiLoop continues to improve with test-time looping beyond 20,000 effective steps, showing that added depth translates directly into stronger reasoning. Our code is available at https://github.com/pixeli99/InfiLoop.

We construct nets of curves in a rectangle $R$ with special properties. The net is formed by two families of curves, "horizontal" and "vertical", each family foliates $R$, and every horizontal curve meets every vertical curve orthogonally. Moreover, every horizontal curve started at the left edge of $R$ arrives to the right edge of $R$ at the same height, and every vertical curve started at the bottom edge of $R$ arrives to the top edge of $R$ at a prescribed in advance longitude. The latter longitude function as a map from bottom to top can be chosen in advance as any small perturbation of the identity. Our motivation for this was to design some elliptic divergence form operators, for instance on one of the domains bounded by a Reifenberg flat curve in the plane, with the scalar coefficient and the corresponding elliptic measure being absolutely continuous with respect to the natural measure on the boundary (an object of a separate paper). The construction of our net of curves in a rectangle can also be interpreted as control problem, has an interest on its own, and, likely, has other applications.

The proximity of the narrow $P_c(4312)$, $P_c(4440)$, and $P_c(4457)$ structures to the $\Sigma_c\bar D^{(*)}$ thresholds motivates their interpretation as hadronic molecules. We treat the $P_c(4312)$ as an $S$-wave $\Sigma_c\bar D$ state with $J^P=1/2^-$ and examine both $J^P=1/2^-$ and $3/2^-$ $\Sigma_c\bar D^*$ assignments for each heavier peak. The selected $S$-wave open-charm transitions to $\Lambda_c\bar D^{(*)}$ are calculated at Born order in the quark-interchange model, including the prior and post rearrangement diagrams and their color, flavor, spin, and spatial matrix elements. Across the input variations considered, the quark-interchange widths robustly favor $\Lambda_c\bar D^*$ over $\Lambda_c\bar D$ for every $J^P=1/2^-$ assignment. At either heavier peak mass, the $\Lambda_c\bar D^*$ width for $J^P=3/2^-$ is suppressed by approximately one order of magnitude relative to $J^P=1/2^-$, providing a potential decay-based discriminator of the spin assignments. The large channel hierarchies arise from spatial overlaps and interference between spin-independent and hyperfine contributions and are not universal heavy-quark-spin-symmetry ratios. We also explore a retarded one-pion-exchange transition at first Born order. It affects $P_c(4312)\to\Lambda_c\bar D^*$ most strongly, while the coherent widths remain sensitive to the cutoff, energy-transfer prescription, short-distance subtraction, and unresolved relative matching sign. The quark-interchange hierarchies are therefore the central predictions, whereas the pion-exchange results quantify matching sensitivity. Measurements of the $\Lambda_c\bar D^*$ modes can directly test these patterns.

Battery state of health (SoH) forecasting is important for battery management, but remains challenging due to nonlinear degradation and heterogeneity across batteries. Existing data-driven approaches primarily use temporal models to learn from numerical battery time series, and higher-level degradation characteristics are often not explicitly represented. These characteristics, however, can provide degradation guidance to support reliable forecasting and make the influence of degradation more interpretable. In this paper, we propose Sera, a semantic representation aggregation framework that complements temporal modelling with degradation semantics. Guided by battery domain expertise, Sera extracts degradation semantics from time series and constructs two complementary representations using rule-based knowledge and LLM-based interpretation. The representations are independently encoded and integrated with the representation learned by temporal models through gated aggregations. Experiments on the mainstream benchmark across multiple prediction horizons and different temporal models show that Sera consistently improves forecasting performance, achieving up to a 37.3% reduction in prediction error over the temporal baseline and enhanced generalizability. Counterfactual analysis examines how forecasts respond to changes in degradation semantics to assess interpretability. The results show that prediction responses are consistent with the meanings of key degradation descriptors across tested horizons. Together, these findings demonstrate that structured degradation semantics and effective aggregation can improve forecasting accuracy and support reliable and interpretable battery health forecasting for advanced battery management.

Existing Open-Ended Deep Research (OEDR) systems primarily generate reports from scratch, making them inefficient for scenarios where research reports need to be continuously maintained as new information emerges. We introduce Incremental Open-Ended Deep Research (Incremental-OEDR), a research setting that treats a report as an evolving research state and incrementally updates it by preserving valid knowledge, revising outdated or incomplete content, and incorporating newly available information. To support this setting, we propose Structured Harness, which represents reports as structured collections of outlines, sections, and supporting evidence, and provides structured retrieval, a persistent structured evidence pool, and structured generation for selective report updating and evidence reuse. We further establish a temporal evaluation framework spanning ten years, with Single-Step Task and Long-Chain Task to evaluate incremental updates over both individual transitions and long-term update chains. Extensive Experiments on DeepResearch Bench and DeepConsult under both the Open-source Configuration (OC) and Proprietary Configuration (PC) show that Incremental-OEDR maintains competitive report quality while substantially improving report continuity and reducing research costs. As shown in Figure~\ref{fig:profile}, it achieves up to 0.51 higher content-level ROUGE-L F1, 0.63 higher outline-level EM F1, 33% lower token consumption, and 61% fewer search calls than OEDR on DeepResearch Bench. For more details, please refer to our project page: https://ioedr-project.github.io/.

Specialization maps are natural maps from equivariant vanishing cycle cohomology to the equivariant cohomology of the ambient space. We study their injectivity problem for moduli stacks of representations of quivers with potentials, including those associated with generalized conifolds and finite subgroups of $SO(3,\mathbb{C})$. Applications to the irreducibility of modules of Reshetikhin type shifted Yangians are also given.

In this article, we investigate the boundary controllability of finite point-vortex systems in bounded multiply connected planar Lipschitz domains. The controls are applied locally on an arbitrary nonempty relatively open subset of the boundary. Zero incoming vorticity is assumed at the inflow part of the boundary. For time-dependent controls, we prove exact trajectory controllability along arbitrary prescribed collision-free vortex paths remaining away from the boundary. For stationary controls, we prove exact endpoint controllability associated with mutually disjoint real-analytic arcs. The proofs rely on harmonic interpolation, localized Runge approximation, and an implicit-function argument. The optimal control minimizing the length of trajectories is also studied. We prove the nonexistence of minimizers within classes of slightly smooth controls, and establish existence when the control time and the size of the control are penalized.

Recently, academia and industry have considered the problem of approximate nearest neighbor search (ANNS) with filters. In this setting, each object consists of a high-dimensional vector and attribute values. Given a query vector, filters for attribute values, and $k$, this problem retrieves $k$ vectors approximately nearest to the query vector among a set of objects passing the filters. This paper considers range filters, i.e., users can specify a range constraint for each attribute. Most existing works do not consider this setting, and they assume (i) only a single attribute or (ii) matching filters that require the same attribute values or categories. Existing techniques for these assumptions are not available for our setting or are trivially not efficient. Although standard solutions, such as pre-filter and post-filter, can handle our problem, they are also inefficient. Some works tackle the same problem as ours, but their techniques necessitate historical query workloads, which significantly limit practical use cases. To remove these limitations, this work proposes Grant, a novel framework that solves this problem efficiently while accepting arbitrary range filters and ANNS data structures. Grant can guarantee a search time sub-linear to the number of objects, which is not held by existing techniques. We conduct extensive experiments, and their results demonstrate that Grant outperforms existing techniques.

Multipolar-fluctuation analysis based on density functional theory combined with dynamical mean-field theory is extended from $4f^1$- to $4f^2$-electron systems. Two methodological developments are presented. First, many-particle states are constructed within the $j$-$j$ coupling scheme in the $j=5/2$ subspace, so as to reproduce the crystalline-electric-field splitting of the $4f^2$ Hund's-rule ground-state multiplet. Second, multipolar susceptibilities and interactions between $4f^2$ multiplets are derived through a mapping from single-particle to many-particle bases. We apply this framework to PrCdNi$_4$, which has a non-Kramers $\Gamma_3$ doublet ground state, and propose a $(3z^2-r^2)$-type antiferroquadrupolar order at $\boldsymbol{q}=(2\pi,\pi,0)$ driven by the competition between nearest- and fourth-nearest-neighbor interactions. The formalism developed here is applicable to general $4f^n$-electron systems.

LLM-based agents now interact with their environments continuously, shaped by such organizational channels as user instructions, peer messages, and long-term memory. Existing safety research has examined these influences, but largely as separate agent components. How such factors jointly shape an agent's safety behavior from a unified perspective remains unmeasured. To bridge this gap, we advocate organizational behavior (OB) as a framework for studying the safety of advanced agents, reorganizing the objects of study, theoretical foundations, and experimental design around the relational structure in which agents operate. We present the first systematic formalization of counterproductive work behavior (CWB), a canonical safety-relevant subfield of OB, as Agentic Counterproductive Behavior (ACB). ACB specifies three organizational antecedents (vertical supervisor relations, horizontal peer norms, and internal cognitive structures) and maps them onto three counterproductive outcome dimensions (unauthorized disclosure, destructive operations, and production deviation). To operationalize ACB, we introduce Workerville, a controlled benchmark that manipulates organizational conditions over shared tasks, applying 16 organizational configurations to 210 tasks to yield 3,360 challenges, evaluated by human-validated agentic judges. Benchmarking 6 frontier LLMs, we find that (I) negative organizational antecedents exhibit non-monotonic amplification when combined, with the unauthorized-disclosure rate rising from 16.5% under no negative antecedent to 60.1% under two and falling back to 50.3% under three; (II) agents reproduce typical behavioral patterns predicted by human CWB research; (III) these results establish OB as a systematic framework for agent safety research, pointing toward a new research agenda.

Parallel magnetic resonance imaging reconstruction is an ill-posed inverse problem under undersampling. Multi-coil acquisition and Hankel lifting expose complementary repeated information: observations of the same anatomy across coils and repeated local k-space neighborhoods in overlapping windows. These dependencies guide recovery of missing k-space data. However, splitting lifted Hankel entries for self-supervision can place the original sample in both input and target, causing data leakage. We propose Hankel Subspace Self-Supervised Reconstruction (HSSRecon), a scan-specific reconstruction framework for parallel magnetic resonance imaging. HSSRecon partitions data by physical acquisition units before Hankel lifting and applies multiplicity normalization to repeated Hankel copies in overlapping windows. Rather than learning a mapping that directly predicts missing data, the network learns a compact complex-valued Hankel subspace operator. Reconstruction is performed over the original k-space variables using a conjugategradient solver with hard data consistency. This design separates structural learning in the Hankel domain from data consistency in the physical domain: the former exploits multi-coil and local Hankel correlations, while the latter solves over unacquired degrees of freedom. We provide theoretical analyses of physicalgroup splitting and multiplicity normalization, and establish positive definiteness, uniqueness, hard data consistency, and a finite-step conjugate-gradient error bound for the system. On fastMRI brain data with three contrasts and three sampling masks, HSSRecon achieves competitive peak signal-to-noise ratio, structural similarity, and normalized mean squared error across six aggregated conditions.

Coding assistants such as Claude Code and Codex have become a major application of LLM agents, yet existing benchmarks remain far from real-world use, particularly in task horizon and interaction length. Code assistants require completing long chains of development work in continuously evolving repositories, while repeatedly clarifying requirements and adapting implementations through multi-turn interaction. To address these gaps, we introduce SWE-Journey, a benchmark for more realistic evaluation of coding assistants. To address the task-horizon gap, we propose a weak-to-strong synthesis pipeline that automatically constructs long-horizon coding tasks. To address the interaction gap, we mine four representative user personas from real interaction data and build a user-simulation agent to reproduce realistic code-assistance interactions. On average, models pass over 75% of tests for requested functionality with software architects, but fewer than 25% with non-coders. These results show that current coding assistants still fall short of enabling reliable coding for non-coders. We further analyze the reasons for this gap and identify asking right, finding right, and fixing right as key capabilities during interaction.

Downlink channel state information (CSI) feedback is essential for beamforming optimization in frequency-division duplex (FDD) massive multiple-input multiple-output (MIMO) systems. However, the feedback overhead increases with the number of antennas and subcarriers, posing a major challenge to practical deployment. Although recent deep learning (DL)-based methods reduce this overhead by compressing CSI at the user equipment (UE) and reconstructing it at the base station (BS), most of them treat CSI as a generic image and rely on convolutional or Transformer architectures. As a result, the low-rank multipath prior is insufficiently exploited, while non-negligible computation is introduced at both the UE and the BS. To address this issue, we propose a low-rank prior (LRP)-guided CSI feedback framework that performs structured sensing and reconstruction directly over rank-one channel components. Specifically, the LRP encoder conducts learnable rank-one sensing to obtain path-aware measurements at the UE, while the LRP decoder reconstructs CSI through rank-one synthesis at the BS, avoiding the iterative recovery required by classical solvers. To compensate for finite-rank approximation residuals, we extend our previous DCRNet into DCRNetV2 by incorporating the proposed LRP backbone and gated dilated-convolutional residual branches. Experiments on multiple datasets and scenarios show that the proposed methods achieve a better accuracy-complexity tradeoff than existing model-based and DL-based baselines. In particular, standalone LRP achieves more than $10$~dB NMSE improvement over model-based baselines with much lower complexity, while DCRNetV2 achieves comparable accuracy to state-of-the-art Transformer-based methods with only about one-fifth of the complexity. The source code is available at https://github.com/tangshunpu/DCRNetV2.

Stop-and-go waves pose a persistent challenge in traffic flow theory, significantly compromising highway safety, operational efficiency, and environmental sustainability. This paper mathematically derives an optimal cooperative driving policy based on the premise that connected vehicles can cooperate and assume distinct roles. In doing so, we depart from the classical assumption of uniform vehicular behavior. Through a rigorous optimization formulation constrained by linear stability conditions, it is analytically proved that the traffic system's throughput is globally maximized when a single designated "buffer" vehicle maintains an enlarged headway to absorb perturbations, enabling the remaining vehicles to form a tightly spaced, high-capacity platoon. This analytically derived optimum is confirmed via numerical spectral analysis and stability-constrained flow optimization using the Intelligent Driver Model and the Full Velocity Difference Model. The results demonstrate that this non-uniform cooperative buffering strategy yields dual benefits. It radically expands the string-stable parameter space to actively dissipate SGWs, while simultaneously achieving significantly higher traffic flow compared to both unbuffered traffic and non-cooperative Jam-Absorption Driving strategies.