Papers

We study the optimization landscape for infinite-horizon discounted Markov decision processes (MDPs) with general state and action spaces under structured stationary policy classes. A general weighted policy-iteration approach to establishing global convergence guarantees for policy gradient methods requires closure under weighted policy improvement at every policy, a property that may fail even when the policy class contains an optimal policy. To address this issue, we propose weaker conditions that guarantee the absence of suboptimal stationary points and establish the Polyak–Lojasiewicz–Kurdyka (PLK) condition for the policy gradient objective with a finite concentrability coefficient. We also establish the PLK condition from a policy-improvement bound that holds at every state, without a concentrability assumption. Our general results encompass settings covered by the earlier framework when the common standing assumptions hold for the same policy class and parameter domain. We further verify our proposed conditions for two operations models: inventory systems with Markov-modulated demand and stochastic cash-balance problems. For both models, the Bellman equation yields approximate convexity of the Q-value functions in the action variable, with deviations controlled by the first-order stationarity measure. These estimates establish exponent-one and, under additional curvature assumptions, exponent-two PLK conditions, which, together with Lipschitz continuity of the policy gradient, imply an $\mathcal{O}(1/\epsilon)$ iteration complexity and linear convergence, respectively, for projected gradient descent using exact policy gradients. To the best of our knowledge, we provide the first non-asymptotic convergence rates for solving infinite-horizon discounted inventory systems with Markov-modulated demand and stochastic cash-balance problems using policy gradient methods.

Post-training plays a pivotal role in enhancing the reasoning capabilities and task-specific expertise of large language models (LLMs). Despite recent advances in post-training methods, such as Group Relative Policy Optimization (GRPO), their practical deployment remains impeded by training instability arising from the reliance on importance sampling. We introduce Group Variance Policy Optimization (GVPO), a novel post-training method that integrates the analytical solution of KL-constrained reward maximization into its gradient weighting scheme. This formulation provides an intuitive interpretation: GVPO's gradient corresponds to the mean squared error between the central distance of implicit rewards and that of actual rewards. GVPO offers two key advantages: (1) it guarantees a unique optimal solution, exactly to the KL-constrained reward maximization objective, and (2) it enables flexible sampling distributions without requiring importance sampling. Beyond general post-training, we show that GVPO naturally extends to on-policy distillation (OPD). Furthermore, GVPO enables the optimization of a broad family of extended OPD objectives, providing a principled foundation for diverse objective design. By unifying theoretical guarantees with practical adaptability, GVPO establishes a new paradigm for reliable and versatile LLM post-training and on-policy distillation.

We establish independence of $q$ as a necessary condition for coefficientwise nonnegativity of normalized principal specialization ratios of monic Macdonald polynomials. For arbitrary pairs of partitions, we characterize independence of $q$. For denominator partitions with largest and smallest parts differing by at least two, every ratio independent of $q$ is a power of $t$. We classify the nonnegative ratios under addition of maximal rows and columns of fixed height. For rows of length at least two, the classification is equivalent to rectangular complementation. For arbitrary initial partitions, the generating polynomials of successive row ratios at the smallest number of variables admissible for all stages have simple negative real zeros with lower bounds for the ratios of consecutive absolute zeros. We derive a reciprocity criterion, bounds for gamma coefficients, and the corresponding Jack limit. For generalized staircases, we determine shortest horizontal and conjugate vertical strip sequences and obtain terminating basic hypergeometric representations.

For a class of growth processes of Laplacian type in the plane, we suggest an interpretation as a “gradient descent” in the space of smooth closed curves. More precisely, we show that boundary of a growing domain moves along a gradient of a certain functional in the space of curves. In the simplest cases this functional is $\log (1/r)$, where $r$ is the external conformal radius of the growing domain.

Microwave breakdown driven by electron cyclotron (EC) waves provides a non-inductive route to plasma initiation in reactor-scale tokamaks. We introduce a three-dimensional Monte Carlo simulation that, for the first time, self-consistently treats nonlinear wave-particle interactions, atomic collisions, and guiding-center transport. The Monte Carlo simulation unveils the key role of parallel Brownian motion in the ionization avalanche mechanism. The predicted breakdown boundary is validated against KSTAR experiments. This work concludes that microwave gas breakdown will be successful under ITER-relevant conditions at a D$_2$ prefill pressure near 2 mPa with 1 MW of injected EC power.

We propose a Bell-measurement node for quantum repeaters based on a planar semiconductor microcavity operating in the strong-coupling regime. Cavity photons hybridize with quantum-well excitons to form polaritons combining properties of photons and matter quasiparticles. A control photon loaded into one polariton mode changes the polarization response seen by a subsequently incident target photon. This conditional rotation is governed by the interplay of self-induced Larmor precession triggered by spin-dependent exciton-exciton interactions and the polarization beats caused by the splitting of transverse-electric and transverse-magnetic cavity modes. We identify conditions of the experiment that enable implementation of a controlled-Z gate and allow to distinguish all four Bell states in the ideal limit. The one-sided scattering scheme provides a lower interaction threshold than a scalar Kerr reference under the same assumptions. At a selected operating point, the bandwidth-induced identification error scales as the fourth power of the pulse bandwidth in the narrow-band limit. An additional fixed input rotation reduces this error even further. We describe the entanglement swapping between remote memories and determine the minimum quality of the elementary links needed to obtain an entangled output.

Mean-variance portfolio optimization (MVO) is a central framework in data-driven asset management. A widely adopted approach is a two-stage framework that first predicts expected returns and then solves the optimization problem based on these predictions, with the predictive models trained by minimizing prediction errors. However, this objective of prediction is not aligned with the quality of the downstream portfolio decision. Decision-focused learning (DFL), which directly minimizes the downstream decision loss within the learning process, has thus emerged as a promising direction. However, existing DFL approaches to MVO rely on surrogate losses or constraint relaxations for tractability, creating a structural mismatch between predictive model training and the constrained MVO solved at evaluation. We propose a single-level optimization formulation that incorporates the Karush-Kuhn-Tucker (KKT) optimality conditions of the lower-level MVO into the upper-level learning problem. This formulation explicitly preserves the budget and short-sale constraints while remaining tractable for standard nonlinear optimization solvers. Rolling-window experiments on real-world ETF (Exchange Traded Funds) data across two asset universes with different correlation structures show that our method achieved the best performance on multiple investment metrics and also demonstrated performance improvement due to the proposed regularization.

Ransomware increasingly fragments its file operations across temporary and intermediate files, scattering the semantic context that links individual I/O events to an overarching encryption campaign. This fragmentation defeats existing detectors that reason over isolated file streams – whether pattern-based methods that match rigid event sequences or learning-based methods that require accumulating statistical evidence across many files. We present DEFEAT, a framework that reconstructs this fragmented, scattered context by grouping causally related file events into File Event Gadgets (FEGs), semantically coherent units that capture the full intent behind sequences of file operations spanning multiple dynamically created files. Unlike provenance graphs (system-wide causal graphs that record relationships among all OS entities, such as processes, files, sockets, and registry keys, across the entire system), FEGs are scoped to the file-operation context of a single user asset, enabling lightweight, targeted analysis without whole-system instrumentation. Each FEG is modelled as an attributed control flow graph (ACFG) and embedded via a graph neural network for unsupervised clustering, enabling analysts to label entire behavioural clusters rather than individual samples, reducing annotation effort by 94%. Evaluated on a corpus of 97,816,471 file I/O events spanning 67 ransomware families, DEFEAT achieves 99.2% detection accuracy and outperforms state-of-the-art methods including UNVEIL, RWGuard, and Peeler by 6.57 to 7.56%. The framework operates at the granularity of a single file encryption: because each ACFG represents exactly one FEG (one user asset context), a cluster label can be assigned as soon as the first file operation completes, enabling detection at the first encrypted file.

Zero-shot time-series forecasting (TSF) is often described as forecasting without target-specific parameter updates, but that training-status condition does not specify what evidence the system may use. A frozen language model prompted with serialized values, a time-series model pretrained on broad forecasting corpora, and a retrieval-augmented forecaster may all satisfy the no-update condition while drawing on different transferable evidence. This paper argues that zero-shot TSF should therefore be governed as an evidence-access claim. We propose a source-first taxonomy that separates three primary evidence sources—frozen LLM prior reuse, parametric time-series pretraining, and retrieval-augmented external memory—from the architectures that implement them. After the source is identified, four additional audit questions remain: task interface, forecast object and scoring, prediction-time context, and resource budget. The resulting agenda is to make zero-shot leaderboards auditable by reporting evidence boundaries and interface assumptions alongside scores, so that benchmark progress reflects transferable forecasting capability rather than undisclosed changes in context, memory, or budget.

Few-shot semantic segmentation (FSS) of strip steel surface defects (S$^3$D) has posed significant challenges distinct from natural scenes. Unlike natural images, S$^3$D task exhibits unique characteristics including low local contrast, uneven illumination, and complex fine-grained texture patterns. Although recent methods based on Segment Anything Model (SAM) have shown promise in FSS on natural images by leveraging SAM's powerful pre-trained representations, these unique industrial characteristics of S$^3$D images lead to performance drop when directly applying SAM to industrial defect scenarios. In this paper, we propose a novel Perceptual Parallel Prompt (P$^3$) framework that empowers SAM, creating the P$^3$-SAM model to address these challenges through two core strategies. First, we develop a Perceptual-Optimized Encoding (POE) strategy that enhances local contrast and preserves critical texture details for S$^3$D segmentation. Second, we introduce the Parallel Prompt Generator (PPG) strategy that simultaneously generates both semantic and spatial prompts, enabling comprehensive guidance for SAM's decoder across varying images. Extensive experiments on three few-shot S$^3$D benchmarks demonstrate that P$^3$-SAM achieves state-of-the-art performance, with particularly notable improvements of 12.00% in mIoU on Surface Defects-4i dataset.

Online agent deployments accumulate execution trajectories at massive scale and behavioral diversity, for which predefined annotation criteria hardly exist. Extracting useful evidence therefore demands costly manual annotation or verifier signals that fail to scale, leaving valuable evidence buried among redundant, incomplete, and failed executions. This raises a question: without post-execution rewards or correctness labels, how can reusable experience be distilled from the trajectories themselves? To address this challenge, we introduce DENSE (Distilling Evidence from Nested Subtask Executions), which organizes trajectory-derived evidence into nested shortcut trees. By consolidating redundant attempts, identifying resolved subtasks, and retaining useful steps alongside outstanding requirements, DENSE transforms noisy execution traces into structured and reusable task-solving feedback. To evaluate whether such feedback helps agents retry the same task, we design REFIT, which measures success-rate changes between the initial attempt and feedback-guided retries. Among feedback methods without external outcome supervision, DENSE achieves the highest strict pass rate across four agent models on Terminal-Bench 2.1, improving over initial attempts by 7.12-21.81 percentage points with 19.0-43.6% fewer agent tokens on retries. In addition, on hard tasks DENSE consistently outperforms self-reflection in cumulative pass rate across multiple feedback iterations on all four models, demonstrating its strong potential for continual agent self-improvement.

Deep representation learning often selects hidden features and fits the final predictor on the same sample, so fixed-feature analysis performed after selection can omit selection cost. We study the conditional empirical Rademacher complexity of deep ReLU representations followed by bounded-norm predictors in additive or Lévy-Brownian RKHSs, termed Brownian heads. For a fixed representation, we derive an exact dual identity and sharp bounds in terms of activation mass, the average norm of the observed hidden vectors. Under same-sample selection, the representation supremum induces a quadratic Rademacher process. Brownian layer-cake and Gaussian-projection identities reduce it to coordinatewise or signed projected threshold traces, separating realized scale from selection complexity. For samples with pairwise-distinct inputs, explicit scalar ReLU families match the finite-trace and VC rates up to universal constants at the realized trace-and-envelope level. Induced-norm contraction also yields architecture-level bounds for rectangular, rank-deficient ReLU networks. Experiments verify the sharp bounds and rates, exhibit a selection gap at fixed activation mass, and assess the predictive feasibility of Brownian heads.

We investigate the stability and localization properties of Minkowski, de Sitter (dS), and anti-de Sitter (AdS) thick branes generated by a complex scalar field $\Phi$. The scalar self-interaction potential contains a temperature parameter $a$. As $a$ approaches its critical value, the imaginary part of $\Phi$ becomes double-kinked and the energy density splits into two peaks, indicating the formation of two sub-branes. The analysis of linear scalar, vector, and tensor perturbations shows no signs of instability in the scalar and vector sectors, while the factorization of the tensor perturbation equation excludes tachyonic tensor modes. The graviton zero mode is localized on the Minkowski and dS branes but is not normalizable on the AdS brane. We further study a real test scalar field $\Psi$ with a coupling function $F(R)$ between its kinetic term and the spacetime curvature. The scalar zero mode can always be localized on the Minkowski and dS branes, with a continuous gapless spectrum of scalar Kaluza–Klein (KK) modes. Massive scalar resonances emerge near the critical value of $a$ for suitable values of $\beta$. Increasing $\beta$ produces more resonances and prolongs the lifetimes of the lower resonances, while their time evolution exhibits the characteristic decay of metastable modes. For the AdS brane, the scalar zero mode is also localized, whereas the divergent effective potential at the boundaries traps the massive scalar KK modes as bound states, yielding a discrete mass spectrum.

We introduce and study the family of super adaptable monoids, which includes, among many others, all conical refinement monoids that are either finitely generated, or countable, primely generated, and regular. Our main result shows that any countable super adaptable monoid can be realized by a von Neumann regular ring, thus answering affirmatively the Realization problem for this class. The result is new even for the subclass of countable, primely generated, regular conical refinement monoids, which we show can be realized by both a von Neumann regular ring and a purely infinite, real rank zero C*-algebra. To prove this result, we develop a systematic framework that connects three fundamental structures: abstract monoids, separated graphs (a specific class of colored graphs), and von Neumann regular rings. By establishing functorial connections among these classes, we translate several algebraic properties of monoids into combinatorial and ring-theoretic notions. This allows us not only to prove the main realization result, but also to obtain several characterizations of super adaptable monoids. As a particular application, these can be used to provide a complete characterization of which regular monoids arise as graph monoids. To our knowledge, this work offers the first self-contained treatment in the literature where the complete realization process —from abstract monoids through I-systems and separated graphs to regular rings and C*-algebras— is developed sequentially and in full detail within a single text.

Software Bills of Materials (SBOMs) have emerged as a key mechanism for software supply chain governance in enterprise architectures. However, their adoption across organizations remains limited due to concerns about exposing sensitive dependency information. To address this limitation, we propose TrustBOM, a scalable architecture for confidentiality-preserving SBOMs integrated into enterprise CI/CD workflows. TrustBOM enables software providers to attest that specific vulnerabilities or restricted licenses are absent from their software without revealing the underlying dependency graph. This is achieved using zero-knowledge non-membership proofs, which are applied selectively based on consumer-defined policy constraints. The architecture ensures that proof generation scales linearly with the number of asserted constraints rather than with the size of the SBOM, enabling efficient operation in large-scale enterprise environments. Empirical evaluation demonstrates linear performance, with an average proof generation time of 0.9 seconds per constraint on commodity hardware, indicating the feasibility of deployment in enterprise platform ecosystems.

We determine the ultimate information rate of a relaxing multilevel quantum sensor under unrestricted adaptive control and construct an explicit strategy that attains it. We consider sensing a weak field that couples the ground state to several excited states that decay back to the ground state, a setting that reduces to amplitude-damping sensing for a single excited state. We show that the ultimate information rate is exactly the mean population lifetime of the bright state coupled to the ground state by the signal. Although the sensing dynamics generally involve several decay modes, an explicit rank-one protocol that monitors the ground–bright coherence using fresh meters and classical feedback attains this rate while keeping its finite-time information deficit bounded by a constant. For independently relaxing identical sensors, independent local monitoring attains the sum of their individual optimal rates, so intersensor entanglement and joint quantum error correction cannot increase the asymptotic rate. Consequently, the free population-decay curve provides an operational measure of the optimal sensing performance under arbitrary adaptive control.

We study real-analytic planar maps, that admit a real-analytic first integral with a non-degenerate extremum, in a neighbourhood of an elliptic fixed point. We prove that, if the map preserves an analytic area form, it is real-analytically and symplectically conjugate to a twist map, both in the resonant and in the non-resonant case. Using an example, we show that without the area-preservation assumption the conclusion fails. We then prove that a second, continuous, first integral is either a function of the first one or forces the map to be periodic, in which case the map is analytically (and, when area-preserving, symplectically) conjugate to its linear part. We illustrate our results, we study the McMillan map, and a planar map obtained from the Adler–Yamilov Yang–Baxter map.

Collaborative robots are increasingly deployed in industrial scenarios characterized by frequent product changeovers. As an intuitive programming method, kinesthetic teaching facilitates rapid robot deployment. However, users may overlook the configuration of the robot during kinesthetic teaching, leading to degradation in operational performance. Operational performance refers to the capability of the robot to generate motion and can be quantified by the Minimum Singular Value of the Jacobian matrix. To address this issue, this paper proposes an online dynamic force guidance method that integrates performance constraint and optimization mechanisms. Specifically, variable admittance control maintains the operational performance of the robot above a predefined threshold, while a virtual force actively guides the user to drag the robot towards configurations with improved performance. Experiments are conducted on a 6-DOF collaborative robot, comparing three typical paths in the task space. To evaluate the quality of the taught trajectories, trajectory playback experiments are conducted to analyze the relationship between the operational performance of the robot and the work efficiency. The results demonstrate that the proposed method effectively enhances the operational performance of the robot and consequently improves the work efficiency, holding significant value for reducing production takt time in industrial deployment.

The Lorenz system is the canonical low-order model of convective instability, yet its dissipative and driving terms have never been checked against, nor constructed from, an explicit thermodynamic bookkeeping. We derive a modification that satisfies the local-detailed-balance condition for macroscopic relaxation toward nonequilibrium steady states, thereby identifying the thermodynamic force, entropy-production rate and frenesy of the resulting flow. The resulting model produces a transition from a quiescent fixed point to a robust, large-amplitude relaxation oscillation, closely analogous to recharge-discharge oscillator paradigms used for the El Nino-Southern Oscillation. The system alternates between a long, nearly reversible recharge phase and a brief, violently frenetic discharge burst, during which essentially all of the cycle's activity and entropy production is concentrated.

Recent work has constructed Abelian Chern-Simons theories as categorical TQFTs, allowing us to naturally incorporate categorical defects and construct defect extensions of Abelian Chern-Simons TQFTs. We first identify the Turaev-Viro realizations of Abelian Chern-Simons theory in the center and doubled pointed modular cases, clarifying the distinction between single bulk realizations and canonical doubled ones. Then we obtain Turaev-Viro partition functions in terms of Reidemeister-torsion densities. Alternatively, the Alterfold construction supplies the associated topological boundaries, domain walls, and condensation sectors, establishing an explicit Alterfold/Chern-Simons dictionary. We show that the finite quadratic module is the invariant controlling the bulk theory, its topological symmetries, orientation-reversal invariance, and defects. We further show that multicomponent Abelian BF theory arises as the extended TQFT of an off-diagonal Abelian Chern-Simons theory, placing it naturally within the same extended framework. Finally, we demonstrate that recently proposed Abelian Chern-Simons dualities do not define a genuine TQFT duality. These results provide a concrete model for defects in Abelian topological orders and suggest a route toward the non-Abelian case.

Human mobility data has become an increasingly important component of urban analytics. Although the range of available mobility data sources has expanded substantially, access remains highly constrained by commercial restrictions, privacy concerns, and institutional barriers. Data protection procedures also often reduce the analytical value of released datasets. Synthetic mobility data has emerged as a promising solution, but existing methods differ substantially in their underlying mechanisms, the information they preserve, the outputs they generate, and the analytical questions they can support. Their comparative strengths and trade-offs remain insufficiently understood for urban analytics. This paper presents a structured review of synthetic human mobility data generation from an urban analytics perspective. We review the literature by methodological family and index it by the mobility outputs each family generates natively and the analytical capabilities those outputs enable. We first provide a taxonomy of synthetic data products, including population and persona representations, activity schedules, trip and tour records, trajectories, and aggregate mobility patterns. We then review the major methodological families, spanning mechanistic models, survey-driven population synthesis, activity- and agent-based simulation, deep generative models, transformer-based mobility language models, and LLM-agentic systems. Building on this synthesis, we introduce a Meaning-Population-Autonomy framework that characterises these methods along three dimensions: behavioural meaning, population grounding and scale, and generation autonomy. We consider these dimensions the principal requirements for downstream urban analytics. Few methods deliver behavioural meaning, population grounding and autonomous generation at once, and fewer still with generation constrained to feasible trajectories.

Medical image segmenters often get worse when sites, scanner vendors, or protocols change. Continual test-time adaptation (CTTA) addresses this problem without target labels, but it can be impossible to update a model on a non-stationary stream and can lead to a lot of errors. We examine a more reasonable and meaningful alternative: parameter-frozen inference enhancement(PIE). We use a source-trained segmenter that learns about anatomy-preserving scale and flip views, maps their predictions back to the native location, and averages the probabilities. We do not modify the weights of the model or the normalization statistics. On a cardiac MRI stream from M&Ms, which is trained on vendor A and evaluated sequentially on vendors B, C, and D, PIE has 0.7786 mean Dice, compared to 0.7680 for source-only inference and 0.7388–0.7416 for five other online-adaptation baselines. The controlled ablations show that performance saturates at 28 views, and confidence weighting, class-prior correction, connected-component filtering, morphological refinement, and inter-slice smoothing have no effect or cause negative transfer. Qualitative results on cardiac MRI and fundus images are also consistent with the frozen ensemble keeping thinner and nested anatomical structures. These results provide a strong, stable baseline for medical CTTA and expose an important failure mode: adaptation and handcrafted refinement can be less reliable than carefully designed inference.

In the Fermi-LAT gamma-ray light curve from January 1 to August 25, 2026, of the changing look blazar B2 1420+326, we report here the detection of transient quasi-periodic oscillations (QPOs) of period ranging around 10-13 d. The source entered the flare state around May 7, 2026, and lasted till around mid-August, 2026 during which it depicted strong QPO signatures. We utilized standard techniques, including weighted wavelet Z-transform analysis, generalized Lomb-Scargle periodogram, and phase dispersion minimization, to search for and confirm these findings. Prior to the flare, the source displayed a mild damped oscillatory feature of similar periodicity with a damping timescale of around 20 d. Based on the period timescale and nature of the QPOs detected, we discuss possible physical mechanisms for the observed salient features.

While underground reservoirs emerge as essential elements to face global warming, these systems represent complex multi-physical and multiscale problems. The considered injection of fluids during hydrogen storage, carbon dioxide sequestration, or geothermal energy recovery involves a modification of the chemical equilibrium of the fluid in the porous reservoir. Chemical reactions can induce microstructural changes of the rock matrix, leading to a reduction of elastic properties of the material, and to potential settlement or stress redistribution. Consequently, it becomes pivotal to establish predictive behavior laws to describe the effect of chemical damage on elastic properties. Facing the difficulties to estimate experimentally the impact of chemical damage on mechanical properties, a Digital Rock Physics approach is proposed in this contribution. This numerical homogenization scheme is used to compare two distinct types of microstructure models: the first one consists in a Discrete Element Model, while the second one employs a continuous description. This continuous formulation is based on a Phase-Field description to predict the evolution of the microstructure subjected to chemical alterations and on the Fast Fourier Transform to estimate the macroscopic properties of the material. Finally, these frameworks establish different softening laws that can be used as constitutive ingredients for a cemented material during its weathering.

Noncommutative resolutions of AS-Gorenstein isolated singularities are investigated by Li–Shen–Wu. However, establishing their existence and constructing such resolutions are generally difficult, even when they exist. In this paper, we study conditions under which a commonly graded AS-regular algebra serves as a noncommutative resolution of an AS-Gorenstein isolated singularity. We investigate projective modules over a noetherian commonly graded AS-regular algebra whose endomorphism rings admit resolutions by the underlying regular algebra. This leads to a more general definition of noncommutative resolutions of balanced Cohen–Macaulay isolated singularities. We show that the existence of such resolutions is equivalent to the existence of cluster tilting modules over balanced CM isolated singularities. The corresponding noncommutative analogue of the Bondal-Orlov conjecture is established in dimensions $2$ and $3$. As an application, we study Hopf actions on commonly graded AS-Gorenstein algebras and investigate noncommutative resolutions of invariant rings. We present three examples of noncommutative resolutions, including one in which the noncommutative isolated singularity is not connected graded.

Standard treatments of Earth matter effects in neutrino oscillation experiments assume that the coherent forward scattering potential experienced by propagating neutrinos is sourced entirely by the Standard Model electron density predicted by the Preliminary Reference Earth Model (PREM). We examine how an anomalous matter potential localized in Earth's inner core, motivated by gravitationally captured dark matter or resonant mixing of active neutrinos with a light (eV-scale or lighter) sterile state, would distort the vu survival and vutove appearance probabilities relative to the Standard Model PREM expectation. Using a full three-flavor matrix-exponentiation propagation code with a four-shell PREM density profile, we parametrize the new-physics contribution as a fractional enhancement ecore of the electron-sector potential confined to the inner core (r<1221 km) and compute oscillation probabilities for near-vertically upgoing atmospheric neutrinos (L=8000to12742km). Trajectories that do not cross the inner core (L<=12500km) are essentially insensitive to ecore, reproducing the Standard Model expectation to better than 0.1%. In contrast, near-antipodal trajectories crossing the inner core (L>=12505km, corresponding to nadir angles within roughly half a degree of the vertical) show significant effects. The muon-neutrino survival probability is displaced by tens of percent for ecore=0.05to0.15 and by more than 100% at particular energies for the {\nu}{\mu} to {\nu}e appearance channel, with a pronounced neutrinoantineutrino asymmetry that mimics, but is physically distinct from, intrinsic CP violation. These results indicate that percent-level new-physics potentials confined to the inner core are within the discovery reach of high-statistics atmospheric neutrino telescopes such as IceCube Upgrade, KM3NeT/ORCA, and next-generation megaton-scale detectors.

Quantization offers a practical path to deploying diffusion models with reduced memory and computation, but aggressive compression can cause quantized outputs to deviate substantially from their full-precision counterparts. Sampling-stage correction methods seek to compensate for such deviations during sampling, but existing approaches rely primarily on local information and underexploit trajectory history, limiting their ability to correct errors that propagate across timesteps. In this work, we formulate sampling with a quantized denoiser as an online estimation problem, using the history of quantized denoiser outputs to recover the underlying full-precision outputs required by the sampler. We propose QuAKE, a Quantization-Aware Kalman Estimator that combines a smooth trajectory prior with a conditional Gaussian observation model. At each sampling step, QuAKE recursively updates the posterior over the output window in closed form and feeds its posterior mean to the sampler. QuAKE is a lightweight plug-and-play corrector that requires no modification to the quantized network and naturally supports arbitrary high-order multistep ODE samplers. Experiments across W4A4-quantized text-to-image diffusion models show that QuAKE consistently outperforms existing methods in reducing the distributional discrepancy from full-precision sampling.

The relative velocity between a particle and the surrounding fluid is a key quantity that determines the various phenomena of particle-laden turbulence, such as particle clustering and turbulence modulation due to particles. We theoretically derive the expression for the relative velocity of a particle laden in turbulence by extending the argument of Balachandar (2009). This derivation provides a simplified form of the analytical model of Berk and Coletti (2024), and it makes the underlying physical assumptions explicit. Using direct numerical simulation data of turbulence in a periodic cube and turbulent channel flow, we demonstrate the validity of the derived expression, and show that it also holds for non-spherical particles and droplets.

We propose a method to overcome the difficulties encountered when applying reinforcement learning to fluid mixing processes. The proposed method has two main features: (i) it does not require detailed measurements of the flow state, and (ii) by effectively exploiting a stable limit cycle of a two-dimensional dynamical system (the Li'enard system), it can stably perform optimization without imposing explicit constraints on the control parameters. As an illustrative example, we optimize a process in which a fluid contained in a cylindrical vessel is mixed by periodically rotating the vessel. The resulting optimal vessel motion is physically reasonable: it reverses its direction of rotation before a solid-body rotation state is established. Furthermore, even when the fluid viscosity increases with time during the mixing process, the method can continuously adapt the control parameters to the changing viscosity.

End-to-End autonomous driving models commonly predict future waypoints from sensor inputs and convert them into vehicle control commands through a downstream controller. However, conventional waypoint-based imitation learning mainly minimizes coordinate-level errors, making it difficult to capture scene-dependent path-speed changes and temporal instability across waypoint outputs. In this paper, we propose an offline teacher-signal generation and learning method for trajectory-output stabilization based on a Scene-Conditioned Interacting Multiple Model (SC-IMM) to mitigate this issue. The proposed method converts expert trajectories into path-speed states and performs IMM updates conditioned on scene cues to generate path-speed teacher labels and mode posterior probabilities. The generated signals are added to the original trajectory loss as auxiliary supervision during training, while the inference structure and waypoint controller remain unchanged. In closed-loop evaluation on 100 short routes in CARLA Town12, the proposed method improved the driving score by 28.0% and reduced Collision/km by 62.3% compared with the baseline, while also improving jerk and trajectory-variation metrics. These results demonstrate that offline teacher signals embedding scene-conditioned motion-model cues can guide trajectory-output driving models toward more stable closed-loop behavior.