Papers

Defenses against jailbreak attacks on Large Language Models (LLMs) operate at different pipeline stages, such as input modification or output guard, but it remains unclear which defenses to deploy at each stage and how to combine them. Prior empirical studies, fragmented by inconsistent attack-success-rate definitions and experimental settings, have evaluated defenses largely in isolation. Here we present the first systematic study, to our knowledge, of defense combinations both within and across pipeline stages, under a consistent threat model of direct, black-box, single-turn attacks. Our decision framework standardizes evaluation through a principled attack-success-rate formulation with controlled query budgets, together with explicit fairness rules. Across 19 attacks and 15 defenses, we find that no single defense is universally best, but well-chosen combinations achieve substantial safety with minimal utility degradation, yielding practical recommendations for layered defense pipelines.

Navigating toward human callers is an important capability for rescue robots operating where visual contact is degraded or occluded. We present AcousticDiffusion, a semantically conditioned, audio-guided diffusion policy for human-directed navigation. A frozen pretrained audio recognizer processes 10.24 s windows, with speech gating and distress-aware prioritization converting recognition outputs into source-level navigation roles. Microphone-array direction-of-arrival measurements are recursively integrated into a robot-centric Bayesian bird's-eye-view belief field. Ego-motion compensation aligns successive observations, progressively constraining source position while preserving bearing-induced range uncertainty. The semantic belief, recent acoustic observations, audio features, and robot state condition a diffusion model that generates waypoint trajectories. On a synthetic-navigation validation set using recorded audio, AcousticDiffusion achieves a mean end-point bearing error of 11.20 degrees, with 91.78% of trajectories aligned within 30 degrees of the caller. Distractor rejection ranges from 89.20% to 98.99%, and the policy favors a HELP-designated caller over a competing speaker in 91.07% of windows. Deployed online on a ZSL-1 quadruped without additional retraining, it achieves a mean bearing error of 64.9 degrees, compared with 98.2 degrees for A* and 90.4 degrees for RRT, with a mean planner compute time of 6.07 ms. Despite imperfect acoustic localization, the reported mean final source distance is reduced from 3.96 m for the classical planners using ODAS-derived (Open embedded Audition System) guidance to 2.48 m, a 37.4% improvement. These results demonstrate the framework's ability to translate uncertain acoustic observations into closer approaches to human callers.

As deep learning models continue to advance, knowledge tracing models have achieved higher accuracy. However, these gains come at the cost of reduced interpretability, which is crucial for practitioners in educational settings to adopt new methodologies. Additionally, deep learning models are prone to overfitting, particularly when dealing with the small datasets that are common in educational applications. In this paper, we propose a novel regularization technique designed to enhance the robustness of deep-learning-based knowledge tracing models, while simultaneously improving their interpretability. Our method addresses both the interpretability and overfitting challenges, making it more feasible for real-world educational applications.

As services are increasingly centralized in public clouds, understanding the nature of Internet Background Radiation (IBR) hitting these particular environments is an important part of understanding their overall security posture. Classical network telescopes, long the cornerstone of IBR research, face hurdles here: their surface area is shrinking, and their well-known address ranges are easily avoided. We present a multi-cloud view of IBR in this paper. We deploy a passive, distributed network telescope with 336 IPs across five major cloud providers. We compare traffic from our cloud telescope with data from two classical telescopes, a large well-known (/9 + /10) and a small unknown /16, to analyze observational biases. To enable a fair comparison across very different telescope sizes, we tune a scan detection algorithm to determine appropriate thresholds. Our findings reveal that IBR in the cloud is strongly provider-dependent rather than geography-dependent, highlighting the necessity of multi-cloud deployments for comprehensive visibility of IBR in the cloud. Our cloud telescope also captures a distinct set of scanners, confirming that scanning activity is not uniform across the IPv4 space, and we confirm that small, distributed telescopes are ill-suited for observing random events like DDoS backscatter. Our work underscores that monitoring must evolve beyond classical telescopes to include diverse, multi-cloud vantage points to accurately capture IBR.

Most multilingual dysarthria-severity systems either train on a single aetiology-language pair or pool heterogeneous aetiologies into one label space. We test that pooling assumption with four matched HuBERT-base contrastive embedding models under a shared backbone, training recipe, corpus registry and held-out evaluation: one mixed-aetiology baseline and three aetiology-specific models for cerebral palsy (CP), Parkinson's disease (PD) and amyotrophic lateral sclerosis (ALS). Training combines clinically labelled speech with ordinal pseudo-labels from a training-free phonological profiling method [1], [2]. On speaker-disjoint, leakage-filtered held-out subsets, the per-aetiology models outperform the mixed baseline across all three target aetiologies: CP (macro F1 0.829 vs 0.676, +22.6 % relative), PD (0.715 vs 0.511, +40.0 %) and ALS (0.788 vs 0.596, +32.3 %). On CP, adding 144 SAP and 44 CDSD pseudo-labelled speakers lifts macro F1 from 0.786 to 0.829 over a clinical-only CP model (+4.3 percentage points). Training data span three to seven languages per aetiology. We position this as a controlled comparison of label-space design choices and discuss pseudo-label calibration, split hygiene, and confidence-thresholded deployment as important limitations for future work.

A pretrained robot foundation policy may execute most of a long-horizon task yet repeatedly fail at a few critical subtasks. Collecting additional full-task demonstrations for supervised fine-tuning (SFT) requires operators to repeat behaviors the policy already performs well. Reinforcement learning (RL) fine-tuning offers a promising path to bridge this gap, but existing approaches struggle to solve long-horizon tasks using only sparse rewards. We present PARTS (Policy Adaptation with RL on Targeted Subtasks), a real-world subtask RL framework that concentrates practice at these bottlenecks while allowing training rollouts to proceed with minimal human intervention. The frozen pretrained policy supplies nominal actions throughout execution, while agent-generated selectors and success verifiers activate residual corrections and provide local outcome rewards. These rewards support learning from successful subtasks even when complete-task successes are scarce. Training combines online RL with success-reweighted retraining, and each retrained residual policy is redeployed to collect further experience. Humans identify bottlenecks during setup and perform physical resets when needed. On bimanual YAM and single-arm Franka tasks, PARTS improves complete-task success from 32% to 61% and from 50% to 95%, respectively, using tens of minutes of real-world RL rollouts per task on average. Compared with existing real-world RL fine-tuning methods, PARTS raises full-task success by more than 25% under the same robot-rollout budget while requiring less human involvement.

Learned world models may have compact interventions even when their recurrent state is high-dimensional, but it is unclear what happens to such a correction after it enters the model. We study this question in a controlled recurrent world model where prior work identified a checkpoint-specific rank-4 interface for one-shot counterfactual velocity interventions. The correction rapidly leaves this fixed entry subspace during autonomous rollout. Nevertheless, a low-rank image obtained by transporting the entry directions through the factual recurrent Jacobian chain continues to capture most of the nonlinear correction. Restarts using the tangent-predicted correction preserve substantial counterfactual future function. This transport/function pattern recurs across independently trained structured-GRU models and a parameter-matched LSTM initialized with a privileged compact correction. We further characterize a finite-horizon future-response operator over the full recurrent carrier. Patching shifts its leading future-sensitive directions toward the matched native-counterfactual organization, and the local operator accurately ranks finite perturbation effects over the registered direction panels at the patched and native-counterfactual basepoints. A separate full-amplitude assay finds substantial factual-endpoint tangent residuals and supports response reconfiguration in two of three checkpoints. Together, these results show that compact intervention structure can persist as a moving, state-dependent local geometry embedded in high-dimensional recurrent dynamics, without implying a fixed or dynamically closed low-dimensional state.

The determination of physical height differences using highly stable atomic clocks has emerged as a novel approach in relativistic geodesy, exploiting the gravitational redshift as a direct observable of geopotential differences. In this study, we investigate the feasibility of satellite-based clock comparisons using the Atomic Clock Ensemble in Space (ACES) onboard the International Space Station, which enables time transfer via microwave (MWL) and optical (ELT) links. Since operational optical data are not yet available, a comprehensive full-scale simulation of realistic ACES observation scenarios is performed, including detailed noise models of clocks and links. A slope-based estimation method is applied to time series of clock comparisons in order to extract the relativistic redshift signal and derive height differences between the ground stations. The performance of the approach is evaluated for quasi-common view, non-common view, and split non-common view configurations, where the latter divides the observation period into shorter intervals. The results show that optical links enable faster convergence and can achieve height accuracies at the decimeter level within a few days and at the centimeter level over longer periods, while microwave links are more strongly affected by noise and bias contributions. Non-common view processing significantly increases observation availability with only minor loss in accuracy, and the split approach provides robust solutions for larger networks. These findings demonstrate the strong potential of satellite-based clock comparisons as a remote-sensing technique for determining physical height differences on a continental scale.

The satisfiability problem for word equations asks whether variables can be replaced by words so that the two sides become equal. For regular word equations, in which each variable occurs at most once on each side, satisfiability is NP-complete. For general quadratic word equations, in which each variable occurs at most twice in total, satisfiability is NP-hard, but its membership in NP remains open. We consider an intermediate class: quadratic word equations with a linear side, where each variable occurs at most once on one designated side. We show that the Nielsen graph of an equation $U=V$ in this class, with total length $N=|U|+|V|$, has diameter $O(N^{12})$, measured over reachable pairs of vertices. Together with the known NP-hardness for regular word equations, this result establishes NP-completeness of satisfiability for this class. We also show that each strongly connected component is isomorphic to the length-preserving reachability graph of a regular equation, and that the condensation graph has depth at most $|U|+2|V|$.For arbitrary quadratic word equations, we further show that Nielsen paths can be shortened to a length polynomial in the equation size and the number of prefix operations between variables repeated on opposite sides. This yields NP membership for subclasses defined by variable occurrence arrangements.

We present high-resolution cosmological hydrodynamic simulations of the Warm-Hot Intergalactic Medium (WHIM) using the GPU-optimized grid-based hydrodynamic code Kratos. Employing a uniform 4096^3 grid in a (100 h^-1 Mpc)^3 comoving volume, we achieve a spatial resolution of ~24.5 h^-1 kpc, sufficient to resolve the Jeans scale of gas at T ~ 10^4 K and n_H ~ 10^-3 to 10^-2 cm^-3. This calculation ranks among the largest grid-based cosmological hydrodynamic simulations performed to date. We find that ~23.4% of cosmic baryons reside in the WHIM phase (T = 10^5 - 10^7 K) at z = 0, significantly below the 40-50% found in earlier, lower-resolution simulations. Through a suite of lower-resolution simulations, we demonstrate that spatial resolution plays a pivotal role in determining the WHIM fraction: resolving gas near its Jeans scale allows it to reach higher densities, where enhanced radiative cooling transfers a substantial fraction of baryons out of the WHIM temperature range. The remaining WHIM resides predominantly in filaments and accretion-shock structures in the vicinity of halos, where hierarchical structure formation and ongoing hydrodynamic accretion provide continued shock heating. Synthetic observations of Lyman-alpha, O VI, O VII, and O VIII emission reveal distinct morphological and kinematic signatures, with O VI tracing filament-halo interfaces and the X-ray lines probing hotter gas associated with massive halos. These predictions underscore the importance of current and future missions such as XRISM, ATHENA, and HUBS for mapping WHIM thermodynamics and kinematics.

We construct a maximally supersymmetric Type IIB background in signature $(5,5)$ on the real ten-manifold underlying complex $\mathrm{AdS}_5$. The spacetime is diffeomorphic to $T^*S^5$ and carries the invariant metric $G=\mathrm i(\bar q-q)$, where $q$ is the holomorphic constant-curvature metric. The real five-form $F_5=(4\mathrm i/L)(\Omega-\bar\Omega)$, built from its normalized volume form, is self-dual, and its stress tensor supports this scalar-flat, non-Einstein metric. We derive these fields from the most general $\mathrm{SL}(4,\mathbb C)$-invariant metric and five-form ansatz: self-duality and the Einstein equation fix their phases up to the overall scale, orientation, and flux sign. The fields obey the reality conditions of the untwisted $\mathrm{IIB}_5$ theory. The holomorphic coframe makes the supercovariant curvature cancellation explicit, and flat transport gives thirty-two global real Killing spinors. Their Killing superalgebra is the factor-exchanging real form of $\mathfrak{psl}(4|4,\mathbb C)$, and we identify the corresponding supercoset.

The non-adiabatic heating of the slow Solar Wind (SW) remains an open problem, with wave–particle interactions as a primary candidate mechanism. Novel in situ Parker Solar Probe (PSP) observations reveal strongly perpendicular anisotropic velocity distribution functions (VDFs), called "hammerhead", correlated with intense wave activity. These VDFs are systematically measured at the Heliospheric Current Sheet (HCS), making the hammerhead an important kinetic signature of the slow SW. In this work, we employ a fully kinetic particle-in-cell approach, complemented by a linear Vlasov solver to cross-validate the simulation results, to investigate the stability of these VDFs, the timescales over which they evolve, and their interaction with plasma waves. Our main findings indicate that the hammerhead distribution is primarily susceptible to drift-type instabilities, while energy is nonlinearly transferred back to the plasma through a combination of Landau and cyclotron resonances, resulting in net heating in the parallel direction, preferentially energizing the beam proton population. Crucially, these nonlinear interactions do not drastically alter the morphology of the distribution. This suggests the possibility that the hammerhead may be generated locally within the HCS in the inner heliosphere and subsequently advected outward, where it is eventually measured by PSP. This work lays the ground for future investigations into the kinetic physics of the HCS.

Increasing demand for high-performance optical devices drives the search for improved fabrication and design paradigms. While photonic circuits have traditionally used structures with two discrete 'binary' height levels, grayscale 'wavy' interfaces, or optical Fourier surfaces, have recently become possible. They provide precise control over the Fourier components that govern the optical response. This capability raises the question: When does a wavy device improve performance and why? Here, we show that wavy integrated processors exhibit superior accuracy and efficiency to binary analogs. Inverse-designed wavy interferometers have the design freedom to minimize outscattering to free space and backreflections. They reach a five-fold lower transmission error than binary counterparts and enable bandwidths up to 300 nm. We formalize these findings for other distinct integrated devices, such as photonic crystals, nanocavities, and beam emitters. A Fourier-optics analysis identifies a trade-off: wavy profiles manipulate light more accurately, whereas binary profiles excel in interaction strength. As such, tailored wavy profiles emit high-quality beams. However, binary profiles remain preferable in photonic crystals and nanocavities, where maximal index contrast is required and binarization-induced higher harmonics are benign. Thus, for quantum information, optical computing, and sensing, optical Fourier surfaces offer a route to miniaturized integrated circuits with improved performance.

3D object detection from LiDAR point clouds faces a fundamental dilemma: voxel-based methods achieve efficiency at the cost of geometric quantization, while point-based methods preserve fidelity but suffer from prohibitive computational bottlenecks. Specifically, point-based architectures are crippled by slow downsampling strategies (e.g., FPS) and expensive dynamic neighbor queries (e.g., k-NN) coupled with costly continuous interactions. To tackle these systemic inefficiencies, we propose PointLAM, a highly efficient and powerful point-based architecture driven by two synergistic innovations. First, to resolve the downsampling bottleneck, we develop the Laplacian Point Sampler (LPS). LPS employs an implicit discrete Laplacian high-pass filter and Doubly Sorted Sampling to achieve fast, structure-aware foreground preservation. Second, to overcome local modeling latency, we design the Local Hadamard Aggregator (LHA). LHA decouples spatial indexing from feature representation using transient grids, and replaces complex continuous interactions with a Hadamard Gating mechanism for topology-aware, attentive modulation. By coupling this local gating with Bi-Directional Mamba (BDM) layers for global sequence modeling, we formulate the Local Attentive Mamba (LAM) block. Powered by this architecture, PointLAM achieves competitive performance on nuScenes and Waymo for point-based detectors. It rivals highly optimized voxel competitors while requiring a fraction of the computational footprint, demonstrating marked superiority in detecting small instances and handling extreme sparsity. Project page: https://pointlam.github.io/.

Let $p$ be an odd prime and let $S\subseteq\mathbb{F}_p^2$ have cardinality divisible by $p$. We prove that, for all sufficiently large primes $p$, no such set has exactly four special directions, and obtain a conditional extension to larger numbers of special directions under an affine-independence assumption on the corresponding projection functions. A separate second-moment argument shows more generally that, for every fixed $k\ge4$, no subset of $\mathbb{F}_p^2$ has exactly $k$ special directions once $p$ is sufficiently large. In fact, the result holds uniformly for $k$ up to a positive constant times $\sqrt p/\log p$. In contrast, for multisets every prescribed collection of at most $p$ directions can occur as the set of special directions of a $\{0,1,2\}$-valued multiset on $\mathbb{F}_p^2$.

In this paper, we prove that symmetric, doubly-periodic, three-dimensional hydroelastic waves of small amplitude, bifurcating from a non-uniform shear flow, are necessarily two-dimensional to leading order. A detailed analysis of the solvability condition at quadratic order reveals that all Fourier coefficients with transverse dependence vanish identically. The dimensional reduction is controlled entirely by the vertical structure of the shear flow and is robust under changes to the dynamic boundary condition.

This paper presents a novel online coverage path planning (CPP) algorithm, called TRACE, for real-time coverage of unknown environments. TRACE is built upon a hierarchical coverage tree that provides a global representation of the evolving connectivity of the uncovered space. As the environment is incrementally revealed and covered, newly discovered obstacles and covered cells may fragment the remaining uncovered space into disconnected regions. TRACE recursively expands the corresponding tree nodes to explicitly represent these regions and organize them for subsequent coverage planning. Based on the updated tree, an incremental global tour is maintained to guide the coverage process. TRACE locally refines only the affected portions while preserving the visiting order of unchanged regions, thereby reducing the computational burden of global replanning and maintaining a consistent coverage progression. Guided by the global tour, a local planner generates back-and-forth coverage paths and switches to global-tour-aware planning to efficiently complete the target regions. Theoretical analysis establishes the computational complexity and complete coverage property of TRACE, and derives an approximation bound for the incremental global tour refinement. The performance of TRACE is evaluated through extensive high-fidelity simulations and real-robot experiments using a mobile robot. Comparative evaluations against six existing CPP methods demonstrate significant improvements in coverage time, path length, overlap ratio, and number of turns.

Gradiometry provides a versatile alternative to passive environmental shielding in quasi-static magnetometry, effectively suppressing background noise through differential signal extraction. Nevertheless, traditional implementations rely on multi-sensor architectures restricted to spatial gradients, where subtracting signals from independent detectors involves imperfect suppression of common-mode noise and artifacts, limiting their sensitivity. To overcome these limitations, we introduce a quantum control sequence that enables intrinsic temporal and spatial vectorial gradiometry of magnetic fields using a single quantum sensor. Our method provides direct access to first and higher-order derivatives of the magnetic field and extended applicability via auxiliary nuclear spin memory. We showcase this protocol on an ensemble of nitrogen-vacancy (NV) centers in diamond and combine it with mechanical control to realize high-precision differential sensing. Through detailed numerical simulations, we demonstrate the performance of our scheme in two critical DC magnetometry applications: (i) vector magnetic anomaly detection and (ii) non-invasive gradiometry of neuronal action potentials.

Computational notebooks make problem-solving visible, but typically only one notebook at a time. Meanwhile, in data science platforms like Kaggle, one competition can accumulate hundreds of notebooks. Effective collection-level analysis requires characterizing recurring solution patterns across all notebooks, as well as isolating specific notebooks for closer examination and learning. However, standard notebooks provide no common basis for this. Their workflows are nonlinear, cells declare no intent, and identical code can serve different ends, leaving hundreds of notebooks as separate documents. In this paper, we present Notrix, an interactive visual analytics tool for profiling hundreds of notebooks as one collection. Inspired by a formative study (N = 11), Notrix classifies every cell into one of thirteen machine learning (ML) stages, turning each notebook into a stage sequence, and clusters those sequences by structure rather than by code. To keep the representation constant as the scope narrows from the whole collection to a single cell, Notrix features three coordinated views—Workflow, Structural Matrix, and Detail—that appear at all four levels of granularity. In a within-subject study (N = 17) using two Kaggle collections of over 400 notebooks each, we observed participants answered questions about all notebooks more accurately with Notrix (median 88% vs. 50%) while opening 80% fewer notebooks per minute. Notably, four of the fourteen answered it without opening a single notebook (interaction logs, N = 14). Participants also reported significantly lower mental demand, temporal demand, and stress with Notrix (Holm-Bonferroni adjusted).

Modern AI workloads increasingly rely on scale across architectures that interconnect multiple datacenters to form a single "AI factory", overcoming the power and cooling constraints of individual sites. However, extending Remote Direct Memory Access (RDMA) across wide area networks (WANs) introduces fundamental challenges: multi-path packet reordering, high latency, and packet loss that severely degrade performance. While erasure coding (EC) has emerged as a promising mechanism for loss recovery, its effectiveness critically depends on efficient packet arrival tracking implemented in hardware. We present COmpact Multi-path Erasure-coded Tracking (COMET), the first fully hardware-offloaded packet-arrival tracking design implemented on an FPGA-based network interface card (NIC) for multi-path RDMA over lossy WANs. COMET employs a scalable cache-based architecture that supports operation at high link rates. Our evaluation shows that COMET sustains line rate operation at 400 Gbps and beyond. Critically, COMET decouples on-chip memory footprint from link Bandwidth-Delay Product (BDP), and its cache-based architecture (COMET Cache) enables supporting 6 times more concurrent connections than state-of-the-art (SOTA) SoC-based designs. These results demonstrate that scalable, fully hardware-offloaded packet-arrival tracking is practical on FPGA-based NICs at current data rates, and its architectural scalability extends to emerging 1.6 Tbps NICs and beyond.

We show that a simply connected compact symmetric space of dimension $n\ge 3$ admits a conformal Killing $2$-form which is not Killing if and only if it is homothetic to one of $\mathbb{S}^n$, $\mathbb{CP}^m$ or $\mathbb{HP}^q$.

Phylogenetic networks provide a flexible framework for representing reticulate evolutionary processes, such as hybridization, introgression, recombination, and horizontal gene transfer. However, their combinatorial complexity makes even basic enumeration problems difficult. Building on our previous work for networks with up to three reticulations, we derive an explicit closed-form formula for the number of unrestricted rooted binary phylogenetic networks with four reticulations on \(n\) labeled taxa. Our approach is based on tree-component graphs. We classify the 79 possible component graphs corresponding to networks with four reticulations into ten groups. We then enumerate the networks associated with each group by combining known counts of one-component networks, forests, and networks with fewer reticulations. Summing these contributions yields the desired formula. This result extends the exact enumeration of unrestricted binary phylogenetic networks to four reticulations and further demonstrates the effectiveness of component graphs for systematically organizing and counting increasingly complex network classes.

Let $G$ be a finite abelian group, and let $k$ be an infinite perfect field containing a primitive root of unity of order the exponent of $G$. We prove that the $G$-equivariant version of Gabber's presentation lemma proposed by Bachmann holds over $k$ if and only if $G$ is cyclic of $p$-power order, for some prime $p$.

Image registration is a vital preprocessing step in multimodal perception tasks, including image fusion, object detection, and semantic segmentation. In Advanced Driver- Assistance Systems (ADAS), spatial misalignment between visible (RGB) and infrared (IR) cameras -caused by non-coincident optical axes and field-of-view differences- introduces non-uniform parallax and visual ghosting. Classical keypoint-based methods are restricted to global homographies that fail under dynamic depth, while unconstrained dense flow algorithms lack structural regularization and suffer from temporal instability. In this paper, we propose XCalib, an unsupervised dense thermal-visible registration framework that bridges this gap. Rather than serving as an absolute metric calibration tool, XCalib leverages virtual pinhole camera parameterization strictly as a geometric constraint space. By optimizing effective relative pose and intrinsics alongside predicted monocular metric depth, XCalib restricts the search space of spatial displacements to physically valid projection geometries. Our key contributions are: (1) a novel registration paradigm that uses camera parameterization as an implicit regularizer for dense cross-modal warping; (2) Normalized Edges Correlation (NEC), a robust structural similarity metric tailored to cross- spectral alignment; and (3) extensive quantitative and qualitative evaluations across public ADAS datasets, demonstrating superior temporal stability and alignment accuracy over unconstrained dense flow baselines.

We consider the stabilization of two-dimensional periodic hydroelastic waves in deep water beneath a massless nonlinear Cosserat sheet by means of a localized pres- sure feedback. For sufficiently regular solutions, under an explicit condition relating gravity, flexural rigidity and the localization profile, we prove a uniform spacetime estimate and a localized dissipation observability inequality. As a consequence, the hydroelastic Hamiltonian decays quantitatively as long as the solution remains in a small geometric regime. The proof combines an energy dissipation identity with a nonlinear multiplier method adapted to the infinite-depth geometry, together with a coercivity analysis of the Cosserat bending energy.

The neutrino-matter interaction cause the final flavor compositions deviating from those expected in vacuum. In this work we consider neutrinos interact with ultra-light scalar dark matters $\phi$ ($m_\phi\ll 1\,{\rm eV}$). When the neutrinos emitted from a distant source and propagate through the dark matter medium, the MSW potential in the Hamiltonian is replaced by the hypothetical $\nu\phi$ effective potential. Two types of neutrino sources are adopted in our calculations. The first is point-like neutrino source, the Active Galactic Nuclei (AGN), which produce $\mathcal{O}({\rm TeV})$ neutrinos primarily through charged pion decay, yielding an initial flavor ratio of ($\nu_e,\nu_\mu,\nu_\tau$)=($1:2:0$). We focus on two specific AGNs, NGC 1068 and TXS 0506+056 in our analysis, whose distances from Earth are much larger than the neutrino oscillation lengths. We consider the range of $\nu\phi$ coupling constant such that the adiabatic condition and mean free path can be satisfied. Hence, by assuming neutrinos are produced randomly from their corona regions, we compute the averaged flavor ratio at Earth with $m_\phi=10,1,0.1\,{\rm peV}$ and $m_\phi=1,0.1,0.01\,{\rm feV}$, generate the ternary plots of neutrino flavors, and compare the results with the IceCube present and future sensitivities. The second source considered is the Diffuse Supernova Neutrino Background (DSNB), it is an isotropic neutrino source with energy $\mathcal{O}({\rm MeV})$. Applying the core-collapse simulation neutrino temperatures to the DSNB flux, we repeat the similar analysis for the DSNB with $m_\phi=10^{-22}\,{\rm eV}$, and we estimate different neutrino flavors lies in the HK/DUNE/JUNO combined sensitivities limit.

Vision-language models (VLMs) can replace human annotators in preference-based reward learning, but sequential API requests and single-environment data collection make training slow and costly. We present RAPID (Reward learning with Adaptive Parallel Image Diversity), a system that couples GPU-parallel rollout with data-aware policy updates, single-request preference labeling, automatic reward stabilization, and representative image sampling. We evaluate these components on five Franka Panda manipulation tasks in IsaacLab. Parallel rollout and adaptive updates provide the first substantial reduction in training time: under matched two-stage prompting, mean runtime falls from 9.18 to 3.13 hours. With all RAPID components enabled, training completes in 1.15 hours using 896 rather than 19,840 API calls per run, and aggregate final success rises from 86.3% to 98.7%. This represents an 8.0$\times$ end-to-end speedup and a 95.5% reduction in API usage. An offline evaluation with Gemma~3 12B and GPT-4.1 mini demonstrates that single-request prompting reduces labeling latency and cost across both models. Code is available at: https://github.com/rapid-vlm/rapid-vlm-rl.

Low critical charge-current density and low energy dissipation are highly desired in magnetic random-access memories, requiring spin sources to exhibit both high charge-to-spin conversion efficiency (CSE) and high charge conductivity. Altermagnets with vanishing net magnetic moment and spin-splitting bands provide promising spin-source candidates for spin-splitting-torque magnetic random-access memories. However, achieving both high CSE and charge conductivity remains challenging in altermagnets. In this work, we introduce Dirac cones into two-dimensional d-wave altermagnets, where their intrinsically high carrier mobility enables tunable charge and spin conductivities with high CSE. Dirac-cone anisotropy provides an effective means of enhancing both CSE and charge conductivity, with cone tilting serving as an additional degree of tunability. Guided by this design principle, we identify a maximum CSE of 92% in Cr2SeTeS. When the Fermi level moves slightly away from the Dirac point, high CSE, high charge conductivity, and the resulting high spin conductivity can be simultaneously achieved. Our study advances the understanding of time-reversal-odd spin transport via Dirac-cone engineering and provides a practical route toward developing spin-source materials that combine high charge conductivity with highly efficient charge-to-spin conversion.

Let $B$ be the bluebird combinator with reduction rule $Bxyz \to_{w} x\left(yz\right)$, let $M$ be the mockingbird combinator with reduction rule $Mx \to_{w} xx$, and let $I$ be the identity bird combinator with reduction rule $Ix \to_{w} x$. A fixed-point combinator, called a sage bird by Smullyan, is a closed term $Y$ such that, for a fresh variable $x$, $Yx$ is equivalent to $x\left(Yx\right)$ under these reduction rules. For a fixed variable $x$, we construct an invariant $\mathrm{Tr}_{x}\left(u\right)$ of a $BMI$-term $u$ with respect to $\to_{w}$. This invariant traces the occurrences of $x$ in the leftmost-innermost reduction sequence of $u$. We then prove that $\mathrm{Tr}_{x}\left(Yx\right) \neq \mathrm{Tr}_{x}\left(x^{r}\left( Yx \right)\right)$ for every $x$-free $BMI$-term $Y$ and every $r\geq 1$. Consequently, there exists no fixed-point combinator in $BMI$-combinatory logic. This provides a negative answer to the problem posed by Smullyan in 1985.

One revisits the Entangled Relativity theory framework, embedding it in a larger set of theories. From their scalar-tensor reformulation, these theories admit General Relativity spacetime solutions as solutions with the same matter content, if this matter obeys an "intrinsic decoupling" condition. One shows that a set of matter Lagrangians satisfying this condition is completely characterized by the way they depend on the metric tensor. One then shows that it is possible to define a perfect fluid Lagrangian fulfilling the intrinsic decoupling condition up to 1PN order. Specifying to weak field spherical stars suggests that Entangled Relativity and its proposed generalization are worthy alternatives to General Relativity. Indeed, the obtained numerical values of the Eddington parameters abruptly satisfy the known observational constraints. However, the non strict compliance of the theory with conservation laws demands to reconsider the link between observational data and the parameters for a definitive clarification of this point.