Papers

We present a minimal, predictive scenario for asymmetric fermionic dark matter (DM) generated through freeze-in. The standard model (SM) is extended by a Dirac fermion $\psi$ that serves as the DM candidate, an auxiliary Dirac fermion $\chi$, a real scalar mediator $\phi$, and a heavier real scalar $S$, all stabilized by a $Z_2 \times Z_2'$ discrete symmetry. The DM asymmetry is generated through the CP-violating out-of-equilibrium decays of the mediator $\phi$, while the heavier scalar $S$ plays a dual role: it generates the physical CP-violating phase required for the asymmetry through its interference with $\phi$, and it mediates the annihilation of the symmetric component of the DM population. We solve the coupled Boltzmann equations for the asymmetric yield $Y_-$ and the total yield $Y_\Sigma$, and identify the scaling relations $Y_-^\infty \propto \mu^2 \varepsilon / m_\phi$ and $\Omega_{\rm DM} h^2 \propto m_\psi \, \mu^2 \varepsilon / m_\phi$, which we verify numerically. The observed relic abundance $\Omega_{\rm DM} h^2 \simeq 0.12$ is naturally reproduced for a wide range of parameters; in particular, for $\varepsilon = 10^{-6}$ and $m_\phi = 200$~GeV, the correct relic density is obtained for $\mu \approx 2.7 \times 10^{-9}$~GeV and $m_\psi \approx 40$~GeV. We further confront the model with current experimental constraints, including direct detection, indirect detection, collider searches, and DM self-interactions, and show that the predicted signals lie far below the sensitivity of current experiments. This scenario provides a self-consistent and testable realization of asymmetric fermionic DM within the freeze-in framework.

We study the spherical collapse of a homogeneous dust cloud on a three-brane in the Shtanov–Sahni braneworld, in which the extra dimension is timelike. In the high-energy regime the brane Friedmann equation acquires a negative quadratic term in the matter density, controlled by a critical density $\rhoc=2\lambda$, where $\lambda$ is the magnitude of the negative brane tension. For a spatially closed interior we integrate the dynamics exactly. Every turning point of the scale factor lies above $\astar=\sqrt{3/4\pi\lambda}$, a bounce exists if and only if a simple condition on the comoving density holds, and all curvature invariants stay below bounds fixed by $\lambda$ alone along the entire cycle. Comoving worldlines are geodesically complete. We then match the cloud across a comoving boundary to a spherically symmetric exterior with a trace-free effective source and solve the Israel conditions in closed form. The exterior is a charged Vaidya geometry whose mass function and tidal charge are fixed by the boundary radius. The tidal charge is positive, opposite in sign to the Randall–Sundrum case, and the exterior cannot be static. A trapped region forms on the boundary if and only if the mass exceeds $2/(3\sqrt{\pi\lambda})$, and in that case the bounce still takes place in an untrapped neighbourhood of the boundary. For trapped clouds we construct an exterior over the full contraction, bounce and re-expansion by joining an advanced and a retarded charged Vaidya patch through a static wedge. The matched metric is of class $C^{1,1}$, the effective energy conditions hold, and the cloud re-emerges through a white hole region into a second asymptotic region. In the general relativistic limit $\lambda\to\infty$ the Oppenheimer–Snyder solution is recovered.

Doc2Query, a popular document expansion technique, leverages sequence-to-sequence models to generate relevant queries, effectively addressing the "vocabulary mismatch" problem in information retrieval. However, these models often suffer from generating either hallucinations unrelated to the document or repetitive content already present in the document. Training sequence-to-sequence models to produce high-quality, novel, and relevant tokens remains a significant challenge. To address these issues, we introduce a novel approach, QGDPO, that employs direct preference optimization (DPO) to guide the generation process. We first fine-tune a base sequence-to-sequence model and subsequently utilize a relevance model to score its predictions. Based on these scores, we construct pairs of winning and losing predictions as relevance preferences for the DPO training. Furthermore, we enhance our pipeline by using the relevance model to filter out poor predictions, retaining only the most relevant generated content for indexing. QGDPO effectively eliminates 50% of irrelevant predictions comparing against Doc2Query baselines, while the relevance filter removes an additional 14.61%. This feature has been successfully deployed to production on Walmart.com for full traffic, with a substantial improvement in relevance and user engagement.

Feed-forward 3D Gaussian Splatting (3DGS) increasingly aggregates multi-view evidence with heavy learned 3D networks. We propose LoCoSplat (Local-Context Splatting), motivated by the observation that a Gaussian is a local primitive: once depth is predicted, what the 3D stage must add (scale, rotation, opacity) depends on the point cloud around each anchor, and a fixed local average of that neighbourhood is enough to supply it, no heavy network required. LoCoSplat realises exactly this average: it splats a 16-d linear projection of the point features into a fine and a coarse grid and reads both back at each anchor with a 0.14M-parameter pointwise MLP; with no learned 3D network and no dynamic sparse computation, its whole encoder runs as one fp16 CUDA graph. On RealEstate10K, LoCoSplat outperforms every prior feed-forward method on PSNR, SSIM, and LPIPS at 6, 12, and 24 views, with a margin that widens as views densify (+3.3 PSNR over VolSplat, the prior voxel-aligned state of the art, at 24 views) and grows further under zero-shot transfer to ACID and fine-tuning on ScanNet. It reconstructs a 6-view scene in 33 ms on one NVIDIA RTX PRO 6000 GPU, the fastest of seven feed-forward methods and $4.2\times$ faster than the previous state of the art, trains $2.7\times$ faster ($5.9\times$ at 24 views), and uses $6.7\times$ less inference memory.

We present a method for the data-driven inference of Hamiltonian and Lie-Poisson dynamical systems from vector-field samples, in which the Hamiltonian and the underlying Poisson tensor are estimated jointly rather than assuming the structure is known. The Hamiltonian is represented as a linear combination of random feature maps with fixed random internal weights, so its estimation is a linear regression problem for any given Poisson structure. We treat three cases of increasing generality: a known Poisson tensor, an unknown constant invertible (symplectic) structure matrix, and Lie-Poisson and affine Poisson structures, with structure matrix linear in the state. In the constant case, the Hamiltonian and structure matrix are recovered jointly from an eigenvalue problem that avoids the trivial zero solution due to the scaling symmetry of Hamiltonian systems. For Lie-Poisson and affine Poisson systems we alternate between an initial Hamiltonian estimate from a generalized eigenvalue problem for an approximate first integral, and a constrained nonlinear least-squares fit of the structure constants that enforces the Jacobi identity and, for affine structures, the cocycle condition, via continuation in a penalty parameter. Orthogonal matching pursuit selects a compact, well-conditioned subset of random features. Because the learned Hamiltonian is a sum of single-feature terms, the model admits an explicit Poisson/symplectic splitting integrator, which we use to validate the learned systems via long-time Poincare sections. We demonstrate the approach on the four- and five-dimensional Lorenz-86 model and a six-dimensional Kirchhoff rigid-body system, recovering Hamiltonians, structure constants and Casimir invariants to high accuracy, and showing the inferred Poincare sections converge to the correct invariant-tori topology as the number of random features increases.

We study the involution-preserving properties of the Active Flux method on unstructured triangular meshes. Two model problems are considered: the two-dimensional linear acoustic system, which satisfies a curl involution on the velocity field, and the two-dimensional Maxwell equations with an in-plane magnetic field, which satisfy a divergence involution. Both systems can be written in conservative form. The question we address is how a single discretization can satisfy the conservative form and the involution at the same time. Within a discrete Active Flux framework on simplices, we derive explicit formulas for a discrete curl operator that is exactly preserved by the scheme for acoustics, and for a discrete divergence operator preserved for the Maxwell system. The analysis applies to general simplex meshes and to any polynomial degree, including extensions of the method beyond third order. Numerical experiments with the third-order scheme confirm the theoretical results. Highlights: 1-Active Flux on unstructured triangular meshes preserves discrete involutions, 2-Explicit discrete curl operator preserved by the scheme for linear acoustics, 3- Explicit discrete divergence operator preserved for 2D Maxwell equations, 4- Results hold on general simplex meshes,% and for any polynomial degree 5- Third-order numerical experiments confirm the theoretical results. \end{itemize}

We develop a price-based framework for stochastic portfolio theory in which trading strategies are generated from nominal price weights and evaluated relative to a price-weighted benchmark. Stock splits and reverse splits induce jumps in the generating weights without changing the value of existing investments. In a semimartingale market with predictable adjustment events, we incorporate the corresponding share adjustments into the self-financing condition and construct additively and multiplicatively generated strategies with explicit jump-corrected wealth decompositions. For additive generation, an entropy-based example exhibits relative wealth tending to $-\infty$ almost surely under repeated splits, demonstrating why the usual argument for long-horizon relative arbitrage does not extend directly. We also show that advance adjustment information alone provides no model-free guarantee of improved performance for functionally generated portfolios. For multiplicative generation, wealth-preserving restarts retain the classical portfolio allocation evaluated at the current price weights. We also establish a correspondence with the capitalization-based framework that preserves absolute wealth and yields a strategy-independent benchmark conversion. We illustrate the framework using daily NYSE data, comparing price- and capitalization-weighted benchmarks and the corresponding diversity-weighted portfolios across price- and capitalization-selected universes.

Identical utterance choices can arise from different communicative causes, and identical interpretations can leave different traces in what a listener learns. Rational Speech Act (RSA) models treat interpretation as inference over speaker meaning, but standard one-shot RSA does not intrinsically distinguish these causal update targets. We develop a coupled active-inference model of dialogue in which a listener's likelihood for an utterance is the policy distribution attributed to the speaker, placing the speaker's expected free energy within the listener's variational free energy. Each utterance is therefore both evidence about and an intervention on a partner. The model separates four updates that RSA approaches typically collapse: inference about the partner's current pragmatic state; learning of partner-specific parameters; prospective evaluation of clarification or repair; and a policy prior shaped by habit and a context-sensitive price of time. Under one-step, exact-inference restrictions, the model recovers the RSA speaker and listener, with RSA as the restricted single-turn limit of the coupled process. Outside these restrictions, the updates obey distinct rules and timescales, predicting which adaptations persist, remain partner-specific, or transfer. Worked examples show audience design reversing after clarification, self-confirming misunderstanding in which both interlocutors have low free energy while disagreeing about reference, and rational closing before uncertainty is resolved. Consistent with critiques of equating natural language with communication, the model treats communication as a downstream use of linguistic structure and recasts production and interpretation as coupled inference: speaking is both an intervention on a partner and an epistemic action that samples evidence for the speaker's model of that partner.

This paper proposes any-scale object detection using arbitrary-scale super-resolution for continuously rescaling object images, while general multi-scale object detection uses discretely rescaled appearance representations. However, a naive usage of super-resolution produces many false-positive detections if many super-resolution images are independently fed into an object detector. Our method suppresses these false positives by predicting scale proposal maps, each of which represents a set of pixels appropriate for each super-resolution scale.

Many wireless control tasks require repeated selection of a single action from a finite feasible set under stringent latency and reliability constraints. While large language models (LLMs) have recently emerged as general-purpose decision engines, their autoregressive generation mechanism is not naturally aligned with such bounded control problems. This paper investigates System-One models, which directly learn probability distributions over explicitly defined decision spaces, as a lightweight alternative for wireless decision-making. We formalize their decision structure and learning objective, identify their applicability across physical-layer control, radio resource management, mobility, network slicing, and network operations, and evaluate their practical behavior through representative wireless case studies. Using Jev as a System-One implementation, we benchmark decision quality and client-observed latency against generative LLMs and conventional baselines. In receive-antenna selection, Jev delivers up to an 8.5x reduction in median response latency relative to the evaluated LLMs, although this gain comes with a loss in decision quality compared with stronger task-specific alternatives. More notably, in intent-conditioned RAN slicing, Jev achieves utility comparable to the evaluated LLMs while providing more than a 3.5x reduction in median response latency. Complementary evidence from edge-service orchestration further shows that faster decisions do not necessarily translate into lower end-to-end service latency. These results expose a fundamental quality/latency tradeoff and position System-One models not as replacements for numerical optimization, but as a promising decision interface for latency-sensitive, bounded, and intent-driven wireless control.

Reinforcement learning with verifiable rewards (RLVR) has been shown to improve the reasoning capability of large language models (LLMs) across diverse reasoning tasks. However, group-based RLVR methods, such as GRPO, assign a uniform advantage to all tokens within rollouts of the same outcome. While existing works refine credit assignment of GRPO based on local signals such as token locations or entropy, they often fail to capture the global semantic novelty of a reasoning behavior relative to the current policy. In this work, we propose a hierarchical credit assignment approach for group-based RLVR methods, called HarA, which identifies and encourages semantically novel reasoning behaviors during RLVR. HarA represents each sampled rollout as a distribution over the hidden states and locations of tokens, and computes the Fused Gromov-Wasserstein (FGW) barycenters of all rollouts with the same outcome, capturing the internal reasoning patterns in the latent space under the current policy. The semantic novelty of a reasoning element can then be measured by its contribution to the FGW distance between the current rollout and the barycenter. While solving the FGW formulation is expensive, we introduce an anchor-guided linearization that turns it into a Wasserstein formulation solvable via the Sinkhorn algorithm efficiently. By reweighing token-level advantage of group-based RLVR methods based on the novelty signals, HarA highlights novel reasoning behaviors at flexible granularities to encourage fine-grained LLM exploration. Extensive experiments across three group-based RLVR methods show that our plug-and-play method effectively enhances the exploration of LLMs, outperforming existing methods across diverse reasoning benchmarks.

Reservoir computing (RC) is a machine learning technique for data-driven modeling of dynamics for forecasting and control, primarily studied in computer science and physics literature with promising applications in neural network learning, physical computing, and neuroscience. Why reservoirs learn and how to choose good reservoir architectures are considered important open questions. We show that the RC problem is mathematically an extension of a classic problem in systems and control theory, the Kazantzis-Kravaris-Luenberger (KKL) observer design problem. As a consequence, many of the questions considered open for RC stand to benefit from a large body of theory in the mature KKL literature, non-exhaustively including on questions of embedding, transverse stability, local and global uniqueness guarantees, and effective data-driven solution constructions. Elaborating on this connection, we show that the surprising forecasting ability of reservoirs is in fact a direct consequence of the well-known observer internal model principle, derive an upper bound on the prediction error over a fixed forecast horizon, and provide a partial explanation for why linear readout training in RC works reasonably well. This work illustrates how classical ideas from systems and control can provide strong theoretical backing and open new questions for modern machine learning methods.

Inter-robot communication is a key enabling technology in cooperative robotics applications. While wireless communication is ubiquitous, simple, and effective, it has inherent limitations: the receiving robot cannot easily localise the source of an incoming signal, clock synchronisation is challenging, and signals are broadcast to all nearby devices rather than targeted recipients. Optical communication provides a complementary channel that is inherently directional and spatially localised, directly addressing these limitations. Event cameras are bio-inspired dynamic vision sensors that respond to changes in image intensity with high temporal resolution, high dynamic range, and low latency, making them well-suited as receivers for high-rate optical communication in cooperative robotic systems. In this paper, we propose the Asynchronous Tracking and Optical Communication (ATOC) system, which integrates LED smart-beacon modulation, event-based detection, optical tracking, and event-data demodulation in a single pipeline. By simultaneously tracking and demodulating multiple LED smart beacons, ATOC transforms a conventional visual marker into a robust communication channel suitable for a wide range of high-impact robotics applications. We validate ATOC in a suite of laboratory studies and two 'applications': a smart-city demonstration and a 3D motion-capture system.

We present two decades of X-ray monitoring of the eclipsing accreting pulsar Hercules X-1 with Fermi/GBM, Swift/BAT and MAXI/GSC, centred on the extended low state (ELS) that began in 2025 May (the first such event in 22 years) and is still ongoing at the time of writing, $\sim$480~d after its onset. The 35-d superorbital clock does not simply stop in the ELS; it fades. The surviving modulation stays phase-locked to the pre-ELS clock until the flux of the main high state (the bright phase of the 35-d cycle) has decayed to a few per cent of its normal level. The in-phase coherence is then lost, around 2026 January–March. We classify the pre-ELS pulse-frequency history into three torque states: spin-up, spin-down and flat. The spin derivative correlates positively with the main-high hard X-ray flux (Pearson $r\simeq0.7$), and the ELS hosts the longest sustained spin-down of the record, $\dot{\nu}\approx-4.7\times10^{-13}$~Hz~s$^{-1}$ over the $\gtrsim$380~d of \fer/GBM coverage. The 8–50 keV pulse-profile morphology is statistically indistinguishable across the torque states at fixed superorbital phase, and the shape parameters show no secular drift over the 18-yr record. These results show that the precessing warped accretion disc keeps its rhythm while the pulsar fades from direct view: the clock keeps its phase, the torque keeps tracking the accretion flux, and the beam keeps its shape. No free precession of the neutron star is required by the current monitoring data.

In this letter, we demonstrate that employing the Lens Aperture as a Knife-Edge (LAKE) with two identical LEDs has the distinct advantage of obtaining schlieren images along two cutoff orientations. It is possible to control LED brightness to obtain a desirable weighted sum of both horizontal and vertical schlieren cutoff images on each camera frames. Powering the LED sources by an affordable and programmable Arduino board enables a rapid switch between the horizontal and vertical schlieren cutoff that is exactly synced to the camera frames, resulting in crosstalk-free schlieren image sequence. Furthermore, with a little postprocessing, this monochrome image series can be easily transformed into a colorful sequence of bidirectional schlieren images.

ShadowMiner v1 is a system that automatically discovers research problems and generates hypotheses from AI papers. It is a nine-stage pipeline. It structures documents into a knowledge graph and finds graph gaps in it - structural blind spots in research. These graph gaps are included in the LLM generation prompt. Each generated hypothesis is then verified by checking whether it is already covered by existing research, scoring its quality, and checking that the facts it relies on are accurately drawn from its sources. This report does not propose a new generation or evaluation technique. It describes our experience of implementing and applying ideas from prior work, and measuring whether each one actually contributed.

Machine learning-based weather prediction (MLWP) has achieved strong performance in global weather forecasting. Recent Hierarchical Equal Area isoLatitude Pixelation (HEALPix)-based methods use the HEALPix (HP) grid to avoid area distortion near the poles of conventional latitude-longitude (LL) grids. However, existing HP-based approaches often use pointwise mapping methods and process HP pixels within separate base faces or local windows. Consequently, the mapping may introduce reconstruction errors and cross-face communication depends on handcrafted boundary handling or shifted windows. We propose the Spherical Spiral Scanning Network (S$^3$N) to address both limitations. First, L2Proj provides a bidirectional method for mapping atmospheric fields between the LL and HP grids through an $L^2$ projection of their continuous finite-element representations. Second, the Attention-Guided Quad-Spiral State-Space Scanning (AQSS) block uses cross-latitude attention to guide selective state-space updates along four global pole-to-pole spiral paths. This design enables continuous information propagation across HP base-face boundaries without additional boundary-processing mechanisms. Experiments show that S$^3$N achieves better results at 4-, 7-, and 10-day lead times, and exhibits slower error growth in long-range forecasting.

Calibrating microscopic traffic models for digital twins is an expensive black-box optimization problem: tuning car-following and lane-changing parameters requires a full simulation run, affording only a tight budget per recalibration window. Matching raw trajectories yields a rugged objective that sparse surrogates cannot learn, reducing sequential acquisition to near-random probing. We present FLAT, which couples what to optimize with where to sample next. An eight-dimensional behavioral fingerprint smooths the parameter-error landscape, making the objective learnable from a few dozen samples; annealed lower-confidence-bound (LCB) acquisition then spends each remaining run where it most reduces error. The surrogate, interchangeable among a Gaussian process (GP), random forest (RF), or multi-layer-perceptron (MLP) ensemble, plugs into the same LCB loop. Across six heterogeneous real-world scenes, FLAT-GP achieves the lowest scene-averaged behavioral error, winning 6/6 scenes against SPSA, GA, and CMA-ES and 5/6 against TPE under the matched budget. Some baselines need up to 4.4 times more simulations to match. Ablations show objective choice shifts final behavioral error by 81% on average, removing sequential LCB raises the six-scene mean by 20%, and surrogate choice shifts it by at most 4.2%, confirming gains trace to objective geometry and sequential allocation rather than surrogate capacity.

Recent advances in generative world models have increased interest in digital models that reproduce both the appearance of real objects and their response to physical interaction. Three-dimensional reconstruction techniques, including 3D Gaussian Splatting, capture detailed surface geometry and appearance from images and videos. However, extending these representations beyond plausible animation to mechanically interpretable models for constitutive behavior, boundary conditions, and inverse parameter identification remains less explored. In this work, a differentiable Lagrangian-coupled 3DGS-smoothed particle hydrodynamics (SPH) model is proposed for forward simulation and inverse analysis of deformable solids. The observed object is first reconstructed from multi-view calibrated visual dataset as a 3DGS rendering model. An envelope-based procedure then generates an independent SPH support for the solid-mechanics model, avoiding the direct use of rendering primitives as mechanical particles. A reference-configuration Lagrangian transfer maps SPH deformation to Gaussian positions and covariances, thereby coupling the physical model and the image observation model while preserving a differentiable computational path. The SPH formulation supports linear elastic, hyperelastic, and Kelvin–Voigt viscoelastic responses, together with fixed, free, and Robin-type boundary conditions. Numerical studies validate the SPH response against finite-element results, assess accuracy and efficiency against a conventional model using Gaussian centers as surface SPH particles, and demonstrate forward simulations on beam, bridge, and liver-shaped examples. Inverse analyses further estimate constitutive and boundary parameters from rendered deformation observations, including noisy cases, demonstrating the feasibility of the proposed model for mechanics-based parameter identification from image data.

Remote pose estimation of noncooperative Unmanned Aerial Vehicles (UAVs) from imagery is critical, as they cannot be influenced or instrumented in advance. Deep-learning-based approaches offer a promising solution; however, their development is constrained by the cost and difficulty of acquiring large-scale real-world datasets with accurate pose labels. Synthetic imagery provides an alternative, but models trained on synthetic data must overcome the synthetic-to-real domain gap to generalize to real-world imagery. This work investigates the inherent synthetic-to-real generalization capability of a Vision Transformer (ViT)-based model for monocular UAV pose estimation. The proposed approach employs a self-supervised DINOv2 backbone and is trained exclusively on labeled synthetic imagery while being evaluated on labeled real-world imagery. Pose ambiguity-aware strategies are incorporated during training and inference to address ambiguities arising from the projection of a three-dimensional target onto a two-dimensional image plane and from target symmetries. An $\alpha$-$\beta$ filter is further integrated during inference to improve pose estimations. To assess the model under operational requirements, it is evaluated in terms of Mean Angular Error (MAE) and inference time, both before and after filtering, using a real-world dataset containing 77,077 labeled UAV images. Before filtering, the model achieves an MAE of $19.18^{\circ}$ and an inference time of $13.25$ ms, whereas after filtering, these values are $8.74^{\circ}$ and $13.42$ ms, respectively.

Seismic hazard and structural analyses require broadband ground-motion time histories, yet the high-frequency band (above 1~Hz) is the hardest to obtain: physics-based simulations are too costly at these frequencies, and stochastic methods lose the deterministic phase, directivity, and long-period coherence that control structural demand. We present HighFEM-2, a latent denoising-diffusion waveform model trained with rectified flow matching on a multi-modal diffusion transformer. It synthesizes three-component acceleration up to 50~Hz, conditioned on a low-frequency waveform below the source corner frequency and on magnitude, hypocenter and station location, and $V_{S30}$. The low-frequency waveform supplies the deterministic source, path, and directivity, so the network learns only the regional high-frequency scattering. Conditioning each station on its own waveform lets HighFEM-2 generate records for all stations of an earthquake that share its source and low-frequency path terms. Trained on 195{,}138 Italian INSTANCE records ($M_w$ 1.4–6.5, within 200~km), HighFEM-2 produces realistic waveforms: it reproduces observed peak amplitudes with small bias ($0.042 \pm 0.088$ log units for PGA, $0.057 \pm 0.090$ for PGV) and matches observed envelopes, Fourier spectra, and integrated velocity and displacement. The conditioning waveform can come from a simulation or a filtered record, and ensembles of realizations sample the high-frequency variability given that input. HighFEM-2 thus provides fast, regionally calibrated broadband time histories for hazard and risk assessment, including near-fault scenarios where recordings are scarce.

Delayed Streams Modeling (DSM) enables LLM-based streaming automatic speech recognition (ASR) by aligning acoustic and text streams on a common timeline. DSM adds the padding token <p> and the word-start token <w> to the LLM vocabulary and predicts them together with normal text tokens using the same softmax. We first show that <w> can be removed while maintaining competitive recognition performance. Based on this result, we propose Factorized DSM (F-DSM), which separates the waiting probability for <p> from the distribution over the original LLM vocabulary. This factorization removes ASR-specific tokens from the text prediction space and allows the large-vocabulary softmax to be skipped on waiting steps. Experiments on the Corpus of Spontaneous Japanese and LibriSpeech show that F-DSM achieves better recognition performance than DSM. It also greatly reduces GPU memory use while maintaining similar training throughput, provides a small inference speed improvement through softmax skipping, and reduces the degradation in text-only perplexity observed with DSM.

This paper studies intrinsic path following for fully actuated rigid-body attitude dynamics over a broad class of attitude paths, namely arbitrary smooth embedded closed curves on $\mathsf{SO}(3)$. We construct a tangential–transverse state decomposition on a tubular neighborhood using normal and pullback bundle geometry. The resulting second-order dynamics admit a smooth feedback transformation that independently assigns tangential and transverse covariant accelerations. A transverse proportional–derivative law renders the path-following manifold invariant and locally uniformly exponentially stable on an explicit forward-invariant neighborhood, while leaving the tangential dynamics freely assignable. This allows independent regulation of progression along the path, stabilization of a selected attitude on the path, or tracking of a time-varying motion. For a representative fiber path, the tubular coordinates are obtained in closed form up to the natural cut-locus obstruction, and simulations illustrate point stabilization and trajectory tracking.

We study the effect of repulsive interactions on skin modes and entanglement in the unidirectional Bose–Hubbard model. Although the ground state of this integrable model undergoes a superfluid–Mott transition with suppressed skin accumulation, the corresponding change in entanglement depends on the selected many-body state. We compare the ground state with the steady state, defined as the largest-imaginary-energy eigenstate in the zero-momentum sector. For sixteen sites at unit filling, we find that the ground-state density approaches uniform filling and its entanglement decreases toward the Mott limit, whereas the steady states exhibit a nonmonotonic entanglement crossover. At intermediate interaction ($U/t=10$), the second Rényi entropy of the steady state develops four spatial maxima. We show that these maxima are accounted for by competing subsystem-number sectors within a dominant configuration manifold of four doublons and four holons, but not by the subsystem particle-number variance alone. The dominant global onsite-pair manifold retains the same entropy peak structure. At strong interaction, projection onto the one-doublon–one-holon manifold closely reproduces the full entropy profile and its two edge peaks. Furthermore, we demonstrate that the many-body winding about the origin remains nonzero at strong interaction; an exact two-site example shows how it becomes ill-defined in the singular projected $U/t\to\infty$ limit. Our results distinguish ground-state skin suppression from state-selective entanglement and point-gap topology.

Let $\pi:G\to \mathcal{U}(L^{2}(G/P,\nu))$ be the boundary representation of a non-compact connected semisimple Lie group $G$ with finite center on its Furstenberg-Poisson boundary $(G/P,\nu)$. Let $\Gamma\subset G$ be a lattice in $G$ (uniform or not). We show that for any continuous functions $\varphi,\psi,\varphi',\psi'$ on $G/P$, $$ \lim_{n\to\infty}\frac{1}{|\Gamma_{n}|}\sum_{\gamma\in \Gamma_{n}}\frac{\langle\pi(\gamma)\varphi,\psi\rangle\overline{\langle\pi(\gamma)\varphi',\psi'\rangle}}{\Xi^{2}(\gamma)}=\langle \varphi,\varphi'\rangle\overline{\langle\psi,\psi'\rangle}, $$ where $\Gamma_{n}$ is a ball in $\Gamma$ with radius $n$ relative to a natural length function and with center the identity element of $G$, and $\Xi$ is the restriction to $\Gamma$ of the Harish-Chandra function of $G$. As a corollary, we deduce that when the real rank of $G$ is one, then the analogous convergence holds for any $\varphi,\psi,\varphi',\psi'$ in $L^{2}(G/P,\nu)$ if and only if the lattice $\Gamma$ is uniform.

Reliable language models should resist unsupported user pressure while effectively using objective contextual information. However, models may exhibit sycophancy by yielding to unsupported user pressure or contextual stubbornness by failing to update their answers when relevant contextual information warrants revision. Evaluating interventions for these failures separately can obscure whether mitigating one failure exacerbates the other. To assess this trade-off, we introduce CoPE-Bench with six conditions per question: a neutral baseline, correct or incorrect user pressure, contextual information consistent with or conflicting with the neutral answer, and a joint condition combining incorrect claims with conflicting contextual information. To regulate the influence of user pressure and contextual information, we propose SPAE (Suppressing Pressure, Amplifying Evidence), a training-free framework that uses the model's own judgments to identify relevant tokens, suppressing user pressure and amplifying contextual information through token-level attention steering. On average across five backbones, SPAE reduces pressure following by 18.8 percentage points and increases joint-condition updating by 5.5 percentage points relative to the strongest baseline in the main comparison. In two-turn dialogue, it improves joint-condition updating by an average of 13.2 percentage points over the strongest prompting baseline. The source data and codes can be found at https://github.com/03Grant/sycophancy-and-stubbornness.

Correction-based offline preference pipelines commonly treat model failures only as rejected responses under the original prompt. This supervision is incomplete for boundary failures: responses that violate the given instruction yet coherently satisfy a nearby intent or constraint setting. We introduce Bidirectional Preference Synthesis (BPS), a data-construction method for standard Direct Preference Optimization (DPO) that makes this missing prompt dependence explicit. For each validated boundary failure, BPS keeps the conventional forward pair under the original prompt and adds a reverse pair under a synthesized achieved prompt, so the same response is rejected where it is wrong and chosen where it is right, without changing the DPO objective, training a reward model, or requiring online sampling. On Qwen3-4B-Instruct-2507, BPS preserves original-side pairwise ranking while raising achieved-side ranking accuracy from 6.8% to 62.3% on held-out crossed anchors, with a similar shift under a Kimi-K2.6 cross-teacher probe. A blind human audit supports the intended reverse preference direction, and downstream evaluations show the clearest separation from Forward-DPO in multilingual multi-turn instruction following, with consistent capability-retention patterns on agentic, tool-use, and code checks.

Satellite and street-view imagery see different parts of a damaged building: roofs from above, facades and debris from the ground. Average accuracy hides whether fusion methods use this complementary evidence. We study the conflict cases, the samples on which independently trained single-view models disagree, as a diagnostic subset. On paired collections from the 2025 Eaton wildfire and Hurricanes Ian and Milton, conflicts make up 9.6-33.2% of the test data. Fusion improves conflict accuracy over either view, yet an oracle that selects the correct single view per conflict remains 0.19-0.32 above the best evaluated fusion method. A linear fusion gate over building-visibility features, calibrated confidences, and disagreement recovers part of this headroom where ground photographs are aimed at the building: on Eaton it beats end-to-end fusion by +0.072 conflict accuracy p < 10^-4 and calibrated averaging by +0.051 p = 0.0001; on the panoramic hurricane datasets it is indistinguishable from end-to-end fusion. Cropping panoramas toward the building raises the conflict gain of fusion, whereas random crops of the same size do not. The spatial density of conflicts, computed without labels, correlates with tile-level wildfire damage Spearman r = 0.615, 25 tiles. Conflict cases thus show how much complementary evidence fusion uses, which view it carries under each hazard and capture setting, and where current methods fall short.

Pretrained flow models are now widely used as generative priors in science and vision, where inference-time guidance enables test-time constraints without retraining. Existing methods use projection, posterior sampling, or iterative optimization of the generative trajectory. We propose LyapuFlow, an alternative based on Lyapunov feedback control. At each sampling step, LyapuFlow predicts the terminal sample towards which the current flow is evolving, and evaluates the constraint violation on this prediction. Then, we compute the minimum-norm control that satisfies a prescribed Lyapunov decrease condition. The resulting control remains inactive when the uncontrolled dynamics already reduce the constraint violation at the prescribed rate. Otherwise, it provides a corrective update within a feedback trust region that prevents the control from dominating the pretrained dynamics. We demonstrate LyapuFlow in both data and latent spaces, outperforming alternatives spanning different mechanisms for test-time constraint enforcement in scientific machine learning and image inverse problems.

The main aim of this article is to obtain explicit formulas for the root multiplicities of a class of finite-rank symmetrizable Borcherds-Kac-Moody algebras g(A) whose real part is of type A. The Dynkin diagram of each such algebra is obtained from the disjoint union of a Dynkin diagram of type A and a finite simple graph G_im, whose vertices correspond to imaginary simple roots, by choosing one vertex in each and joining the chosen vertices by a single edge of multiplicity c>0. More generally, we consider the same construction with the type A diagram replaced by the Dynkin diagram of an arbitrary finite-rank symmetrizable Kac-Moody algebra g_re. The resulting symmetrizable Borcherds-Kac-Moody algebra g(A) has real part g_re and imaginary part represented by G_im. Our starting point is a graph-theoretic description of the quotient of the denominator products of g(A) and g_re. This quotient is a multivariate independence polynomial of G_im, with the variable corresponding to the attachment vertex weighted by normalized highest-weight characters of g_re. We prove that the coefficients of powers of this quotient factor into a generalized chromatic polynomial and a tensor-power weight multiplicity. Using this, we obtain explicit formulas for all root multiplicities with nonzero imaginary component in terms of generalized chromatic polynomials of G_im and weight multiplicities in tensor products of highest-weight modules of g_re. When G_im is chordal, these formulas specialize to finite divisor sums involving products of binomial coefficients; paths and complete graphs provide particular examples. In particular, we obtain explicit formulas for the root multiplicities of g(A) when its real part g_re is of type A.