Papers

We present JWST NIRSpec and MIRI observations of the metal-poor T dwarf WISEA J155349.96+693355.2. The combined NIRSpec/prism plus MIRI/LRS spectrum (R ~ 100) provides 0.6-12 $\mu$m spectroscopic coverage that reveals the presence of H$_2$O, CH$_4$, and NH$_3$ absorption features, enhanced collision-induced H$_2$ absorption, and weak or absent CO and CO$_2$ absorption, all indicators of a metal-poor low-temperature atmosphere. Comparison of the spectral data to a suite of atmosphere models yields an effective temperature $T_{\rm eff}$ = 960$^{+28}_{-32}$ K and a solar-scaled metallicity of [M/H] = -1.16$^{+0.28}_{-0.25}$ dex. We find indications that the uncertain trigonometric parallax, likely subject to Lutz-Kelker bias, may be overestimated. Adopting the photometric distance yields a bolometric luminosity log$_{10}L_{bol}/L_{\odot}$ = -5.29 $\pm$ 0.10, from which we infer $T_{\rm eff}$ = 974 $\pm$ 56 K, consistent with the temperature derived via our atmospheric model fits. From the moderate-resolution (R ~ 3,000) NIRSpec/G395H spectrum, we measure a radial velocity of -163 $\pm$ 5 km/s, which combined with the source's measured proper motion and estimated distance indicates a highly eccentric (e $\geq$ 0.9) Galactic orbit consistent with an accreted halo object. The UVW kinematics of WISEA J155349.96+693355.2, its bulk metallicity, and its modest alpha enrichment ([$\alpha$/Fe] = +0.15$^{+0.04}_{-0.06}$ dex) are all equivalent to stars identified in the Gaia-Enceladus Milky Way substructure, making it the first brown dwarf with kinematic and chemical evidence of association.

Speculative decoding accelerates Large Language Model (LLM) inference by using a lightweight draft model to propose candidate tokens for parallel verification by a target model. Drafting and verification, however, exhibit different service characteristics and favor different batch configurations, making fixed draft-target coupling inefficient under concurrent workloads. Existing distributed designs can physically separate the two stages, but often retain request or batch affinities that prevent their capacities from being shared globally. We present NebulaSD, a many-for-many, or M-for-N, speculative decoding system that organizes draft and target workers into independently schedulable resource pools and dynamically reconstructs stage-specific batches from shared request pools. Such dynamic reassignment removes fixed worker locality, requiring request states to be made available at newly selected workers without introducing migration stalls. NebulaSD addresses this challenge through worker-triggered batch reconstruction and asynchronous KV-state preparation overlapped with model execution. We evaluate NebulaSD from both system and scaling perspectives, showing that dynamic pooling improves request-round processing rate by 50.4% over a physically disaggregated baseline and 72.6% over co-located execution on a four-GPU deployment while substantially increasing effective GPU utilization. Profile-driven simulations further show approximately proportional compute-side capacity scaling under idealized state movement.

The public visibility of Artificial Intelligence (AI) is growing rapidly, driven by the positive impact of its applications across diverse fields of knowledge. In this new chapter, courses that cover the foundations of AI and machine learning become essential for understanding their role and potential in contemporary society. Therefore, understanding fundamental concepts and elementary algorithms through the close integration of theory with practice is essential in AI courses. In this essay, we report our experience designing machine learning exercises for automated assessment tools in programming. It is worth mentioning that we are not developing a novel form of automated grading system. Instead, we propose a perspective that frames machine learning problems as input-output assessment tasks. From this perspective, each exercise admits a unique and deterministic answer and enables automated programming assessment tools (e.g., VPL for Moodle, Codeforces, and MOJ) to effectively support AI education. We believe this essay can encourage instructors to foster educational innovation by adopting more dynamic and interactive approaches to AI courses that integrate theory and practice. Importantly, this essay does not introduce an innovation in the use of AI for education; rather, it introduces an innovative approach to improving the learning of AI, particularly, machine learning.

Sparse AutoEncoders (SAEs) offer a promising way to inspect language model representations, but it is still unclear what kind of linguistic structure their latents expose. We use part-of-speech (PoS) categories as a controlled test case to study whether morpho-syntactic information is encoded by individual latents or by structured groups of features. We find that PoS distinctions are highly recoverable from SAE activations, but do not align with one-to-one latent / category mappings. This recoverability is not reducible to lexical memorisation, and Open and Closed PoS classes differ substantially. Categories are supported by compact groups of sparse latents, with substantial variation across tags. These groups remain stable on held-out data, while also showing overlap between related categories. Our results show that SAEs localise morpho-syntactic information in a distributed and category-dependent form rather than through atomic grammatical features.

Predicting optoelectronic response across thermodynamic state space requires coupling finite-temperature nuclear dynamics to electronic structure at scales where direct first-principles calculations are impractical. Machine-learning force fields and Hamiltonian-learning models provide scalable predictions, but integrating them into reliable and interpretable workflows remains challenging. Here, we introduce FLOW-OTTER, a modular, model-agnostic framework that automates molecular dynamics, Hamiltonian prediction, observable extraction, reliability assessment, and Hamiltonian-level interpretation. We demonstrate FLOW-OTTER in halide perovskites, soft semiconductors whose anharmonic fluctuations strongly modulate electronic response. Using FLOW-OTTER, we show that independently learned nuclear and electronic models remain predictive when composed end-to-end, reproducing experimental temperature- and pressure-dependent band-gap trends using models trained only on first-principles targets at zero pressure. By resolving the nonlinear evolution of Pb-$s$/Br-$p$ antibonding at the valence-band maximum, it identifies the microscopic origin of the asymmetric pressure response. FLOW-OTTER thus establishes Hamiltonian learning as a general route from thermodynamic trajectories to experimentally grounded optoelectronic mechanisms.

Let $n$ be an integer with $n\ge2$, and let $E$ be a real algebraic number field of degree $n+1$ over ${\mathbb Q}$. Let $\alpha_1, \ldots , \alpha_n$ be real numbers in $E$ such that $(1,\alpha_1,\ldots,\alpha_n)$ is a linear basis of $E$ over ${\mathbb Q}$. Let $\nu_1, \ldots, \nu_{n-1}$ be real numbers satisfying $$0<\max_{1\le i\le n-1}\nu_i\le2\min_{1\le i\le n-1}\nu_i,\quad \nu_1+ \ldots +\nu_{n-1}=1. $$ We establish that there exist a real number $C$, depending only on $\alpha_1, \ldots , \alpha_n$, and infinitely many integers $Q \ge 2$ satisfying the inequalities $$Q^{1/n}\Vert Q\alpha_i\Vert \le C (\log Q)^{-\nu_i}, \quad 1\le i\le n-1, \quad Q^{1/n}\Vert Q\alpha_n\Vert\le C.$$ This answers partially a conjecture of Peck, who proved in 1961 this statement in the particular case where $\nu_1 = \ldots = \nu_{n-1} = 1/ (n-1)$. We also improve a result from 2004 of de Mathan and Teulié on a question of simultaneous Diophantine approximation involving a non-Archimedean valuation.

We prove that every complete nonsteady gradient Kähler-Ricci soliton with constant scalar curvature is rigid. This establishes Cao's rigidity conjecture for Kähler-Ricci solitons in arbitrary dimension and gives the corresponding result for expanding solitons, without additional curvature assumptions. The proof uses a rigidity criterion for gradient Ricci solitons with constant scalar curvature, expressed by the vanishing of the Lie derivative of the Ricci tensor along the soliton vector field. In the Kähler case, this vanishing follows from constant scalar curvature, the closedness of the Ricci form, and the classical real holomorphicity of the soliton vector field.

Vision-language models such as CLIP achieve strong zero-shot classification, yet under distribution shift, visual embeddings drift from fixed text embeddings. Training-free calibration avoids the per-sample optimization of prompt learning, but prior feature calibration gives each image the full bias of one hard cluster. We propose Domain Recentering with Confidence Calibration (DRC), a training-free method adapting CLIP from a set of unlabeled target images. DRC fits a Gaussian mixture once and subtracts from each embedding a posterior-weighted average of component means. It then removes residual class preference with a log-prior correction, estimating the prior from confidence-weighted predictions. Among compared methods, DRC achieves the highest average accuracy on cross-domain datasets, exceeding zero-shot CLIP by 4.13 and 5.07 points with ViT-B/16 and ResNet-50, with gains over CLIP also holding under ImageNet distribution shifts.

Quantum incommensurate systems have attracted widespread interest due to their unique physical properties. Related studies have made notable progress in recent years. To gain deeper insight, it is both significant and challenging to study a broader range of physical observables for these systems, which requires a comprehensive understanding of their spectral properties and wavefunction behavior. Based on the regularized model recently proposed, this work introduces a regularization framework, rendering physical observables for incommensurate systems mathematically well-defined and computationally accessible with theoretical guarantees. Based on this framework, taking the density of states and the electron density as representative examples, we establish their mathematical characterization, derive their planewave approximations, and provide a rigorous convergence analysis. Numerical experiments validate the effectiveness of our method, demonstrating its practical applicability for generic 1D and 2D incommensurate systems.

We present a compact all-multiplicity formula for tree-level late-time wavefunction coefficients of scalar QED in four-dimensional de Sitter space, with a conformally coupled charged scalar pair and positive-helicity photons. The result admits both a recursive representation in terms of shifted lower-point coefficients and an explicit finite sum involving only kinematic factors. The flat-space amplitudes vanish for two or more photons, yet the de Sitter coefficients remain nonzero. These zeros enforce cancellations of the total-energy residue and, through factorization, of proper partial-energy residues involving multi-photon subamplitudes. Using a common-reference construction, we derive an exact lower-point recursion from Abelian Ward identities. Conformal symmetry provides an independent check: the partition sum solves the inhomogeneous Ward identity, while the total-energy term is a homogeneous solution. We also outline an analogous recursion in massless fermionic QED.

Configuration space~(C-space) of a mechanism is a real variety describing the set of feasible configurations that it can attain. To understand the behavior of a mechanism, it is crucial to identify and scrutinize especially the singular points of its C-space. They usually appear when the variety intersects itself, leading to different branches of motion. There exist many approaches to detect those intersections if they are transversal. However, the problem remains challenging if there are tangential, cuspidal, inter-dimensional or a combination of these intersections. This paper exploits an approach acquired from tropical geometry to analyze the neighborhood of any point on C-spaces of 1-degree-of-freedom~(dof) mechanisms. This is done by finding the approximate rational parametrization of the curve(s) passing through the given point using Puiseux series. The proposed approach is shown to succesfully detect the transversal branchings in two foldable four bar mechanisms and a cusp in the configuration curve of the double Watt mechanism.

The structurally complex tetragonal sigma ($\sigma$)-phase provides a unique platform for investigating the interplay between chemical disorder, electronic structure, and superconductivity, particularly in Re-based alloys where unconventional superconductivity has been widely discussed. Here, we synthesize and investigate the superconducting properties of $\sigma$-phase Re-X (X = V, Nb, and Ta) alloys with compositions Re$_{0.76}$V$_{0.24}$, Re$_{0.56}$Nb$_{0.44}$, and Re$_{0.60}$Ta$_{0.40}$. These compositions lie within the narrow stability range of the $\sigma$ phase, underscoring the critical role of valence electron concentration in phase formation. To examine the influence of enhanced chemical disorder, we further synthesize the structurally complex medium-entropy alloy Re$_{0.56}$Nb$_{0.19}$Ta$_{0.19}$V$_{0.06}$, derived from these binary systems. Magnetization, electrical resistivity, and specific-heat measurements establish bulk type-II superconductivity in all compounds. Analysis of the electronic heat capacity is consistent with a fully gapped, weak-coupling BCS superconducting state. In contrast, the normal-state resistivity exhibits an unconventional negative temperature coefficient, and the superconducting transition temperature values obtained from resistivity measurements are higher than those determined from other measurements. Our results demonstrate that both the $\sigma$-phase Re-X alloys, spanning 3d, 4d, and 5d transition-metal substitutions, and their medium-entropy counterpart constitute an attractive family of model systems for investigating the effects of structural complexity, chemical disorder, and spin-orbit coupling on superconductivity in Re-based materials.

We consider a hierarchy of models that can be obtained from the original Nyzhnyk system by imposing conditions on parameters, introducing potentials or pseudopotentials, performing limiting processes with respect to a scaling parameter, applying differential substitutions and interpreting parts of independent and/or dependent variables as complex or real. Therefore, among the Nyzhnyk models, one can distinguish between symmetric and asymmetric, dispersive and dispersionless, standard and modified, as well as real, complex, mixed and specific models. One can also distinguish single partial differential equations or systems of such equations, as well as linear or nonlinear Lax representations in the dispersive or dispersionless cases, respectively. For each specified model, we compute the maximal Lie invariance pseudoalgebra and, in the symmetric case, find the point- and contact-symmetry pseudogroups using the megaideal-based version of the algebraic method. It is shown that relations between these pseudoalgebras and between these pseudogroups are induced by relations between the corresponding models. Defining (finite-dimensional) subalgebras are singled out in all these pseudoalgebras in the symmetric and specific cases. Based on the established correspondences between the dispersionless and dispersive models, we completely classify one- and two-dimensional subalgebras of the maximal Lie invariance pseudoalgebra of the (dispersive symmetric potential) Nyzhnyk equation and one-dimensional subalgebras of the maximal Lie invariance pseudoalgebra of its linear Lax representation.

Integrated codes simulating interactions between Alfvén waves and fast ions in tokamak plasmas use perturbative models for the relatively slow processes of instability growth, saturation, chirping and bursting, and transport. Faster processes by which an Alfvén mode's spatiotemporal structure forms are assumed to have been completed within the mode's oscillation period, $\tau_0 \equiv 2\pi/\omega_0$. This separation of time scales underlies the computational efficiency of perturbative models, where the Alfvén mode's time-dependence is reduced to that of a scalar signal $s(t) = A(t)\sin(-\omega_0 t - \phi(t))$ with variable amplitude $A(t)$ and phase $\phi(t)$. For this, accurate input data in the form of a mode's spatial structure $\delta\Phi({\mathbf x})$, damping rate $\gamma_{\rm d}$, and initial frequency $\omega_0$ are required. For modes residing in dense or continuous spectra, $\delta\Phi$ and $\gamma_{\rm d}$ could be estimated from the form of the continua and fast ion orbits, but it is difficult to guess the seed frequency $\omega_0$. Here, we report results of numerical experiments showing that it is possible to find $\omega_0$ using a prompt frequency shift that occurs during the first few $100$ time steps of a simulation. Restarts with the shifted frequency iteratively converge to a value of $\omega_0$ that seems to maximize the resonant drive, suggesting an auto-optimization process. The need for iteration is attributed to the fact that the terms required for rapid frequency adjustments were truncated when deriving the perturbative model. Meanwhile, the fact that partial auto-optimization is possible at all is attributed to the fact that series truncation alone (without filter) does not strictly enforce slowness. Remnants of and cross-talk with faster dynamics still occur in numerical implementations. This entails potential for both uncertainty and utility.

A forest in an edge-colored graph is rainbow if its edges have pairwise distinct colors. We prove that for every $\varepsilon>0$ and all sufficiently large integers $m$, every properly edge-colored simple graph with $km$ edges, where $1\leq k\leq 2m$ and every color class has size at most $m$, contains at least $(1-\varepsilon)m$ pairwise edge-disjoint rainbow forests, each with exactly $k$ edges. The range $k\leq 2m$ is best possible: for every $k>2m$ there are such graphs containing no $k$-edge forest. Thus the conjecture of Montgomery, Pokrovskiy, and Sudakov fails beyond this range, while our theorem establishes its predicted conclusion throughout the largest possible range of $k$. The number of forests is asymptotically optimal. The proof uses an orientation dichotomy, hypergraph matching, matroid intersection, and martingale concentration.

We introduce Flow-Based Distribution Matching (FBDM), which learns self-supervised representations through non-adversarial velocity regression. The encoder and auxiliary field are trained jointly by regressing analytic velocities along spherical paths from learned representations to reference targets. Both augmented views share a reference target, with a capacity limit on each reference center and an explicit view-alignment loss. A structured reference supports more components than the auxiliary flow dimension while retaining geometric separation. The auxiliary field is discarded after pretraining; only the backbone is retained downstream. On three ResNet-18 benchmarks with matched auxiliary parameter counts, FBDM provides 2.17 to 2.36 times higher per-epoch training throughput than critic-based Distribution Matching (DM), with about 0.21 GiB additional peak allocated GPU memory. Experiments show that FBDM is competitive with existing SSL methods from CIFAR to ImageNet and performs comparably to DM on their shared benchmarks. Under stated conditions, our population analysis bounds nearest-centroid classification error on the initial normalized pretraining representation by the FBDM loss.

ArGuard is a shared task on harmful content detection in Arabic memes and LLM prompts. It includes two tracks: Track A focuses on multimodal hate detection in Arabic memes, while Track B addresses harmful prompt detection for Arabic LLM safety evaluation. In total, 58 teams registered, 35 participated in the final evaluation, and 27 submitted system-description papers. Participating teams explored models such as AraBERT, Jais, and Qwen3-VL. The best systems achieved macro-F1 scores of 0.823 on A1, 0.419 on A2, 0.984 on B1, and 0.790 on B2. Fine-grained meme classification in A2 was the most challenging setting, partly due to sparse labels and train-test distribution shifts.

A quantum measurement transforms a quantum statistical experiment into a classical one, but different measurements generally yield experiments with different statistical information. We introduce quantum-to-classical sufficiency: relative to a prescribed measurement class, a measurement is sufficient if its induced classical experiment Blackwell-dominates those from every other measurement in the class. This separates two components of quantum inference: reducing the measurement class through symmetry or decision-theoretic arguments, and identifying a canonical measurement in the reduced class by an experiment-level sufficiency condition. We establish a factorization criterion where the parameter enters the group-averaged experiment only through weights of an orthogonal block decomposition. The block-label measurement is then sufficient relative to all invariant measurements. For tensor-product models under unitary conjugation, this identifies weak Schur sampling as the sufficient measurement for spectral inference. Combined with risk-preserving symmetrization, it reduces invariant Bayes and minimax problems over all measurements exactly to classical decision problems based on Young diagrams. We derive sharp first-order asymptotic Bayes and minimax risks for smooth spectral functionals. We establish an analogous reduction for a noncompact group action in a continuous-variable setting. We consider displaced thermal states and functionals of the thermal parameter with an unknown displacement. Within displacement-invariant measurements, total residual photon number measurement emerges as the sufficient measurement. Combined with Hunt-Stein reduction, this yields exact finite-sample reductions of Bayes-minimax and minimax problems over all measurements to classical decision problems based on the total residual photon count, which follows a negative binomial distribution.

Event cameras provide a high dynamic range and preserve brightness-change cues in lighting conditions where conventional RGB frames may be noisy or saturated. To benchmark event-guided restoration across a broad illumination range, we organized the SEE Challenge 2026 with the Event-Based Multimodal Vision Workshop at ECCV 2026. The task conditions restoration on one or more RGB frames, synchronized events, and a scalar target-brightness statistic provided by the organizers. It uses SEE-600K, which contains 610,126 image-event observations from 202 real-world scenes spanning low-light, normal-light, and high-light conditions with illumination variations of up to 1,000$\times$. The challenge follows an open-system protocol: participants may use different temporal contexts, architectures, pretrained weights, test-time augmentation, and post-processing strategies. PSNR determines the ranking, and SSIM is reported as a secondary metric. Around 70 teams registered interest and 15 valid CodaBench submissions were received. Six distinct teams completed organizer-side identity and technical verification, provided method descriptions, checkpoints, inference code, and instructions, and are included in the verified open-system ranking reported here. Beyond the ranking, this report analyzes exposure subsets, semantically distinct test cases, a shared failure pattern, system design choices, and inference strategies. The top systems obtain closely spaced average scores, while the best-performing method varies across cases and metrics; under severe underexposure, all verified systems retain visible local errors.

We develop a three-sector model in which Marx's aggregate equalities (total price equals total value; total profit equals total surplus value) restrict a sequence of fully realized states during capital reallocation and technical diffusion. Anchoring one sector's profit rate to the contemporaneous average selects a trajectory without modeling pricing intentions. Before innovation, capital inflow lowers the receiving sector's price and excess profitability along calibrated stable paths, without a demand function. Stability is predictable from value and technical data; near a stable equalization, the receiver's profit rate falls even where its price rises. A labor-saving, capital-intensifying innovation creates temporary extra surplus value and revalues committed capital. At constant real wages the final uniform rate rises, as Okishio predicts; when rising wages keep exploitation constant, it falls.

Enterprise governance requires decisions, evidence, and accountable authority; it does not require every review task to retain its current human implementation. We develop a task-substitution framework for Digital Governance Frameworks (DGF), treating each gate as an executable contract. Substitution requires sufficient accessible information, valid decision and authority checks, and a reduction in total human work after exceptions, verification, correction, and maintenance are counted. We derive a residual-work threshold and show why automating most cases can still increase labor. Forward deployed engineering connects these conditions to an architecture for agents, rule engines, evidence services, and escalation. DGF-Bench supplies controlled evidence from 300 synthetic projects and 899 evaluable model-project runs. Gemini 3.8 Flash, GPT-5.6 Luna, and DeepSeek v4.1 Flash achieve strict gate success of 94.98%, 83.29%, and 74.18%; complete-route success is 76.92%, 42.33%, and 24.67%. A deterministic control passes all 1,700 gates given the supplied rules and structured facts, locating the comparison in execution of a supplied decision kernel. Evidence audits and 135 repeated runs distinguish correct decisions from reliable execution. A document counterexample establishes an information-sufficiency obstruction. These results support the technical feasibility of replacing human execution of specified governance-review tasks with agents and software. The framework specifies a workforce test based on the complete human effort required at fixed output and quality; the present measurements concern review performance. Sources, dossiers, traces, and analyses are public.

Elemental substitution within existing structural frameworks is a widely applied strategy for developing advanced materials. Yet, optimizing target properties while maintaining phase purity usually demands extensive trial-and-error, which becomes substantially inefficient when navigating a complex design space. Here, we introduce a strategy that simultaneously and dynamically assesses composition-dependent synthetic accessibility and target properties via aggregated cost functions that guide autonomous experimentation in a truly self-driving and self-learning mode. Specifically, we developed a cost-guided autonomous solid-state synthesis (CASS) framework and demonstrate its application in the discovery of Na superionic conductor (NASICON) solid electrolytes. CASS successfully optimizes ionic conductivity and phase purity, leading to the identification of 18 promising compositions in 78 trials conducted in an autonomous laboratory, the A-Lab. Among these, we identified two fast-conducting NASICONs yielding total (bulk) ionic conductivity of 0.7 (3.96) and 0.3 (3.17) mS/cm. The successful deployment of CASS reinforces the potential of coupling self-driving autonomous laboratories with physics-informed generative models to accelerate materials discovery. Moreover, the interpretability of the design factors, informed by outcomes from both successful and failed syntheses, enables model refinement and generation of new chemical insights.

Speech LLMs are costly to deploy in resource-constrained settings. Layer pruning can cut this cost, but existing scoring metrics transfer poorly to speech LLMs: they assume a decoder-only architecture with homogeneous token sequences, whereas speech LLMs add encoder and adapter components and process multimodal sequences. We propose Speech Block Influence (SBI), the first layer-importance scoring framework designed for speech LLM pruning that consists of two component-specific scores: SBI-Enc measures the effect of encoder-layer removal at the adapter's output to better reflect downstream impact; SBI-Dec measures layer-wise input-output similarity over text-token positions only to avoid audio-token dominance. Across three speech LLMs, SBI improves pruning robustness, with stronger encoder performance at higher pruning rates and more reliable decoder layer selection by scoring text tokens rather than the audio-dominated full sequence. We further find that text-only calibration yields decoder rankings highly correlated with those from speech-text calibration, suggesting a cheaper alternative to measure decoder layer importance.

Future radio access networks are expected to rely on large-aperture antenna arrays, for which an increasing portion of the coverage region may fall within the radiative near-field. In this regime, conventional far-field models become inaccurate, and the received spatial signature depends jointly on the propagation range and azimuth, making reliable uplink channel acquisition a key physical-layer challenge. Many existing near-field estimation works rely on simplified line-of-sight-dominant or single-path channel models, which fail to capture practical non-line-of-sight (NLoS) multipath environments. In contrast to prior single-path near-field partial relaxation (PR) formulations, in this paper, we develop a PR-based framework for near-field NLoS uplink channel estimation by modeling each user channel as a superposition of multiple spherical-wave components characterized by their ranges and azimuths. For the known-pilot case, we develop a greedy PR-based maximum likelihood estimator that iteratively extracts dominant propagation paths while mitigating multi-user interference. For the unknown-symbol case, we propose a PR-based rank-adaptive covariance-fitting approach that captures the multipath structure. We further derive the corresponding Cramér-Rao bounds for both cases. Numerical results show that the proposed multipath PR-based methods achieve strong estimation performance, remain close to the corresponding bounds, and outperform near-field two-dimensional multiple signal classification across the considered scenarios, supporting their relevance for future large-aperture uplink systems.

Consumer-facing skincare AI must coordinate visual evidence, product information, tool use, and user-facing actions within explicit evidence and safety boundaries. This study evaluates SkinAgent AI, a non-diagnostic multimodal framework that combines visual concern routing with grounded and auditable LLM-based orchestration. The architecture includes routing for Acne, Pores, and Wrinkles; photograph-based skin-type estimation; count-informed ordinal acne-severity support; typed tools; database-grounded recommendation and action functions; deterministic safety, privacy, and evidence checks; approval before state-changing actions; and structured trace and replay mechanisms. Visual-model performance and system-level agent behavior were evaluated separately. Across three seeds, the skin-condition routing model achieved 99.84% +/- 0.07% accuracy. Skin-type estimation achieved 88.85% accuracy, while count-informed acne-severity support achieved 84.59% accuracy with a quadratic weighted kappa of 0.9076. On a locked but non-independent 240-case system benchmark, intent accuracy was 80.00%, exact tool-set match was 62.92%, and strict task completion was 47.08%. No violations or successful cross-user leakage events were observed in the finite safety and privacy test suites. Tool-selection errors, incomplete grounding of product attributes, and unreliable failure fallback nevertheless remained. These findings support the feasibility of bounded, database-grounded, and traceable agent orchestration for non-diagnostic skincare assistance. They do not establish clinical readiness, external generalization, formal privacy guarantees, or universal safety. Independent validation, expert assessment, robustness and fairness testing, and prospective evaluation in real-world settings remain necessary.

This paper presents a grasping model for cloth manipulation specifically tailored to ease the deployment of robotic control methods. The model is robust, fast and easy to implement avoiding at the same time contact and friction considerations between the gripper and the cloth in favor of simple positional constraints. The gripper is described by its pose, jaw state, and an attached grasping volume. Two kinds of grasping volumes are considered: an axis-aligned box to simulate a pinch grasping and a square pyramidal volume to simulate point grasping. When the gripper closes, the discrete cloth positions lying inside this volume are selected, stored in the local gripper frame, and then transported with the gripper motion. A simple squeezing step is also included to progressively move the selected cloth positions toward the center of the grasping region, avoiding an instantaneous displacement at closure. The model can be used in any simulator as it only requires access to discrete cloth positions and a mechanism for imposing target positions as constraints. We implement our grasping model in conjunction with a constraint-based inextensible cloth simulator, where grasping is implemented as moving positional equality constraints coupled with stretch, shear, collision, and table contact projection steps. The same gripper trajectory is applied on a robot arm to fold a real piece of cloth, serving as a simple bridge between simulation and physical cloth manipulation and showcasing the realism and practicality of our idealized grasping model.

In this paper, we study a two-dimensional Poisson–Nernst–Planck–Navier–Stokes system in a periodic channel driven by wall motion and an imposed tangential electric field, with two ionic species that may have unequal diffusivities. At the electroneutral Couette state, a diffusivity-weighted ionic energy combines with the triangular structure of the linearized operator to yield exponential linear stability for fixed \((A,E_0)\) and all positive diffusivities. Small-data nonlinear exponential stability is then obtained on a complex interpolation space adapted to the graph domain of the generator by combining analytic-semigroup smoothing with quadratic estimates. For unequal diffusivities, the nonzero streamwise modes of the linearized ionic subsystem satisfy a bounded-channel enhanced-dissipation estimate at rate \(D_{\min}^{1/3}|A|^{2/3}\) under an explicit strong-shear condition. When \(D_+=D_-\), a Fourier-mode factorization separates the common Couette advection–diffusion operator from a contractive drift–reaction semigroup, so the same scalar mixing rate is retained without any smallness condition on the electrostatic coupling. We also construct an exact Poisson–Boltzmann/electroosmotic Couette family for prescribed wall potentials. For sufficiently small wall-potential amplitude, its linearized generator is a small graph-domain perturbation of the electroneutral generator, which yields linear and nonlinear exponential stability on the same interpolation scale.

Track reconstruction in drift chambers is essential for momentum measurement and particle identification at electron-positron colliders. While Transformer architectures have transformed many sequence-processing domains, their application to tracking in high energy physics is still being explored. We present a model that combines a BERT encoder with a Transformer decoder to perform hit-to-track association through autoregressive sorting. The model is evaluated on the DCTracks open dataset that provides realistic drift chamber simulations with varying particle types, momenta, track multiplicities, and noise conditions. Across single-track, two-track, and multi-track samples, the model achieves high hit and track efficiencies while keeping the rates of clones and fakes very low. It also works well in the reconstruction of displaced vertices. These results show BERT-based sequence-to-sequence models as a promising approach for track reconstruction in low-background, precision-oriented experiments.

Dynamical simulation of quantum many-body systems is a central task in quantum chemistry and requires efficient wavefunction representations. Although tensor-network and neural-network quantum states have achieved considerable success in ground-state calculations, entanglement growth hinders their application to quantum dynamics. Quantum computing may offer a route to quantum advantage, but algorithms such as Trotterization generally require fault-tolerant quantum computers, whereas near-term hardware supports only limited circuit sizes. Variational quantum simulation (VQS) instead represents the evolving state with a single shallow, fixed-size parameterized quantum circuit (PQC). Here, we introduce nonorthogonal variational quantum simulation (NOVQS), which applies linear combinations of parameterized quantum states to real- and imaginary-time evolutions. We design a shallow, hardware-friendly ansatz tailored to second-quantized electronic-structure Hamiltonians, together with resource-efficient protocols for measuring the matrices and vectors in the parameter equations of motion. Error analysis and resource estimation are also provided. Numerical simulations of hydrogen chains and the nitrogen molecule demonstrate that a collection of shallow, or even single-layer, PQCs can match or outperform a much deeper PQC in VQS. In particular, we identify a trade-off between circuit number and depth. NOVQS therefore provides a flexible route to enhancing wavefunction expressivity under circuit-depth constraints, making it a promising framework for quantum dynamics simulations on near-term quantum processors.

Quasilinear models make gyrokinetic turbulent-transport predictions sufficiently fast for integrated modelling, but their predictive capability is limited by two factors: the physical and geometrical applicability of the linear solver, and the validity of the saturation rule used to close the model. We present the Predictive Quasilinear Solver (PQLS), a quasi- linear gyrokinetic transport solver formulated in general magnetic geometry. Its implementation as an eigenvalue solver retains electromagnetic and collisional effects, provides access to dominant and subdominant modes and is differen- tiable with respect to all plasma parameters. Linear benchmarks against GENE reproduce the growth rates, frequencies, and eigenfunctions. We additionally formulate the saturation-rule closure as a Bayesian inference problem that distin- guishes uncertainty in its fitted coefficients from the residual model-form uncertainty. The approach is demonstrated by calibrating the SAT3 rule on PQLS quasilinear weights against published nonlinear CGYRO cases. In addition to improving the robustness of the calibration, the new method also quantifies the uncertainty in each of the fit coefficients. Such uncertainty is propagated through transport calculations to produce error-aware profiles that are compared to the ones obtained from the full gyrokinetic simulation, showing excellent agreement.