Papers

Large language models (LLMs) are entering wildfire management, where overstated evaluations can cost property and lives. How do they perform on wildfire tasks, with and without grounding? Bare means a model receives the task input alone. Grounded means it also receives one task-specific addition: for smoke detection, a smoke-free reference frame from the same camera. AI4Fire runs six core models bare and grounded on five wildfire tasks, zero-shot; a sweep adds 29 more. Our literature search on fire tasks found 138 works; none combines this roster, task coverage, and paired bare and grounded runs. We report three findings. (1) Grounding helped most where the addition carried the answer: a read-only SQL tool lifted every core model's database accuracy from at most 16 to at least 88 percent. (2) Simple rules were hard to beat: no core model outperformed repeating today's staffing count, and two open-weight models mostly copied the median of similar earlier fire-days, a worse forecast. (3) Public releases carry hazards: a fire-danger column separates the holdout perfectly, and 67 aerial fire frames carry smoldering or fire-free labels read from a clipped thermal maximum. We release prompts, responses, scores, code, and the survey record.

We present GHARP (Real-time Gaussian Head Animation from Large-scale Reconstruction Prior), a method that animates 3D human heads in real time from a few input images of a subject and a driving expression signal. We decouple the problem into an identity stage that builds a representation of the subject's geometry and appearance offline, and an animation stage that predicts expression-dependent residuals on top of it at runtime. This separation offers a favorable trade-off with respect to fidelity, quality and runtime: the identity stage can be expensive while the animation stage runs a lightweight network, optimized for mobile devices. Our method performs animation in a semantically structured latent space of a pretrained reconstruction model, where expression changes remain spatially contained, making residual prediction efficient. This reconstruction prior provides a consistent spatial layout, allowing fusion of multiple input views into a compact, fixed-size canonical Gaussian representation. While this two-stage design improves the runtime-quality trade-off, it still inherits a problem common to all expression-driven avatar methods: expression codes describe only the face and thus omit body pose and clothing position, making these regions underspecified in the input. The animation network faces an ill-posed mapping and resorts to averaging over conflicting body appearances, producing blur and temporal flicker. We address this with a body alignment network that learns to align the person's body in the target image with the input reference images, removing the ambiguity from the training signal. Our method achieves state-of-the-art quality on the Ava-256 benchmark while running up to 13x faster on an A100 GPU with 8x fewer Gaussians.

Polariton formation and quantum vacuum dressing are usually regarded as intertwined consequences of ultrastrong light–matter coupling. Here we show that they can be selectively controlled through the relative direction of rotation of the cavity field and the matter excitation. We consider cyclotron resonance coupled to a chiral cavity supporting a single circularly polarized mode. We quantize the electromagnetic field within the cavity, derive from first principles the Hamiltonian, and investigate how mode hybridization and vacuum dressing are rendered as we tune the magnetic field. When the cavity field co-rotates with the cyclotron motion, energy is exchanged resonantly between light and matter, producing a large Rabi splitting and conventional polaritons, while the ground state remains the bare vacuum. Reversing the magnetic field makes the cyclotron motion counter-rotating with the cavity field. The same cavity mode then couples through anomalous processes that create and annihilate light and matter excitations in pairs: the conventional polaritonic anticrossing disappears, while the ground state acquires squeezing and virtual excitations. Our results show that spectral hybridization and quantum vacuum dressing are distinct manifestations of ultrastrong coupling, offering an additional degree of freedom for engineering cavity quantum materials.

The development of accurate and efficient computational methods for fluid-structure interaction (FSI) problems remains an active research area due to its broad engineering and biomedical applications. Partitioned loosely coupled (LC) schemes are attractive for large-scale simulations because of their modularity and low computational cost. However, their applicability is limited by stability issues in high added-mass regimes, where fluid and structure densities are comparable, as in hemodynamics. In our recent work, we proposed an LC algorithm based on Dirichlet-Neumann interface coupling, the LC-DN-{\alpha} scheme, which improves the stability of the coupling through a stabilization parameter {\alpha}. Although theoretical stability has been demonstrated under suitable constraints on {\alpha}, its optimal value depends on the physical and numerical characteristics of the FSI problem and is difficult to determine in practice. In this work, we propose a data-driven strategy to automatically estimate the optimal value of {\alpha}. We develop a feed-forward neural network to predict {\alpha} from the descriptors of the FSI problem. The model is trained on a dataset of baseline FSI problems in large added-mass regimes and evaluated on more realistic configurations to assess robustness and generalization. Numerical results show that the proposed model generalizes across the considered FSI scenarios and provides robust estimates of {\alpha} under moderate variations in the problem setup, including changes in geometry and loading conditions, that do not completely alter the interaction mechanics represented in the training set. The methodology is intentionally developed and validated on simplified FSI problems, providing a proof of concept for data-driven strategies to improve the efficiency of partitioned FSI simulations and supporting their extension to more complex and realistic scenarios.

Reinforcement learning environments are now a primary lever for improving large language model (LLM) capabilities in post-training, yet most agentic benchmarks remain static: the world moves only when the agent acts, the reward is a terminal verdict, and the pass bar is set arbitrarily. We introduce StoreBench, a live-commerce environment in which an agent runs a mid-size online apparel store on a production-grade commerce backend, testing long-horizon planning and economic judgment under uncertainty. Customers order around the clock, suppliers reprice and fail, and market shocks arrive with partial or no warning. The agent acts through the same 29 merchant tools a human operator would use, under a windowed operation budget that makes simulated time a function of actions taken, so model latency cannot influence simulated time. Pass thresholds are calibrated against scripted anchor policies, the reward is hardened against a catalogue of reward hacks, and every episode replays identically given a sequence of actions. We evaluate seven frontier LLMs on 11 scenarios of 30 to 45 days and a full simulated year, over three world seeds at matched reasoning effort. No model matches the scripted smart-triage policy on average: the best, DeepSeek-V4-Pro, passes 49% of task-seed cells against the heuristic's 97%. Human experts working through the same tools and budgets outscore every model (mean composite 0.708 vs. 0.700). Over a full simulated year under the Claude Code harness, most models show dramatic performance improvement. In a GRPO post-training run, Qwen3.5-27B trained on only five disjoint tasks raises its mean composite on the held-out evaluation tasks from 0.136 to 0.373. We release five example training-split tasks, ten sample trajectories, and the scoring and verification tooling; the full environment and evaluation suite are withheld to keep the benchmark uncontaminated.

Dark matter subhaloes are a fundamental prediction of the standard cosmological model, but their low-mass population remains difficult to constrain observationally. Recent work has shown that interactions between subhaloes and dwarf spheroidal galaxies can induce large-scale, out-of-equilibrium fluctuations in their stellar surface density fields, identifiable through Fourier-based statistics. We apply this formalism to deep Subaru photometry of the Draco dwarf, extending previous tests to $\gtrsim\mathcal{O}(10^{4})$ resolved stars. We use artificial-star tests to assign completeness weights to individual stars and incorporate them into a discrete likelihood-based framework. Relative to a cored axisymmetric Plummer model, Draco shows radial corrugations that appear in the filtered normalised residuals as a central stellar depletion surrounded by a ring-like excess. The same feature appears in Fourier space as a low-frequency peak with a signal-to-noise ratio ${\rm S/N}=6.0$. Independently, fitting the surface density $\Sigma$ with an $\alpha\beta\gamma$ model yields ${\rm d}\Sigma/{\rm d}R>0$ at $>99.9\%$ confidence – i.e., a profile that decreases towards the centre. Although these signatures cannot yet be regarded as robust evidence for dark subhaloes, they suggest that central stellar depletion in dwarf galaxies may reflect genuine physical structure rather than crowding alone.

In quantum mechanics of genus-1 potentials, perturbation theory around the topologically trivial saddle and non-perturbative instanton saddle are constructively related via the P/NP relation. Whether such a relation persists for higher-genus potentials has remained an open problem. In this work, we resolve it using tools from complex algebraic geometry combined with exact WKB. The classical energy conservation relation $p^2 = 2(E-V(q))$ defines a genus-g Riemann surface $X$, whose periods are naturally organized by its Jacobian variety $J(X) = \mathbb{C}^{\text{g}}/\Lambda$ via the Abel-Jacobi map, and independently by the solution basis of the associated Picard-Fuchs equations. We show that the physically relevant, WKB-active cycles are generally related to the Picard-Fuchs basis by a linear transformation that lies in $\text{SL}(2\text{g},\mathbb{Z})$ but not in the symplectic group $\text{Sp}(2\text{g},\mathbb{Z})$ once $\text{g}\geq2$, in contrast to the genus-1 case, where the two groups coincide. This mismatch is the obstruction that has made an explicit higher-genus P/NP relation elusive. We show that it can be resolved by passing to a modified Riemann bilinear identity, built from a suitably transformed intersection matrix, which the WKB-active periods do satisfy exactly. The result is an all-orders quantum P/NP relation valid for potentials of arbitrary genus, which we derive explicitly for genus-2 (quintic and sextic) and genus-3 (septic and octic) anharmonic oscillators.

Climate policy changes production costs and can reshape the geography of production. We study this spatial channel in a focused model where mobile industrial activity responds to local attractiveness, agglomeration strengthens attractiveness, and carbon policy lowers the profitability of emissions-intensive locations. The model yields an explicit carbon-policy threshold separating spatial instability from stability. When emissions intensity rises with local industrial density, spatial concentration raises aggregate emissions even with unchanged total industrial activity. Embedding these dynamics in a two-jurisdiction policy game creates a wedge between regional and collective incentives: each government values the industrial activity located within its own territory, while aggregate welfare internalizes the effect of policy on economy-wide concentration. With transferable utility, cooperation can select a policy on the stable side of the threshold while the non-cooperative policy lies on the unstable side. A numerical exercise illustrates this regime change. European regional evidence, based on industrial emissions and gross value added at NUTS2 level, documents a positive emissions-activity association and identifies the empirical quantities needed to measure relocation and concentration effects. The analysis links climate cooperation to the spatial organization of production and establishes a benchmark for policy-induced industrial relocation.

The purpose of this work is to introduce and discuss a general framework for performing a linear instability analysis on multiplex networks, in the absence of inter-layer diffusion, with the specific motivation of detecting the Turing instability. The formalism itself is presented in full generality for $m$-unknowns, each of which arise as solutions of non-autonomous reaction-diffusion equations, on $n$-node temporal multiplex networks. In addition, we identify a class of multiplex networks for which the Laplacian matrices commute. In this setting analytical progress can be made and the explicit inequalities, governing instability, are derived. In order to explore our formalism we apply it to the specific example of a two-species predator-prey model on archipelagos. Here nodes are interpreted as islands and edges correspond to the transportation network for each species. Moreover, the non-autonomous setting is used to model a variety of natural phenomena.

In this paper we investigate adapted complex polarizations associated with left-invariant Koszul connections on real Lie groups. Adapted complex polarizations arise as a natural generalization of a complex structure on a manifold. We investigate the existence of such polarizations on the tangent bundle of certain Lie groups with left-invariant connections. We use a practical criterion for the existence of global adapted complex polarizations in terms of the $i\mathbb R$-completeness of the geodesic flow of the connection. In the Lie case, the Euler-Arnold vector field determines the behaviour of the geodesic flow. By combining what we know about the Euler-Arnold vector field of certain Lie groups and the $i\mathbb R$-completeness criterion, we provide several examples of left-invariant ac-polarizations defined on the tangent bundle of Lie groups. Along with pseudo-Riemannian metric connections, we discuss biinvariant connections in several cases. Our focus is on solvable Lie groups, because non-solvable Lie groups are better understood. The main result of this paper is that we demonstrate several not locally symmetric but $\mathcal P$-complete connections.

The conditional colour-magnitude distribution (CCMD) model connects the galaxy colour-luminosity distribution with dark matter halo mass, which has been constrained in Xu et al. by jointly modelling the two-point correlation functions of a large galaxy sample of the Sloan Digital Sky Survey (SDSS) in fine bins of galaxy colour and luminosity. We test the CCMD model of the galaxy-halo connection using counts-in-cells (CIC) statistics by comparing CCMD model predictions with SDSS measurements for 16 subsamples of galaxies in various luminosity and colour bins. In addition, we compare SDSS measurements with three analytic models of CIC distribution. The CCMD model provides a good description of the SDSS CIC measurements for most subsamples. The exceptions are the void probability functions and CIC distributions with cell radius 5 Mpc/h for a few subsamples in the luminous bins, which are not well described by the analytic models, either. For these subsamples, in general, removing the Sloan Great Wall from SDSS or introducing velocity bias to the CCMD model does not resolve the disagreement. Finally, with the help of the CCMD mocks, we investigate the cause of the variations of the galaxy counts in cells. For small cells or cells with a low mean galaxy count, the variation reflects effects from fluctuations in both the matter environment and the halo occupation distribution, while for large cells or cells with a high mean galaxy count, the environment effect dominates. We discuss the implications of our work for investigating galaxy assembly bias.

We study the asymptotic behaviour of fractional Gagliardo-type energies with a periodic, bounded and bounded away from zero coefficient depending on both variables, in the joint limit of vanishing fractional exponent $s\to0^+$ and vanishing period $\varepsilon\to0^+$. Assuming without loss of generality that $\varepsilon^{-2s}$ converges to some $\lambda\in[1,+\infty]$, we prove that the energies, rescaled by $s$, $\Gamma$-converge in the weak topology of $L^2(\mathbb{R}^d)$ to a multiple of the squared $L^2$ norm, whose constant is given by an effective coefficient $a_\lambda$ defined through a cell problem. The coefficient $a_\lambda$ results from the competition between the averaging of long-range interactions, which drives the Maz'ya–Shaposhnikova limit, and the nonlocal energy of microscopic oscillations at finite range. It interpolates between the harmonic mean ($\lambda=1$) and the arithmetic mean ($\lambda=+\infty$) of a coefficient obtained by averaging interactions “at infinity”. The proof relies on Riemann–Lebesgue-type lemmas for singular kernels, on a blow-up argument for the lower bound, and on the logarithmic-scaling properties of the energies. The case $1<p<\infty$ is also discussed.

Actuator degradation turns quadruped locomotion into a coordination problem requiring joints to compensate for lost actuation. Prior work suggests that morphology-aware graph policies improve learning and generalization under body perturbations. We ask whether these benefits can be strengthened by explicitly modeling higher-order mechanical structure. We represent the Unitree Go1 as a cell complex with limb- and body-level rank-2 cells and apply Hodge-based message passing. Under degradation training, the node-edge-face Hodge actor achieves the highest return on unseen actuator degradations, with higher survival and lower velocity-tracking error. These results support higher-order morphology as a useful inductive bias for whole-body compensation under actuator degradation.

Spiking language models face a tradeoff between representing continuous semantic features over short temporal windows and retaining costly nonlinear attention operations. We introduce Spora, which jointly designs spike encodings and attention operators. Binary temporal weights let $T$ spikes represent compositional values with up to $T$ bits of capacity, compared with $O(\log_2 T)$ bits for spike-count readout. Unipolar Binary Spiking (UBS) uses thresholds and spike-triggered residual decay to produce non-negative integer codes; Bipolar Binary Spiking (BBS) separates sign and magnitude and learns a scale for signed activations. These representations support accumulation-and-shift dot products and integer-exponent mappings in attention. With four time steps, Spora achieves 76.6 average GLUE score and 44.1 CoLA MCC, improving over SpikeLM by 1.2 and 6.2 points, respectively. Extending BBS to six steps raises these scores to 78.2 and 47.4. Conditional-decay analysis, matched-budget activation-quantization comparisons, event-workload statistics, and fixed-point evaluation further characterize the connection between encoding fidelity and computational cost.

Offline goal-conditioned reinforcement learning (GCRL) has produced a diverse set of goal-reaching algorithms, yet no single algorithm performs best across environments, goals, and even different phases of the same task. Rather than deploying only the best-performing policy, we ask whether a set of frozen goal-conditioned policies can be used collectively as a portfolio, deciding at every state which policy should act. Choosing a policy at each state is not straightforward. The policies' own value functions cannot be compared directly: they may use different scales, and some policies have no value function. We need to judge each policy by the states it is likely to reach, even though we can execute only one policy at a time. We also need to avoid switching so often that control becomes unstable. To address these challenges, we introduce World-Model Policy Arbiter (WMPA), a test-time framework that, given a set of frozen policies as input, rolls out each frozen policy in a learned state-space world model, evaluates the imagined futures with a shared goal-conditioned value function, and executes the highest-scoring policy for a short commitment interval before the next round of arbitration (policy selection). WMPA assumes access to a bank of frozen goal-conditioned policies and requires neither policy retraining nor privileged task-specific knowledge. Under the official OGBench evaluation protocol on 18 state-based datasets spanning maze navigation as well as cube, scene, and puzzle manipulation, WMPA improves the macro-average success rate from the 44% achieved by the best policy selected per dataset to 58%, with statistically significant gains on 12 datasets. These gains include +33 percentage points on cube-double-play and +36 percentage points on scene-play.

In memetic symbolic regression, candidate structures are compared after coefficient calibration, so the calibration protocol itself contributes to evolutionary selection. Standard centralized calibration evaluates each structure at its pooled-sample optimum, ignoring how stable this calibration is under covariate shifts, and can thus favor structures whose fit relies on sample-specific coefficients. We propose Dirichlet-Sinkhorn Constant Averaging (DSCA), a calibration strategy that partitions the optimization data into equally sized subsets with different covariate distributions, calibrates each candidate independently on every partition, and evaluates it at the mean of the resulting parameters. We show that the excess loss of DSCA relative to centralized calibration vanishes at the population level for correctly specified, identifiable expressions, whereas under misspecification it persists when partition-specific calibrations do not aggregate to the pooled optimum. On synthetic benchmarks and ten real-world datasets, DSCA improves functional recovery and the accuracy-complexity trade-off over centralized Broyden-Fletcher-Goldfarb-Shanno and Levenberg-Marquardt calibration, under selection by negative log-likelihood and by the Akaike and Bayesian information criteria. Mechanism analyses associate the DSCA excess loss with the generalization gap and show that the effect is not reproduced by repeated centralized fitting. These results indicate that controlled heterogeneous calibration provides a complementary source of selection pressure in symbolic-regression search.

We introduce a systematic way to use the subobject functor on any category to set up a framework for generalized topological and measure-theoretic notions and inferences. Concretely, we develop two connected calculi: 1. a fourfold calculus derived from notions of smallness and density that generalizes, among other things, topological density and measure-theoretic negligibility, 2. a twofold calculus that generalizes topological and measure-theoretic inference procedures. The resulting theory is distinguished in that it is inter-categorical in character: its definitions can be applied not just to describe smallness and density of subobjects in any category and classify morphisms by how they handle them, but also functors and natural transformations by how they transform them, somewhat like the theory of limits but with a simple and intuitive conceptual meaning that applies to each incarnation. We provide examples of these notions from many areas of mathematics and outline their unfolding throughout category theory as we add assumptions that bring them closer to classical settings, which we complete by generalizing the topological duality of open and closed subsets and outlining a possible framework for a categorical measure theory. Central to this interaction is the subobject 2-functor and the 2-category of upward-bounded preorders, so in the last part of the text we provide a study of their function. Concretely we derive representation theorems for this 2-functor and 2-category and a conjecture about how to generalize these theorems to n-categories.

Speculative decoding is a popular technique for increasing the speed and reducing the costs of large language model (LLM) inference by verifying multiple draft tokens in a single target-model forward pass. The resulting benefit depends on the ability of the drafter to approximate the target model's distribution. In this work, we study Speculative Rejection Attacks (SRAs), a novel class of attacks that cause draft and target models to disagree more often, resulting in fewer draft tokens being accepted per draft cycle. This leads to more target model forward passes needed per generated token, slowing down inference and increasing costs for the victim. We introduce two attacks which append an adversarial suffix to attacker-controlled content to degrade speculative decoding on a victim's prompts. Both attacks optimise the expected length of the accepted speculative prefix, estimating per-depth acceptance from the target's probability of the drafted proposals (Speedbump-P) or from the overlap between the draft and target distributions (Speedbump-D). In some cases, attacks degrade speculative decoding to the point of being slower than autoregressive decoding. The degradation reduces the output quality - regularisation restores output quality but gives up most of the degradation, trading effectiveness for stealthiness. Additionally, the suffixes remain effective under sampling, and transfer across drafters (Speedbump-P) or across target models sharing a drafter (Speedbump-D). These findings identify the draft-target interaction of speculative decoding as a realistic attack surface through which adversarial inputs can inflate inference costs.

We introduce the notion of a projection scheme on a Banach space and develop an abstract framework for studying the fixed point property (FPP) for nonexpansive mappings. Within this framework, we obtain a general criterion ensuring the FPP for a broad class of nonreflexive Banach spaces. In particular, every uniformly convex Banach space admits an isometric embedding into a nonreflexive Banach space having the FPP. This criterion unifies and extends several previously known fixed point results for nonreflexive Banach spaces with Schauder bases. As an application of our general result, we characterize the FPP for variable Lebesgue spaces. More precisely, we prove that if $(\Omega,\Sigma,\mu)$ is a $\sigma$-finite measure space and $p:\Omega \to [1,\infty]$ is a measurable function, then $L^{p(\cdot)}(\Omega)$ has the FPP if and only if it does not contain an isometric copy of $\ell_1$. Consequently, as in the classical Lebesgue space framework, the presence of an isometric copy of $\ell_1$ completely characterizes the failure of the FPP in variable Lebesgue spaces. This result yields a class of nonreflexive Banach spaces with the fixed point property that contains isometric copies of infinitely many $L^p$-spaces, $1<p<\infty$. Although nonreflexive sequence spaces with the FPP are known, to the best of our knowledge, these variable Lebesgue spaces provide the first examples of nonreflexive Banach function lattices with the FPP.

This paper introduces the class of weakly quasar-convex functions, a notion that simultaneously generalizes weak convexity (and hence convexity) and quasar-convexity. We establish the theoretical foundation for this new class by developing its basic calculus, deriving characterizations, and systematically investigating its closure and structural properties. Building on these tools, we analyze the proximal point algorithm for minimizing a weakly quasar-convex function under local quadratic growth, proving $Q$-linear convergence of the iterates and $R$-linear convergence of the function values to a global minimizer, with iteration complexity $O(\ln(\varepsilon^{-1}))$; we also give a fully computable, minimizer-free stopping rule via the proximal residual with matching complexity. Finally, we construct a weakly quasar-convex function, satisfying all our standing assumptions, that is neither weakly convex, quasi-convex, star quasi-convex, nor quasar-convex for any modulus, showing that our convergence theory applies to instances outside every prior framework in the literature.

Managing machine learning workloads as a network service introduces a resource-orchestration problem distinct from conventional model training; which nodes should be allocated to a task, how communication and computation budgets should be divided among them, and how service quality should be sustained as connectivity and node availability change with mobility. Deploying Generative Adversarial Networks (GANs) in Internet of Vehicles (IoV) environments is a demanding instance of this problem; resource constraints, dynamic network topologies, and competing optimization objectives mean that traditional GAN architectures cannot simultaneously achieve high accuracy, efficient resource use, low delay, and low communication overhead. This paper introduces an adaptive multi-discriminator Wasserstein GAN (MD-WGAN) framework that integrates reinforcement learning with game-theoretic coordination to address these challenges jointly. In our framework, roadside units host generators paired with Deep Q-Network (DQN) agents that select discriminator subsets and manage distributed training across mobile vehicular nodes, while a game-theoretic coordination step allocates training epochs between generators and discriminators. A unified optimization objective ties adversarial learning quality to resource efficiency, communication overhead, and latency under vehicular constraints, allowing the framework to continuously adapt its training behavior as network conditions change. Evaluation on real-world NGSIM trajectory data shows that the framework attains prediction accuracy comparable to state-of-the-art GAN baselines - the lowest RMSE (1.029) and MAE (0.894) among all evaluated methods - while markedly improving resource efficiency: average CPU utilization is reduced by roughly 28% and mean memory usage by roughly 6%, at competitive communication overhead and latency.

We develop two new general frameworks for constructing finitely generated residually finite groups that possess infinite girth, satisfy no non-trivial group law, and contain no non-abelian free subgroups. While such groups are rare–with known examples including the first Grigorchuk group, Thompson's group F, and certain iterated wreath products–our approach provides a more explicit construction of such groups realized as subgroups of direct products of symmetric groups. We investigate the relationship between the girth of a group and that of its finite quotients, leading to the introduction of a "marked" girth criterion, which we establish as a necessary and sufficient condition for infinite girth within the class of residually finite groups. Furthermore, we explore the connection between girth and profinite completions and provide a congruence realization result for infinite girth in linear groups. We conclude by posing several open questions that arise naturally from our results.

The Tensor-Memory Equilibrium (TME) model of "FP8 is All You Need (Part 1)" calculates the execution time of Ozaki Scheme II emulation of fp64 as the maximum of a tensor-core term and a High-Bandwidth Memory (HBM) traffic term, plus a per-output reconstruction term. However, it omits the per-input deconstruction cost: every streamed fp64 operand must be scaled, rounded, and reduced modulo each of the $r$ moduli on SIMT pipes before any matrix multiply can issue. In this note we add this fourth term, calibrate its constant from the cuBLAS emulation path, and derive a closed-form operational-intensity threshold $\mathrm{OI}^{*} = c_q r P_{\mathrm{fp64}}/(8P_{\mathrm{int}})$ below which emulation cannot match native fp64 regardless of tensor-core throughput. On the NVIDIA B300 GPU the threshold is $\mathrm{OI}^{*}\approx 0.56$ FLOP/B. As a result, GEMV, SpMV, and low-batch GEMV, which are the memory-bound kernels the original paper claims to accelerate, are limited to 0.3-0.9x of native performance, and the 7-point stencil to 1.8x rather than the claimed 3.1x. Dense GEMM is unaffected as expected. We also show that precomputing and storing the residues moves the same cost into the bandwidth term, and we state the instruction count that an implementation would have to achieve to invalidate the bound.

The logarithmic law of the wall is one of the most robust scaling laws of classical turbulence, yet whether it survives in a quantum fluid such as superfluid $^4$He (He II) remains unknown. At a solid wall, the viscous normal-fluid and inviscid superfluid components of He II obey fundamentally different boundary conditions, yet their motions can be coupled through quantized vortices. How these competing effects organize the near-wall flow remains an open question in quantum-fluid hydrodynamics. Here we visualize the near-wall velocity field in mechanically driven He II pipe flows at Reynolds numbers up to $1.26\times10^6$ using molecular tagging velocimetry and independently determine the wall shear stress from pressure-drop measurements. We find that the normal-fluid velocity develops a robust logarithmic profile, showing that a law-of-the-wall description persists despite the two-fluid nature of He II. Strikingly, however, the resulting wall law is distinctly nonclassical: the effective Kármán constant is less than half its classical value and remains nearly unchanged across the temperature range studied. This invariance, despite substantial changes in the relative fractions of the two components, indicates strong dynamical coupling between them. These results not only reveal how a universal law of classical turbulence is transformed by quantum hydrodynamics, but also provide valuable guidance for the design and operation of cryogenic cooling systems involving He II transport.

Motivated by the recent finding of Ji and Abbot (2025) that at moderate eccentricity (0.2–0.3) the planetary Snowball bistability vanishes, we have introduced stochasticity of the orbital eccentricity into a simplified version of the energy balance model of Eisenman and Wettlaufer (2009) to study a general Earth-like planet. The stochastic eccentricity is approximated by two dominant time scales, annual and daily, and is introduced through an analysis of the Kepler equation. We find that, apart from mean eccentricities less than 0.1, the stochastic dynamics are quantitatively and qualitatively different than the deterministic dynamics. For a fixed mean eccentricity, fluctuations in eccentricity amplify the asymmetry and intermittency in radiative flux forcing, thereby increasing the time it takes to reach a stable seasonal cycle. As the mean eccentricity increases above 0.1, the stochastic transition from the Snowball to non-Snowball state occurs over a range of mean stellar flux $\Delta S^*$ rather that at a particular stellar flux $S^*$ for the deterministic case. Moreover, as the mean eccentricity increases from 0.1 and the variance of the annual eccentricity fluctuations increases, $\Delta S^*$ increases and the probability of the system reaching a steady state and experiencing a Snowball to non-Snowball bifurcation decreases. This implies that there may be many high eccentricity planetary systems in a perpetually transient state.

Using the Mellin-Barnes representation of de Sitter propagators, we reduce tree-level heavy-particle exchange to finite integrals that converge at physical kinematics on mass-independent contours, and one-loop diagrams to the same integrand times a flat-space momentum factor. We evaluate the triple-exchange bispectrum without analytic continuation and compute the complete one-loop triangle at general mass. These give cosmological collider predictions, the oscillatory signatures of heavy particles during inflation, for multiple exchanges and loops at arbitrary masses.

E-commerce search pages are critical touchpoints for millions of online shoppers. While traditional search engines return a ranked list of results, modern E-commerce search pages increasingly incorporate recommender system modules – for example, secondary stacks that surface alternative product groupings at specific positions. When introduced appropriately, secondary stacks can improve user engagement; however, suboptimal placement may disrupt browsing flow and degrade the primary results. Unlike traditional search ranking, where evaluation techniques such as interleaving are well established, evaluating page-level layout changes e.g., when and where to insert a secondary stack remains challenging without costly online A/B testing. To address this, we study offline methods for evaluating whether a given layout decision – specifically, the inclusion of a secondary stack at a particular position – is beneficial to users. We investigate language models as scalable evaluators by comparing direct prompt-based, prompt-derived feature, and representation-based methods. Our results show that representation-based approaches consistently outperform prompt-based judging in predicting user engagement, suggesting they provide a reliable foundation for offline layout evaluation in E-commerce search.

Despite decades of data quality (DQ) research, a gap remains between DQ dimensions, such as accuracy or completeness, which the literature defines in textual form, and DQ tools, which typically implement low-level checks that are not aligned with these dimensions. ISO/IEC 25024 and ISO/IEC 5259 attempt to bridge this gap by defining DQ metrics for each dimension. However, these DQ metrics are hardly used, because the standards leave open how to implement them: for example, the metric for syntactic accuracy counts syntactically accurate values, but does not state how to decide that a value is syntactically accurate. This simply moves the problem to another level without solving it. As a result, DQ assessment currently cannot build on the standards. In this paper, we make the ISO DQ metrics executable. We classify all data-level metrics of both standards into (i) generalizable metrics that need no input beyond the data, (ii) parameterized metrics whose parameters can be learned from clean reference data or set by an expert, and (iii) non-generalizable metrics that need qualitative judgment and cannot be automated. For the metrics that can be automated, i.e., categories (i) and (ii), we propose implementation guidelines that resolve what the standards leave open. We realize the guidelines in dqmeasure, an open-source library of 20 metrics that learns these parameters from reference data instead of relying on manually defined rules. Our experiments on real-world and synthetic datasets show that the metric scores decrease monotonically with an increasing number of injected errors, decline together with downstream ML performance, and scale linearly with the number of rows, which enables automated DQ monitoring based on the standards.

Large Language Models (LLMs) are considered to be a more efficient and cost-effective alternative to human judgment for Machine Translation (MT) evaluation. With MT evaluation spanning a large number of language pairs, domains and levels of annotation granularity, LLMs must be thoroughly evaluated across these dimensions before being reliably used as alternatives to human evaluation. In this paper, we evaluate the performance of LLMs for two prominent MT quality evaluation schemes: Multidimensional Quality Metrics (MQM) and Error Span Annotation (ESA) by comparing their agreement with human annotators. We present results on a long-context test set of 70 language pairs and the publicly available WMT23 and WMT25 data, investigating both score and error span annotation agreement across a variety of language pairs and domains. Our results show that while LLM agreement with human annotators exceeds agreement between human annotators for some evaluation tasks, both vary substantially across annotation schemes, language pairs and domains and remain unreliable for most settings. Furthermore, we identify challenges facing both human and LLM annotators: humans are particularly challenged by fine-grained MQM annotations and low-resource language pairs, while LLMs struggle with minor errors, wrong language variants and error span annotation. Our results highlight the potential for improvement for both human and LLM annotation performance, possibly through human-LLM collaborative annotation pipelines that address the reliability issues identified in this work.

The radar cross section (RCS) is a fundamental parameter in radar engineering, as it determines the detectability of a given target by a radar system. Increasing interest in very low observability (VLO) targets has arisen recently. As a result, especial effort is put in developing numerical tools to accurately predict the electromagnetic scattering from perfect electric conductor objects. Integral equation methods are commonly used in this task. However, full-wave computation of the monostatic RCS at a single angle can be rather time-consuming, nevermind computing the RCS in a specific angular interval with a fine sampling. In fact, this is what is needed in industrial applications. A reduced-order model for fast full-wave monostatic RCS evaluation is proposed to easily achieve fine details in the desired angular domain. An open source integral equation solver, taken into account as a black box, is used to get the electromagnetic scattering in non-penetrable objects at some specific incident angles. Several radar targets will show the possibilities and capabilities of these model order reduction approaches.