Papers

Intracranial electroencephalography (iEEG) is widely used to record electrical activity directly from electrodes inside the human brain, making it an attractive modality for neural decoding. However, progress in iEEG decoding, especially toward general-purpose foundation models, remains difficult to measure reliably: datasets are task- or institution-specific, limiting evidence of generalization across tasks and recording environments, and preprocessing choices can strongly influence performance, making model improvements difficult to distinguish from preprocessing gains. Thus, we introduce iMINDBench, an iEEG Multi-Institution Neural Decoding Benchmark that evaluates models on a shared suite of fifteen decoding tasks across three naturalistic movie-watching datasets. The benchmark additionally defines standardized preprocessing tracks and fixed evaluation splits to support consistent model comparisons. Using iMINDBench, we find that the evaluated pretrained systems generally outperform baselines within their respective preprocessing tracks, while strong spectral baselines remain competitive across institutional datasets. In our scaling study, adding up to 25 times more supervised data from other subjects or institutions yields only small or task-dependent gains over within-session training. Together, these findings highlight the need for iEEG models that improve on strong preprocessing baselines and make more effective use of data across subjects and institutions. Project website: https://imindbench.github.io/

We consider the determinant maximization problem under partition constraints: Given an $n\times n$ PSD matrix A and a partition matroid $M$ on $[n]$, find a base $S$ of $M$ that maximizes $\det(A_{S,S})$. We give an $e^{O(k)}$-approximation algorithm to find such a set $S$, where $k$ is the rank of $M$. This improves upon the current $k^{O(k)}$-approximation, and matches the current $e^k$-estimation guarantee, up to $O(1)$ factors in the exponent. Our algorithm is based on rounding the geometric max-min relaxation due to Nikolov-Singh'2016, using a continuous potential-driven process, and several new structural and analytic properties of this relaxation.

We propose Cat's Cradle diagrams as a framework to analyse the perturbative accuracy of gauge quantum field theories across different kinematic regions, interpolating between the power countings of fixed-order perturbation theory and soft-collinear resummation. We propose a simple intermediate counting aimed at quantifying parametric accuracy for resolved jet substructure, nested so that $N^nLO \subset N^nLJ \subset N^{n+1}LL$. We show that the resolved jet-substructure power counting, $N^nLJ$, emphasises contributions that are important for resolved substructure but which are lumped together with less dominant terms in both of the two other countings. It may therefore furnish a complementary quantification of relative parametric accuracy in a kinematic region "midway" between the hard (fixed-order) and soft-collinear (resummation) regimes. We discuss practical applications to parton showers, in the context of a hard process with a characteristic hard scale of order 200 GeV.

Communication is a central bottleneck in distributed optimization, but its effect is often summarized by the spectral gap of a chosen mixing matrix. Since this gap depends on link weights as well as topology, it can obscure the intrinsic effect of the network. We clarify this distinction by relating graph diameter to the inverse spectral gap. We establish a universal one-sided bound that is tight up to constant factors, while constructions across several graph families show that no topology-only converse or universal two-sided scaling law exists. This motivates a class-based view of network optimality: diameter-dependent guarantees can be assessed uniformly over graph classes, whereas spectral-gap optimality requires both the graph–matrix pair class and the admissible algorithm class to be specified. Based on the resulting diameter-based minimax benchmark, we introduce Tree-Routed Gradient Tracking (Tree-RGT), a distributed method for stochastic nonconvex optimization. The method pipelines model dissemination and gradient aggregation over a rooted shortest-path spanning tree. Each iteration uses one round of one-hop communication and one stochastic-gradient evaluation per agent, without inner consensus or multi-gossip. Under smoothness and unbiased bounded-variance oracles, Tree-RGT achieves uniformly optimal network dependence over connected graphs of prescribed size and diameter. It recovers centralized stochastic scaling after $\mathcal{O}(nD_{\mathcal{G}}^2)$ transient iterations, where $n$ is the number of agents and $D_{\mathcal{G}}$ is the graph diameter. This network-dependent transient-iteration bound improves upon or matches those reported for representative distributed methods. Thus, topology-aware routing attains diameter-optimal guarantees without relying on a prescribed mixing matrix.

Synthetic data can reduce the cost of collecting and annotating training data for robotic perception, but generating sensor observations that preserve the characteristics relevant to downstream perception remains challenging, particularly for sonar imagery. In this work, we investigate whether conventional image-fidelity metrics adequately reflect the downstream perception performance of GAN-generated synthetic sonar data. We employ a Pix2Pix conditional generative adversarial network with four discriminator configurations characterized by different receptive fields: PixelGAN, PatchGAN-16, PatchGAN-70, and ImageGAN. The models are trained using sonar imagery from two datasets and evaluated using conventional image-fidelity metrics, including Structural Similarity Index (SSIM), Peak Signal-to-Noise Ratio (PSNR), and Mean Squared Error (MSE). To complement these pixel-level measures with task-oriented evaluation, YOLOX-S, YOLOX-L, and Faster R-CNN detectors are trained exclusively on real sonar imagery and subsequently evaluated on the GAN-generated images using identical test samples and annotations across all discriminator configurations. The results reveal a discrepancy between image-fidelity and downstream object-detection performance: the configuration achieving the best SSIM, PSNR, and MSE does not consistently yield the best detection performance. In particular, PatchGAN configurations achieve strong downstream detection results despite not achieving the highest pixel-level similarity scores. These findings suggest, for the datasets and models considered, pixel-level image-fidelity metrics alone may not consistently capture the task-relevant realism of synthetic sonar observations and motivate the use of task-aware evaluation for synthetic sensor data intended for robotic perception.

Graph-based retrieval-augmented generation (RAG) can help answer questions that require information from many documents. However, building a graph often requires many language-model calls during ingestion. It is therefore important to ask whether its quality gains justify the additional cost. We present EffiRAG, a graph-based RAG system designed to reduce this cost. It uses the graph to locate relevant passages and generates answers from the original text. This design preserves source information while keeping graph construction and query processing lightweight. We evaluate EffiRAG on UltraDomain, which contains 120 open-ended questions from four domains. Compared with LightRAG-hybrid, EffiRAG produces the preferred answer on 93 questions. LightRAG is preferred on 7, and the remaining 20 are splits. EffiRAG also reduces total system cost by 57 percent, from USD 0.952 to USD 0.408. The cost includes language-model calls during ingestion and querying. The advantage remains as the corpus grows. At 10 and 20 documents per domain, EffiRAG uses a lightweight, non-LLM filter to skip low-salience chunks. It remains preferred over LightRAG-hybrid. It costs 4.2 times and 4.5 times less, respectively. The comparisons identify different quality-cost trade-offs. Graph-based RAG systems should therefore be evaluated by both answer quality and cost. The results favor graph structure that locates and preserves source evidence.

In this paper, we develop a systematic construction framework of hyper-derivative algebraic geometry codes via local expansions, extending hyper-derivative Reed-Solomon codes from the rational function field to general algebraic function fields. Using the residue theorem, we determine their Euclidean duals and illustrate that the duals naturally reverse. We further give criteria for reverse self orthogonality and reverse self duality in terms of two classes of bilinear forms. Finally, we provide an asymptotic bound on the rate and relative distance via function field towers.

In the simulation of conjugate magnetohydrodynamic (MHD) flows, where a moving conducting fluid dynamically interacts with bounding conducting walls under the influence of an externally imposed magnetic field, resolving the electrical conductivity transition across the fluid-solid interface introduces significant numerical challenges. In problems with high magnetic Reynolds numbers, it is required to solve the full magnetic induction formulation that has a curl-of-curl term for the magnetic diffusion, which is simplified to a divergence term in the original vector-valued lattice Boltzmann method (LBM). The resulting LBM scheme is valid for simulating fluid domains with a constant conductivity where its bounding solid walls are modelled using appropriate boundary conditions. The current study shows that this scheme is also valid for two-dimensional conjugate MHD simulations where the magnetic field component with non-zero gradients is perpendicular to the conductivity gradient. For general conjugate MHD simulations, we improve the LBM scheme by considering the difference between the curl-of-curl term and the simplified divergence form, which is treated as a source term in the LBM evolution algorithm and computed using an efficient LBM discretisation scheme. The improved LBM scheme automatically satisfies the required conjugate constraints at the fluid-solid interface as a volume integral inside two adjacent boundary layers. The accuracy of the improved LBM scheme is verified by piecewise analytical solutions in benchmark problems with an abrupt conductivity jump across the fluid-solid interface.

Community-based fact-checking is promising in countering misinformation, yet its scalability is constrained by slow rating accumulation. To address this challenge, platforms such as X implement platform-directed rating mechanisms, specifically through “Needs Your Help” algorithmic prompts, to target unresolved notes. Using a dataset of over 220 million rating contributions – including 1.9 million platform-directed ratings – on X, we examine contributors' response to note prompts, notes' resolution, and raters' spillovers. We found (i) at note level, population-sampled ratings concentrate on recent notes with certain helpfulness and high disagreement. Once sampled, population-sampled rating was associated with faster and more transitions to resolved statuses. (ii) At rater level, following raters' first observed population-sampled rating, raters exhibit significant yet modest increases in daily ratings, rating pace and tag usage, while other behaviors show no change. These highlight the promise of algorithmic nudges to guide volunteer attention toward contested content, accelerating consensus while sustaining rater engagement.

Recently, radiative coolers have been widely explored for reducing cooling loads or lowering temperatures in buildings, and at urban scales as a heat-mitigation measure. However, the potential impacts of radiative cooler deployment on the chemical composition of the atmosphere remain largely unexplored. A defining feature of recently-designed radiative coolers is their high ultraviolet (UV) reflectance, which is required for sub-ambient cooling under strong sunlight. Yet, wide adoption of such UV-reflective radiative coolers could substantially increase the UV actinic flux in the atmosphere above. This, in turn, may affect tropospheric ozone concentrations, particularly in urban atmospheres with high NOx concentrations. Here, as a case study, we use a 0-dimensional photochemical box model, constrained by field measurements of meteorological conditions and chemical concentrations in the urban environment of Houston, Texas, to explore the potential impact of the widespread use of UV-reflective radiative coolers on ozone concentrations. Our calculations show that complete deployment of radiative coolers may increase tropospheric ozone levels by as much as 30% during specific meteorological conditions in Houston. Informed by the wavelength-dependent modelling results, we propose specific designs, namely pigmented radiative coolers with different UV reflectances, and UV-absorptive visible-reemitting fluorescent radiative coolers, that could minimize negative ozone formation while retaining appreciable cooling performance. Our results motivate further study on the effects of widespread deployment of radiative cooling designs like superwhite roof paints on air quality, and materials that simultaneously minimize adverse photochemical impact and maximize cooling performance.

Research agents working in separate sessions need to know what others have tried and which results they can build on. Agora stores their contributions as an append-only directed acyclic graph (DAG) in Git. Each commit records a result, insight, hypothesis, verification, or report and links it to prior work. Searchable views show leading results, neglected branches, and verification status; diversity-aware recommendations suggest experiments beyond the current leaders. We report a run of nearly 12 days in which 13 language-model workers, with no assigned tasks or central planner, used Agora to solve a weight-transfer problem. Given 141 pretrained donor models and a frozen 119.6M-parameter attention–SSM hybrid whose dimensions match no donor, the workers had to initialize the target without training data or gradient updates. They published 1,703 contributions and reduced the development evaluator score from 3.39 to 1.899 bits per byte, closing 62% of the gap to a trained GPT-2 124M. The best method compresses donor next-token statistics into the target's embedding and output head, then adds short-range context through sparse edits to attention, feed-forward, and state-space blocks. Its 145-commit ancestry spans 15 accounts. Participants also posted 165 verifications of 95 targets, each by an account other than the target's author, with no reported failures. The run documents how agents reused and verified shared work. Measuring the effect on discovery per unit of compute requires a matched comparison.

We present a general method for estimating entropy differences between arbitrary probability distributions using stochastic interpolants. The marginal distributions which bridge the base and target obey a continuity equation, which allows a straightforward calculation of the entropy difference in terms of an inner product of the probability flow velocity and score fields. This formulation has several advantages: (i) no computationally expensive divergence calculations of either field are required, (ii) the score field need not even be learned directly if model transferability is not required, and (iii) on-the-fly estimates are produced nearly for free during training. When tractable base distributions are chosen (e.g. Gaussian chain, ideal gas, etc.), statistical thermodynamic entropies are then immediately recovered. Notably, the analysis relies only on the definition of the Gibbs-Shannon entropy, rather than any particular statistical mechanical ensemble, so that generalized non-equilibrium entropies may be computed. The method is demonstrated on several systems of increasing complexity: (i) a 40-dimensional Gaussian mixture model, (ii) the classical XY model of N spins arranged in one dimension, (iii) the 13-atom Lennard-Jones cluster, and (iv) an active Brownian polymer.

Global localization with scanning millimeter-wave radar remains challenging because place-recognition descriptors often discard spatial structure needed for accurate pose retrieval. We present ReRadar, a radar global localization pipeline that extracts rotation-equivariant intermediate features using steerable convolutional neural networks, forms rotation-invariant descriptors through group pooling and NetVLAD aggregation, and combines descriptor retrieval with landmark-based matching to estimate the robot's three-degree-of-freedom (3-DoF) pose. Across fixed database-query evaluations, ReRadar with target-dataset adaptation achieves 99.37% Recall@1 on OORD Bellmouth, 91.44% Recall@1 with 80.99% F1_max on Mulran DCC01, and 99.38% Recall@1 on falling-snow Boreas sequence. Without target-dataset data, the cross-dataset model reaches 98.07% Recall@1 on OORD, performing comparably to the evaluated state-of-the-art methods.

For decades, the NASA Astrophysics Data System (ADS) has been the foundational digital library for astronomy. As ADS becomes the Science Explorer (SciX), the team prioritizes preserving this trusted experience while increasing their support for space and Earth sciences broadly, without diluting its commitment to astrophysics.

We develop a microscopic theory of spin-torque ferromagnetic resonance (ST-FMR) in nonmagnetic-metal/ferromagnetic-metal heterostructures. Tracing out the conduction-electron degrees of freedom in the FM, we derive an effective interfacial exchange coupling between the localized spins and the conduction electron spins in the adjacent electron system. Based on this interaction, we calculate the current-induced driving torque, resonance-frequency shift, damping modulation, and resulting dc voltage. As a concrete example, we apply the formulation to a disordered Rashba two-dimensional electron gas and demonstrate that the ST-FMR spectrum reflects the dynamical spin responses of the adjacent electron system. Our formulation applies to a broad class of heterostructures and establishes a unified microscopic framework connecting ST-FMR spectra directly to the electronic spin responses of adjacent systems.

Vehicular federated learning (VFL) enables privacy-preserving collaborative model training for intelligent transportation systems, where communication resource allocation and gradient sparsification techniques have been explored to reduce communication overhead. However, vehicle mobility leads to rapidly varying channel conditions and transmission capacity, rendering predetermined resource allocation and sparsification decisions ineffective. In this paper, we propose FedPGT, a progressive gradient transmission scheme for VFL over time-varying channels, where vehicles progressively transmit high-magnitude gradient entries in response to instantaneous channel conditions. We establish a convergence bound that characterizes the impact of transmitted gradient entries and reveals diminishing-return behavior governed by a power-law decay. Motivated by this result, we formulate a stochastic optimization problem for online decision-making, where the main challenge lies in a cumulatively coupled, non-separable objective. To handle this challenge, we introduce per-slot surrogate transmission variables to decouple the long-term dependence across time slots and convert the original objective into an additive per-slot optimization problem, enabling a Lyapunov drift-plus-penalty approach for online scheduling. We further develop a low-complexity resource allocation algorithm for efficient online implementation. Experimental results demonstrate that the proposed scheme achieves a 3.65% accuracy improvement on the CIFAR-10 image classification task and a 12.66% reduction in average displacement error on the Argoverse trajectory prediction task compared with state-of-the-art baselines, demonstrating its applicability to diverse learning tasks under highly dynamic vehicular environments.

Detection-free methods for image-to-point cloud registration are prone to erroneous correspondences caused by domain and modality discrepancies, limited sensitivity of feature extractors, and the presence of non-overlapping regions. The Masked Autoencoder (MAE) has shown strong performance in visual representation for images and point clouds. It may be helpful to apply this approach to image-to-point cloud registration, a task that requires unified feature extraction and accurate cross-modal correspondences. Standard MAE's random masking may overlook key regions due to limited camera views, reducing registration effectiveness. To address this, we propose the Intermodal Dual-MAE Framework (ID-MAE) with a Similarity-based RL Masking Strategy (SRLM), which adaptively masks informative positions by leveraging cross-modal similarity and reinforcement learning, thus narrowing the modality gap. Our method enhances cross-modal representation learning by enforcing representation consistency during feature extraction, thereby enabling more reliable 2D-3D correspondence estimation. Experiments on RGB-D Scenes v2 and 7-Scenes benchmarks show that our method achieves state-of-the-art performance in image-to-point cloud registration.

Let $k$ be an arbitrary field and let $S=k[x,y,z]$ be the polynomial ring with three variables $x,y,z$. We study integrally closed $(x,y,z)$-primary ideals of $S$ that are homogeneous for a positive weighted grading but need not be homogeneous for the standard grading. From a numerical semigroup $H$ of embedding dimension three, we obtain such ideals as inverse images $I_h=\varphi_H^{-1}(t^hk[t]\cap k[H])$. If $\ell$ is the least degree of a defining relation of $k[H]$, then $I_\ell$ is monomial, whereas $I_{\ell+1}$ has a binomial generator in the cases considered here. We determine $I_{\ell+1}$ for numerical semigroups of embedding dimension three and multiplicity three or four. The six-generated cases arising in multiplicity three and in the symmetric multiplicity-four case form two explicit families. For every ideal $I$ in these families, we prove that its Rees algebra is a Cohen–Macaulay normal domain. In the non-symmetric multiplicity-four case, $I_{\ell+1}$ is seven-generated; for $H=\langle4,9,15\rangle$, we prove that its Rees algebra is again a Cohen–Macaulay normal domain.

We present a detailed analysis of FRB 20250607A, a two-component FRB detected by the CRAFT survey which exhibits several unusual component-dependent properties, including apparent differential DM, circular polarisation, and time-resolved RM, together with a polarisation-angle (PA) swing that persists through the scattering tail. The scattering timescale and scintillation bandwidth are inconsistent with a single thin-screen origin, placing the dominant scattering screen within $\lesssim61$ kpc of the FRB source. The lack of PA flattening in the scattering tail is a behaviour not previously reported in an FRB or pulsar and difficult to reconcile with a simple propagation origin. We find no statistically significant differential DM between the two components despite an apparent, and visually obvious, difference in their dispersion, indicating that current uncertainties may be overly conservative. If this apparent differential DM is true, it implies a substantial difference in electron column density between the component sightlines. Two further CRAFT FRBs, FRB 20190611B and FRB 20210407E, show similarly large component-dependent electron-column differences, corresponding to electron-density contrasts consistent with plasma local to a neutron-star magnetosphere. We find no evidence for generalised Faraday rotation, and FRB 20210407E shows component-dependent PA structure. Taken together, these results suggest that the observed DM, circular polarisation, and PA behaviour may arise from differing sightlines through an inhomogeneous or evolving magnetosphere rather than from independent phenomena. Multi-component FRBs therefore provide a sensitive probe of plasma structure local to their progenitors.

In the present study, the consistent and conservative Phase-Field method is extended to the six-equation model for compressible multiphase flows. Based solely on the conservation laws and the second law of thermodynamics, the six-equation model with the Phase-Field mechanism is first derived. In addition to satisfying Galilean invariance and consistency of reduction, the model is general to admit an arbitrary number of phases with different formulations of the Phase-Field mechanism. The isobaric closure by the pressure relaxation and the incompressible limit of the proposed model are analyzed. The derivation and analysis identify additional terms arising from the Phase-Field mechanism, which are absent from previous studies. The consistent and conservative numerical approach is adapted to the proposed six-equation model with moderate modifications, retaining the interfacial equilibrium condition, conservation, and flexibility and robustness of incorporating different formulations of the Phase-Field mechanism with bound preservation. As the new component of solving the six-equation model, both the pressure and pressure-temperature relaxations are theoretically analyzed in a general multiphase setup, with a proof of the existence and uniqueness of a thermodynamically admissible solution for these two relaxations. Various two-phase compressible flow benchmarks are performed to demonstrate the method, and good agreement with exact solutions is achieved.

Fine-tuning a Vision-Language-Action (VLA) model for a new deployment environment is expensive, yet most methods apply uniform-capacity adapters to every network region as if every region requires equal adjustment. This paper tests that assumption on five architecturally diverse VLAs (OpenVLA-OFT, $\pi_0$, SmolVLA, DTP, Octo; 93M-7B parameters). Measuring per-region adaptation cost as normalized parameter displacement under region-isolated fine-tuning reveals an adaptation spectrum in which appearance shifts concentrate cost in the vision encoder, instruction shifts in the language backbone, and novel-object shifts in the vision encoder together with the action head, across all five architectures. To exploit this structure, we introduce a pipeline that observes, diagnoses, allocates, and adapts. From ten unlabeled target observations and without fine-tuning, the diagnostic estimates per-region cost by combining reference-free gradient and Monte Carlo Dropout signals with a Centered Kernel Alignment score against a cached source reference; the allocator converts the estimates into variable-rank LoRA adapters under a parameter budget and freezes well-calibrated regions; and standard LoRA fine-tuning trains the resulting adapters. The diagnostic ranks regions within each deployment at a median Spearman of 0.91, and the allocation matches or exceeds uniform LoRA at every budget we tested on LIBERO and CALVIN. On a physical xArm-7, the pipeline matches full fine-tuning under an instruction-wording shift with 0.04% of its trainable parameters, and on five held-out scenes evaluated without retraining it leads every baseline, with 11-23 successes of 30 rollouts against 8-18 for the strongest parameter-efficient baseline at equal or larger budgets and 2-11 for full fine-tuning. These results suggest that adaptation cost in VLAs is structured enough to measure before fine-tuning begins.

Integrated sensing and communication (ISAC) is a cornerstone of sixth-generation (6G) networks, yet conventional fixed-antenna systems lack the spatial adaptability to cope with dynamic users and targets. The emerging pinching antenna (PA) offers a flexible, low-cost solution by dynamically reconfiguring radiation points along waveguides, introducing large-scale spatial degrees of freedom. This article develops a unified architectural perspective for PA-enabled ISAC. We first discuss the unique advantages of PAs over existing flexible solutions, and then propose a PA-enabled ISAC framework that accommodates both uplink and downlink communication while being compatible with passive and active sensing targets. Within this framework, we identify representative application scenarios, discuss major design challenges, and highlight critical enabling techniques. A numerical case study demonstrates how PA reconfigurability affects the communication-sensing rate trade-off. We also outline open issues to guide further PA-ISAC research for future 6G networks.

In this paper we tackle the problem of aligning LiDAR point clouds also known as the point cloud registration problem. We propose a new algorithm, PESTO, that exploits tetrahedra as "universal features" for LiDAR data, i.e., features that are agnostic to the environment where the LiDAR sensors are deployed. We show empirically that PESTO is competitive with existing solutions for aligning LiDAR point clouds, especially in environments with occlusions. Moreover, we establish PESTO's formal correctness by proving worst-case bounds on the alignment error.

Direct entangling gates between spatially separated qubits are a key capability for modular quantum computing. They enable nonlocal quantum circuits to be executed across different processor modules, thereby alleviating single-chip scaling constraints and expanding the effective connectivity of the system. Here, we propose and numerically investigate a remote adiabatic controlled-Z gate between two frequency-tunable superconducting qubits connected by a multi-mode coaxial cable. The cable-mediated interaction enables the gate to be implemented without additional tunable couplers, preserving a simple circuit architecture. For a 30-cm cable, simulations incorporating ten standing-wave modes yield an optimized gate infidelity of $4.71\times10^{-7}$ with a duration of 376.7 ns in the absence of decoherence. The gate maintains high fidelity under qubit-frequency fluctuations and millimeter-scale cable-length variations. Calculations for 50- and 100-cm cable situations demonstrate its potential for even larger cryogenic systems. These results provide a promising approach to extending high-fidelity quantum operations beyond individual chips and toward scalable, interconnected superconducting quantum processors.

Linear probes can decode safety-relevant concepts such as truthfulness from language-model activations, but probe accuracy may show only decodability, not that the features the probe weights causally drive model behavior. We demonstrate that this gap cannot be closed from the geometry of probe weights alone: the features geometrically aligned with probe direction need not be the ones the model uses, so causal relevance requires intervention. We introduce a feature-level diagnostic that decomposes a deployed True/False probe into sparse-autoencoder (SAE) features, ranks those features by both probe alignment and by gradient sensitivity of the model's behavior, and ablates the resulting shared, probe-only, and random feature sets under a coherence gate. On the truth probe of Buerger et al. (2024) (TTPD), applied in the instructed truth/deception setting of Long et al. (2025) for Gemma2-9B-Instruct, the two rankings overlap only weakly (about 12%, Spearman rho = 0.10), and ablation dissociates them sharply: features the probe shares with the model flip the output far more (up to 27%) than equally sized probe-only (6%) or random (1%) features at full coherence, while probe-only features instead perturb the probe's own readout. The dissociation holds across five seeds and a held-out split, and an activation-aware selection of features flips behavior nearly three times as often as the probe's geometric top features (17.6% vs. 6.1%). In this setting, therefore, the geometric projection of a probe's weight vector alone does not identify the features the model causally uses; however, combining probe information with feature activation statistics recovers substantially more behaviorally causal features, and coherence-gated SAE intervention is needed to separate them from probe readouts.

Classically, the angular resolution of a telescope is limited by its diffraction limit: the ratio of its observing wavelength to its diameter. Developments in quantum information theory over the past decade have uncovered quantum-optimal measurement methods that enable super-resolution imaging beyond the diffraction limit, and many laboratory-scale demonstrations of these methods have been performed over the past decade, culminating in the recent first on-sky demonstration of quantum-enhanced astronomical imaging. In this Review, we detail the current progress and achievements of these theoretical and experimental results, discuss the astronomical science cases motivating the development of higher resolution telescopes, summarise the technological challenges facing the practical application of quantum imaging technologies to astronomy, and present a vision for the realization of quantum telescopes with imaging resolutions orders of magnitude better than current observatories.

Warm-start transfer accelerates delayed generalization (grokking) on algorithmic tasks, yet how internal blocks, residual-stream depth, group non-commutativity, and continual optimization govern transferred circuits remains unclear. Across ten scale-matched batteries ($>1{,}800$ trajectories) on abelian $\mathbb{Z}_{113}$ ($L \in \{1,2,3\}$) and non-abelian $S_5$, transferring internal blocks alongside boundary carriers lifts early accuracy by $5.46$ percentage points and cuts validation confirmation latency by $616.7 \pm 129.0$ steps in 1-layer models ($p_{\rm Holm}=0.004$). In 2-layer models, full transfer compresses trigger onset $8.1\times$ ($+704.7 \pm 84.3$ integral units), whereas omitting donor embeddings incurs a $-4{,}475.6 \pm 76.8$-unit interface deficit. On ternary $S_5$ conjugation ($aba^{-1}$), two-token architectures stall in a unary conjugacy-class shortcut ($\mathrm{Cl}(aba^{-1})=\mathrm{Cl}(b)$, $34.48\% \pm 6.19\%$ peak accuracy). Combining three-token causal routing ($[a,b,a]$) with Peter–Weyl width ($d_{\rm model}=256 > \sum_\lambda d_\lambda^2 = 120$) reaches $97.97\% \pm 1.01\%$ ($80.08\% \pm 7.81\%$ at equal $427\text{k}$ parameters) with $3.2\times$ faster onset than scratch. Under continued AdamW optimization, unshielded circuits relapse by $11.85\%\text{--}80.58\%$ drawdown, and depth-$L$ stability follows a graded $(L-1)$-prefix pinning hierarchy ($75.99\% \to 31.26\% \to 0.32\% \to 0.04\%$ drawdown at $L=3$). Finally, Closed-Loop Curvature-Anchored Consolidation ($\eta_k = \eta_0(1 - A_{\rm val}(k))$ with monotonic AMSGrad second moments) holds saturated drawdown to $0.00\%\text{--}0.12\%$ and lifts pre-saturation terminal accuracy by $+1.67\text{--}5.37$ points over fixed-budget micro-stepping ($87.06\% \pm 0.94\%$ vs.\ $81.69\% \pm 0.90\%$ on $S_5$, $p_{\rm Holm}=0.002$).

Open-source video generative models ship almost exclusively as PyTorch/CUDA reference implementations. This leaves Cloud TPU pods without a production-ready inference path, despite offering large, cost-effective accelerator memory pools ideal for long-sequence spatiotemporal attention. We present vidax, an open-source JAX/Flax inference engine and zero-copy PyTorch-to-JAX weight translator for modern video generation architectures. vidax covers a diverse set of spatiotemporal models — including Diffusion Transformers, omnimodal Mixture-of-Transformers, 3D VAEs, text encoders, and native samplers — with zero PyTorch dependency in the execution path. The framework unifies 1D tensor parallelism with DeepSpeed-Ulysses sequence parallelism on a single JAX sharding mesh, integrates TPU flash-attention kernels, and implements per-layer weight offloading to support reference resolutions that exceed single-device memory. We benchmark compile times, latency, and peak memory utilization on TPU v4-8 hardware, and document real-world numerical bugs surfaced during checkpoint translation. vidax is released open-source as a baseline for JAX and TPU video generation research.

The topology of a bounded three-dimensional domain can strongly affect electromagnetic fields. In a topologically complex domain, a curl-free field may not have a globally single-valued scalar potential and a divergence-free field may also fail to have a global vector potential. These topological effects lead naturally to harmonic vector fields in the Helmholtz decomposition. This paper gives a geometric and constructive study of such fields using vector analysis, circulation integrals, cutting surfaces, scalar Laplace problems, Stokes' theorem, and Green's identity. The first Betti number is interpreted through independent handle-type circulations, whereas the second Betti number counts enclosed voids. Direct proofs are given for the dimensions of the Neumann and Dirichlet harmonic-field spaces. The analysis is also extended to anisotropic lossless media. The results provide a simple topological interpretation of harmonic fields and physical DC modes in bounded electromagnetic resonators.

Conventional video metrics such as PSNR, SSIM, and VMAF measure visual distortion or perceptual quality, but they do not directly capture semantic preservation: whether compression retains a video's objects, actions, and temporal narrative. Existing Quality-of-Experience (QoE)-driven bitrate-selection and resource-allocation methods primarily aim to minimize rebuffering and bitrate switching while maximizing perceptual video quality, without explicitly considering semantic preservation. To address this gap, we introduce video semantic fidelity (SF), a metric that quantifies how well a compressed video preserves the semantic content of its source. An offline multimodal large language model (MLLM) generates structured descriptions of the reference and compressed versions of the video, and a separate text-only large language model (LLM) evaluates their semantic correspondence. The resulting content-dependent SF–bitrate profiles are cached and queried by the online bitrate selector without invoking MLLMs at runtime. Evaluations on three subjective QoE benchmarks show a consistent positive association between SF and mean opinion scores (MOS). A separate human semantic-rating study evaluates semantic preservation and shows that SF correlates more strongly with human judgments than conventional video metrics. We then embed these profiles into a 5G MEC-assisted video-on-demand (VoD) resource-allocation framework at the base station. When wireless resources cannot support high bitrate levels for all users, the framework uses the SF–bitrate profiles to select bitrate levels jointly across users and reduce the semantic loss caused by the required bitrate reductions. NS-3 simulations with the 5G NR module show that the proposed framework achieves higher average and worst-user SF than the evaluated baselines, with a widening advantage as the wireless resources available to each user decrease.