Papers

In aerial RGB–IR object detection, effectively exploiting complementary information across modalities is critical for robust perception under complex illumination and environmental conditions. Existing multimodal detectors mainly focus on spatial-domain interaction or frequency-specific feature enhancement, while the cross-modal interaction patterns of different frequency components remain insufficiently explored. Moreover, spectral discrepancy itself may contain both useful complementary cues and unreliable modality-specific responses, making indiscriminate frequency fusion suboptimal. To address these issues, we propose FoCal, a frequency-oriented framework for aerial RGB–IR object detection. First, a Frequency-Aware Dual-Domain Calibration (FADC) module is developed to explicitly model frequency-dependent cross-modal interaction. Low-frequency components are collaboratively consolidated into a shared structural consensus, whereas high-frequency components preserve modality-specific information through selective cross-modal exchange. The resulting frequency-aware cues are further transferred to the original feature domain to regulate cross-modal calibration. Second, we introduce a Discrepancy-Guided Spectral Modulation (DGSM) module, which characterizes cross-modal spectral imbalance using confidence-weighted relative amplitude discrepancy and transforms it into a bounded signed gate for adaptive enhancement, preservation, or attenuation of the joint multimodal spectrum. Extensive experiments on DroneVehicle, ESCVehicle, and ATR-UMOD demonstrate the effectiveness of FoCal, yielding $\mathrm{mAP}_{50}$ values of 83.5%, 54.8%, and 64.6%, respectively. Meanwhile, with only 3.0M parameters, FoCal achieves 113.6 FPS while preserving leading detection accuracy, highlighting a favorable accuracy–efficiency trade-off. Code is available at {https://github.com/universeliang/FoCal.

Let $K$ be a hyperbolic knot and let $\rho_n$ be the $n$-dimensional irreducible representation induced from a lift of its holonomy representation. Motivated by Goda's asymptotic volume formula and the complexified Volume Conjecture, we study whether the higher-dimensional twisted Alexander invariants associated with $\rho_n$ detect the complex volume of the knot complement. We compute $$ \frac{\pi}{2} \log \left( \frac{A_{K,n-2}(1)A_{K,n+2}(1)} {A_{K,n}(1)^2} \right) $$ for all hyperbolic knots with at most six crossings. Our numerical experiments indicate that these values approach $$ \operatorname{Vol}(S^3\setminus K) +i\,2\pi^2\operatorname{CS}(S^3\setminus K) $$ modulo $i\pi^2\mathbb{Z}$. Based on these computations, we propose a complexified analogue of Goda's asymptotic volume formula.

In cross-lingual zero-shot text-to-speech, the accent of the reference leaks into the target speech. We propose accent analogy guidance (AAG), a training-free sampler term that subtracts an accent direction estimated from the model's own predictions for one synthetic voice rendered in both languages, so the voice cancels and only the accent remains. By a blind LLM accent judge on real dubbing data, reweighting classifier-free guidance between reference and text, and its variants, stay near one identity-accent trade-off curve; we score a method by its speaker similarity above that curve at equal accent ($\Delta$SIM). Across four open TTS models AAG lies above the curve: on OmniVoice $\Delta$SIM is +0.11 to +0.27 on three test sets (accent 3.51 to 4.28 on a 1-5 scale at speaker similarity 0.29, where reweighting keeps 0.02); MaskGCT and CosyVoice 2 also lie above their curves, and on F5-TTS it is more native than any reweighting setting. An LLM-free language-ID measure and a twelve-listener panel agree. A premise test and the reach of a model's own curve indicate in advance whether and roughly how much AAG can gain, predicting the one model where it gains nothing (X-Voice).

For every $n\ge2$ and $1<p<\infty$, we show an existence of a nonconstant homogeneous entire $p$-harmonic function whose only critical point is the origin. The angular parts of these functions are axially symmetric and even across the equator. To establish the solution we use contraction argument at the pole in conjuction with a shooting argument. It follows that in every dimension $n\ge3$ and for $p<2$ sufficiently close to $2$, the natural gradient has pointwise Hölder exponent strictly below one at the origin. Such examples settle the remaining scalar case of the conjecture of Balci, Diening, and Weimar by disproving both the $C^1$ assertion and the linear $L^2$ mean oscillation estimate.

Audio-Visual Semantic Segmentation (AVSS) aims to identify, segment, and classify sound-emitting objects in video frames. Previous Transformer-based AVSS approaches largely inherit design principles from image segmentation models. Recent studies show that these image segmentation models contain redundant components that contribute little to the segmentation performance. Following this insight, we propose Encoder-only Audio-Visual Segmentation (EASE). EASE runs at up to 365 FPS, 3x faster than prior State-of-the-Art (SotA) AVS models at comparable accuracy, and trains in under 11 GPU-hours. Furthermore, we achieve SotA AVSS performance across different backbones and input resolutions. Our results demonstrate that AVSS can be both simpler and faster, providing a scalable foundation for future research and real-time applications. Code, model weights, and samples are available at https://ease-avs.notion.site

For an $[n,K]$ Reed–Solomon code, the covering radius is $n-K$. Gandikota, Ghazi, and Grigorescu proved deterministic NP-hardness of bounded-distance decoding when the decoding radius is $d$ below the covering radius for every $1\le d\le c\log n/\log\log n$, where $c>0$ is an absolute constant. We prove that, for every fixed rational $0<\alpha<1/2$, bounded-distance decoding is NP-complete under deterministic polynomial-time many-one reductions over explicitly represented finite extension fields for the additive gap $d=\lfloor n^\alpha\rfloor$ below the covering radius. The hard codes have odd block length~$n$, dimension $K=(n+1)/2-d$, decoding radius $(n-1)/2$, and rate tending to $1/2$. The alphabet size is subexponential in the evaluation set size: for a fixed $0<\eta<1$ depending only on $\alpha$, it is $2^{\Theta(n^\eta\log n)}=2^{o(n)}$. The proof passes through moments subset sum on $n-1$ nonzero field elements, with required subset size $(n-1)/2$ and $d$ prescribed moments. The arithmetic ingredient is a uniform positive-completion theorem over prime fields $\mathbb{F}_q$ with $q\ge d^{2+\rho}$, for any fixed $\rho>0$. A sharper form follows from a higher-dimensional point-count estimate based on Deligne's theorem; the weaker form used in our reduction is proved more elementarily using additive-character orthogonality, the one-variable Weil bound, a moment identity of order $2d$, and Newton identities. A universal completion pool, an extension-field quotient construction, and a deterministic linear-size simultaneous power condenser complete the reduction.

The constant-Q transform (CQT) represents audio on a logarithmic frequency axis. Its nonstationary Gabor formulation is exactly invertible, but the unequal numbers of time coefficients in its bands complicate GPU computation. An exact factorisation combines spectral selection, conjugation, windowing, and reordering into a fixed map between one packed Fourier transform and the shorter band inverse transforms. The factors give waveform reconstruction, real adjoints for backpropagation, and bounds on arithmetic depth and block width; overlapping slices permit streaming with bounded memory. Tests on two GPU models show that Flash-CQT reduces analysis-synthesis round-trip time by factors of two to eight relative to a baseline computing the same CQT. The proposed implementation also uses over 30% less peak temporary workspace and reaches a negligible reconstruction error, with a signal-to-noise ratio of about 130 dB, in single-precision floating-point arithmetic. These advances make Flash-CQT a practical, computationally efficient front end for spectral analysis and modern audio machine-learning systems.

LiDAR place recognition is a key front end for loop closure and global relocalization, yet indoor retrieval remains difficult when attitude or sensor mounting height changes between mapping and query sessions. Scan Context stores the maximum height in each polar cell; indoors, broad ceilings can suppress the lower and mid-level geometry that distinguishes adjacent rooms and corridors. We present UpDown-SC, a training-free polar descriptor that first canonicalizes gravity and then represents two complementary surfaces: the upper envelope of lower/middle structures and the lower envelope of overhead structures. Their physical split is estimated once from a cell-balanced map height distribution and reused by every query. A mask-aware, non-uniform two-channel distance retains discriminative lower-level evidence while limiting sensitivity to its cross-session variation, without treating unobserved cells as zero-height measurements. Conventional Scan Context shortlisting and circular yaw alignment are retained, so retrieved hypotheses directly initialize geometric verification. Experiments across repeated indoor sessions, mounting-height changes, mixed outdoor-to-indoor trajectories, and an outdoor transfer sequence show more reliable first-choice retrieval on the indoor and mounting-height-varied sessions. A paired test finds a significant gain over Scan Context on the in-house sessions. UpDown-SC also gives the best or second-best F1max and AUPR under threshold-based acceptance while retaining a lightweight CPU front end. Continuous replay confirms that the retrieved hypotheses support metric prior-map localization. Code and evaluation artifacts: https://github.com/jiejie567/updown-sc.

Significant research effort has been directed in recent years towards establishing both asymptotic and non-asymptotic convergence guarantees for two-timescale actor–critic algorithms, where the actor recursion is run on a slower timescale than the critic recursion. This work derives a uniform all-time concentration bound for the actor–critic algorithm with function approximation in the long-run average-reward setting. This bound helps us analyze the behavior of the actor parameter with high probability. We show that, after some finite time, the actor parameter enters a safe region and remains within it thereafter with high probability. Specifically, with probability at least $1-\epsilon_1-\epsilon_2$, the actor error $\Vert \theta_k-\theta^{*}\Vert$ is $O\left(\frac{n_0^{3/4}}{k}\frac{1}{\sqrt{\epsilon_2}}+\left(\frac{1}{n_0}\right)^{1/4}\log^{1/4}\left(\frac{1}{\epsilon_1}\right)+\left(\frac{1}{n_0}\right)^{1/4}\right)$ for all $k\geq n_0$ and sufficiently large $n_0$. We also present experimental results demonstrating that the aforementioned actor error diminishes with the number of actor-parameter updates.

Distribution Matching Distillation (DMD) enables high-quality diffusion sampling in only a few steps, but its optimization dynamics remain dominated by coarse, low-frequency signals, delaying the recovery of fine-grained details. We identify a pronounced concentration of spectral amplitudes at low frequencies in the DMD directional error, where dominant low-frequency components overwhelm weaker mid- and high-frequency signals. To address this issue, we propose Spectral Amplitude Purification for Distribution Matching Distillation (SAP-DMD), a plug-and-play approach that adaptively modulates the amplitude spectrum of the DMD directional field. By suppressing the dominant tail of the amplitude spectrum, SAP-DMD reduces low-frequency dominance and promotes more effective recovery of fine structures and textures. Experiments on PixArt-$\alpha$, SD3, and SD3.5 demonstrate that SAP-DMD accelerates training convergence and improves generation quality under both 2-step and 4-step sampling.

Modeling the full inspiral-merger-ringdown signal requires combining analytically controlled inspiral physics with the nonlinear strong-field information supplied by numerical relativity. We represent the numerical relativity contribution beyond an analytic inspiral waveform as residual amplitude and phase corrections. Retaining the leading-order frequency-domain amplitude and the $3.5$ Post-Newtonian TaylorF2 phase, we use a Kolmogorov-Arnold network to learn these residual corrections from SXS waveforms. After training, the learned corrections are stored as explicit spline functions, so waveform evaluation no longer requires the network itself. On $75$ simulations excluded from training and model selection, the model achieves a median flat-noise mismatch of $2.7\times10^{-5}$. Our results demonstrate a machine learning driven waveform modeling strategy in which numerical relativity augments, rather than replaces, analytically known waveform structure.

Recent stellar-kinematic and neutral-hydrogen observations indicate a dark matter deficit within the central tens of kiloparsecs of nearby galaxies that grows systematically with stellar mass relative to hydrodynamical simulations. We show that this mass dependence can arise from exothermic inelastic dark matter with strongly velocity-dependent scattering. Two nearly degenerate dark matter states interact through vector and scalar mediators with opposite-sign contributions and unequal ranges. The resulting coupled-channel dynamics suppresses $s$-wave conversion at low velocity while retaining a $p$-wave enhancement at several hundred $\rm km/s$, where down-scattering injects kinetic energy and lowers central dark matter densities. At dwarf velocities, conversion remains suppressed while elastic scattering can still drive core formation consistent with the observed dwarf-clustering pattern. A late dark-sector phase transition, along with the low-velocity suppression, preserves a large excited-state fraction until halo formation. Using representative halos spanning the four observed stellar-mass bins, we find that the model accounts for the inferred dark matter deficit in all four bins, including its systematic growth with stellar mass, while its impact weakens toward cluster velocities. The growing discrepancy with hydrodynamical simulations may therefore offer a glimpse of previously hidden dark matter microphysics.

We study the existence and nonexistence of meromorphic solutions of nonlinear differential equations involving differential polynomials and finite sums of exponential-type functions with entire coefficients of controlled growth. Employ ing Nevanlinna theory, we obtain structural results showing that any admissible meromorphic solution must reduce to an exponential-type function governed by the dominant exponential term. The obtained lemma and main result strengthen several existing results in nonlinear complex differential equations with exponential terms.

Industrial robots are rarely used for machining tasks due to their limited path accuracy. This accuracy is mainly limited by inaccuracies in the drive trains. Compliance and transmission errors occur in the joint gearboxes. While transmission errors have been extensively studied for strain wave gears, there is little research on these errors in cycloidal drives. This gearbox type is commonly used in industrial robots for medium to heavy payloads. It is proposed to model the mainly periodic transmission errors using a Fourier series where amplitude and phase are defined as a polynomial function of the main influence factors load-torque, velocity, and temperature. Measurements of the transmission errors were conducted using an experimental setup representing a single robot joint. In the evaluation of the measurement data, harmonic frequencies were related to mechanical properties of the cycloidal drive. These frequencies were used to identify the parameters of the polynomial Fourier series model. Compared to validation measurements, the derived model shows an average root mean square error of 0.026 mrad. It is proposed to use the output of the resulting model in a feedforward control approach to compensate the transmission errors and to increase the path accuracy of industrial robots.

Let $\nu$ be a Krull valuation of arbitrary rank on a field with valuation ring $R_\nu$, and let $\theta$ be a root of the irreducible polynomial $F(x)=(x^k+c)^m-ax^n\in R_\nu[x]$ and $1\leq n<km$. We establish necessary and sufficient conditions for the integral closedness of $R_\nu[\theta]$, expressed explicitly in terms of the coefficients $a$, $b$, and the integers $m$, $n$, and $k$. In particular, when $\nu$ is the $p$-adic valuation on $\mathbb{Q}$, our results yield criteria for determining the primes dividing the index $[\mathbb{Z}_K:\mathbb{Z}[\theta]],$ where $K=\mathbb{Q}(\theta)$ and $\mathbb{Z}_K$ denotes the ring of integers of $K$.

Recently, two-dimensional transition-metal borides (MBenes) have attracted substantial interest due to their promising properties for electrocatalytic applications. Here, we explore their potential as novel two-dimensional superconductors and topological materials through first-principles calculations on both pristine and surface-functionalized hexagonal MBenes. We conduct a thorough examination of the structural, electronic, phononic, superconducting, and topological properties of 36 compounds with formulas M$_{2}$B$_{2}$ and M$_{2}$B$_{2}$T$_{2}$ (M = Sc, Ti, V, Zr, Nb, Hf, Ta, Mo, W; T = F, O, OH). Our analysis identifies 21 superconducting MBenes, including four with critical temperatures (T$_{c}$) exceeding 10 K, with Ti$_{2}$B$_{2}$O$_{2}$ exhibiting the highest predicted T$_{c}$ of 24 K based on the McMillan formalism. For the two most promising compounds, Ti$_{2}$B$_{2}$O$_{2}$ and V$_{2}$B$_{2}$(OH)$_{2}$, we further solve the anisotropic Migdal-Eliashberg equations, obtaining zero-temperature superconducting gaps of 6.1 and 3.6 meV and anisotropic T$_{c}$ values of approximately 32 and 25 K, respectively. Symmetry-indicator-based analysis further reveals nontrivial normal-state band topology in several superconducting MBenes. These results indicate the coexistence of phonon-mediated superconductivity and nontrivial normal-state band topology within this material family, making MBenes promising platforms for future investigations of the possible emergence of topological superconductivity.

Reasoning-capable language models often produce long chains of thought when direct answers suffice, wasting inference compute. Dual-mode models offer both thinking and direct-answer modes, but typically leave mode selection to users. Automating this selection while improving responses under both modes is challenging because routing targets evolve with the policy, strong initial mode preferences can destabilize exploration, and sequence-level objectives entangle routing with response learning. We introduce CounterRoute, an online reinforcement-learning framework that jointly learns routing and mode-conditioned responses in one shared policy directly from a native dual-mode checkpoint, without method-specific SFT warm-up. Paired current-policy counterfactual rollouts estimate cross-mode routing credit, which is assigned only to the routing token, while within-mode GRPO trains response tokens. A paired-to-self-routed curriculum stabilizes early training with forced coverage of both modes before progressively transferring training to autonomous routing. Across nine benchmarks, CounterRoute better balances accuracy and efficiency than heuristic and learned adaptive-routing methods, improving macro-average accuracy at both Qwen3 scales while cutting mean generated tokens by 51% and 41%, respectively, versus always-thinking checkpoints. On instruction-following and commonsense benchmarks where direct answering is strong, think rates fall as low as 1% while response quality improves. Despite training only on math and synthetic instruction-following data, its routing and response quality generalize to held-out tasks in coding, science, knowledge, and commonsense.

European electricity trading in the EU operates as a constrained multi-layer system in which legal design, exchange microstructure, and network physics are executed jointly across forward, day-ahead, intraday, and balancing horizons. This paper develops a functional architecture for AI-supported trading that is aligned with market-coupling mechanics, cross-zonal transfer constraints, and compliance obligations under REMIT, MiFID II, MiFIR, and EMIR. The contribution is a formal system specification composed of a decision-state vector, residual-exposure accounting, constrained optimization objective, executable-action permission gate, and fail-closed AI control logic with auditable records. The analysis maps major Nominated Electricity Market Operator (NEMO) venues and related exchange operators into an operational venue topology and identifies where cross-border coordination fails in practice: interface-level timing, permission heterogeneity, and balancing-layer coupling. The resulting framework proposes how AI can be deployed as a bounded decision component inside regulated market operation with explicit governance, rather than as an unconstrained prediction layer.

Let $f$ be a regular endomorphism of $\mathbb{A}_{\mathbb{C}}^N$ and let $C\subseteq\mathbb{A}_{\mathbb{C}}^N$ be an irreducible curve. Suppose $C$ has an infinite intersection with the $f$-orbit of a point $x\in\mathbb{A}^N(\mathbb{C})$. Then the normalization of $C$ is isomorphic to either $\mathbb{A}^1$ or $\mathbb{G}_m$. We prove that $C$ is $f$-periodic in the latter case, as expected by the dynamical Mordell-Lang conjecture.

3D foundation models recover video cameras and geometry in one forward pass, but some of the strongest are up to scale. Joint people-scene reconstruction then requires two missing outputs: metric scale and persistent person identity. We ask whether one up-to-scale foundation representation can support both through lightweight adaptation. Exact metric labels are scarce, but unlabeled in-the-wild video is abundant. We use people in curated web video to initialise the solution: a posed metric body and 2D keypoints give an approximate, closed-form scale pseudo-label. These pseudo-labels pretrain a Scale Readout, which is then fine-tuned together with a lightweight adapter using exact metric supervision from standard real-video training splits. At inference the head predicts metric scale from foundation-model tokens, without the ruler or its teachers. For person identity, we probe the pretrained foundation model alone and find evidence that its intermediate query-key features encode person correspondence across frames. In most evaluated moving-person clips, a mid-layer token prefers that person over the vacated location and other people. A tiny projection reads this correspondence; together with metric pelvis motion and proposal confidence, it drives dustbin-aware Sinkhorn association of per-frame bodies. WildHSR combines both readouts to reconstruct metric cameras, scene and people from monocular video. Each window is predicted feed-forward; analytic association and Sim(3) composition connect windows. On EMDB-2, WildHSR is the first feed-forward method in the published comparison to beat the best optimization-based WA-MPJPE and RTE while leading feed-forward methods on all three world-frame metrics. On RICH, it leads feed-forward people-and-scene methods on WA-MPJPE and W-MPJPE. The complete pipeline runs at 10.1 fps on one GPU.

General-purpose machine-learning interatomic potentials (MLIPs) for organic reactions need to be accurate on both the minimum energy path (MEP) for static evaluation of basic properties and the broader configurational space for simulating reaction dynamics. Existing general datasets for gas-phase organic reactions rely on quasi-static relaxation that confines configurations to the MEP vicinity, so models trained on them could fail on direct molecular-dynamics trajectories; the gap is methodological, not a question of dataset size. We introduce a spatiotemporally resolved, system-independent collective variable (CV): Cartesian RMSD within randomly partitioned local domains against an expanding list of time-averaged reference geometries. The CV drives metadynamics as the main exploration engine, supplemented by structural relaxation towards transition state (TS) to augment the coverage around TS. Within a concurrent-learning workflow, this produces OpenRxn26, a dataset of 1.8~M DFT-labeled configurations covering neutral singlet unimolecular reactions in the H/C/N/O chemical space ($N_\mathrm{heavy} \leq 30$), containing reactive atomic environments underrepresented in community datasets. Trained on OpenRxn26, a DPA3 model (denoted DPA3_rxn) achieves transferable accuracy on barrier heights and reaction energies. On off-MEP reactive trajectories, DPA3_rxn is the only model in the benchmark suite to reach 1.0 kcal/mol energy accuracy compared with the labeling method, where the domain MLIP leading on static benchmarks degrades several-fold (e.g. MACE_OMol25), showing the insufficiency of quasi-static sampling and MEP-anchored benchmarks for guaranteeing dynamics reliability of MLIPs. OpenRxn26 thus provides MD-ready reactive training data for gas-phase neutral singlet organic reactions, verifying the generality and efficiency of the sampling strategy.

In this paper, we study the existence, nonexistence, and growth be haviour of solutions of certain nonlinear differential and differential-difference equa tions involving the term fnf'. Using tools from Nevanlinna theory, we establish several nonexistence results for finite-order entire solutions of equations involving ff'. In particular, we describe all the possible entire solutions of the type f2f' in which right hand side involves exponential polynomials. The results obtained by us improve several known results in the literature and we have them demonstrated through illustrative examples.

Accurate state estimation (tracing) of Deformable Linear Objects (DLOs) such as cables is a critical challenge for data centers, manufacturing, construction, homes, and surgery, where precise cable management directly impacts operational safety and efficiency. However, resolving the state of multiple monochrome cables amid foreground and background clutter poses challenges due to occlusions, overlap, and ambiguous crossings. We present Two-way Routing And Cable Estimation (TRACE), which combines bi-directional cable tracing with interactive perception primitives-Divergence Push and Cluster Dilation-to actively resolve ambiguities. Evaluation with 110 physical experiments suggests that TRACE can increase the percentage of cable length correctly traced in complex scenarios (with up to 4 cables and 40 crossings) from ~60% with the strongest prior method, HANDLOOM 2.0, to ~90%, outperforming RT-DLO, Nano Banana Pro, and ChatGPT 5.2 as well. For a trial run on a workstation with an NVIDIA GeForce RTX 4090 GPU, the average computation time is 0.4 seconds per cable. Project website: https://trace-paper.github.io/.

Fully continuous diffusion language models (dLMs) denoise continuous representations without intermediate discretization, then decode all response tokens in parallel at the final step. Their performance on challenging reasoning tasks remains less established than that of autoregressive (AR) LLMs and masked dLMs. We scale Embedded Language Flows (ELF) to mathematical reasoning and code generation on GSM8K, MATH-500, HumanEval, and MBPP. We introduce ELF-REG, which improves learning with representation alignment and entanglement (REPA+REG), where a frozen AR teacher supervises intermediate denoiser features and supplies a global representation that is jointly denoised with the response. ELF-REG-L achieves 55.96% pass@1 on GSM8K at 64 network function evaluations (NFE), and 13.39% on MATH-500 and 22.56% on HumanEval at 128 NFE. It outperforms the evaluated comparable-scale dLMs in pass@1 on GSM8K and code, and improves MATH-500 pass@1 from 10.55% for the ELF-L baseline to 13.39% with ELF-REG-L. Without few-step training, the same task-specific checkpoints support strong low-NFE performance through early-stop, which decodes an intermediate clean prediction without completing the denoising trajectory. At 16 NFE, ELF-REG-L reaches 41.21% HumanEval pass@10, outperforming recent continuous dLMs of comparable scale.

Medical Named Entity Recognition (NER) for low-resource languages remains a challenging task due to high linguistic variability and a scarcity of domain-specific annotated corpora. This work presents a comprehensive empirical benchmark evaluating three fine-tuned transformer encoders-BanglaBERT, multilingual BERT (mBERT), and XLM-RoBERTa-against GPT-4o mini under zero-shot and few-shot prompting configurations for Bangla medical NER. In contrast to prior studies that evaluated large language models on limited subsets of only 50 samples, we conduct a large-scale evaluation across the full test set of 3,179 samples, providing statistically robust and reproducible baselines. Our fine-tuned XLM-RoBERTa model achieves an F1- score of 0.5959, establishing a new state-of-the-art and surpassing the previously reported best result of 0.5848. Crucially, we demonstrate that the language-specific BanglaBERT model consistently underperforms its multilingual counterparts with an F1-score of 0.4937, indicating that pretraining domain diversity can outweigh language specificity in highly specialized clinical settings. Furthermore, we present a detailed per-entity-type analysis for this task, revealing that Medicine and Specialist categories are recognized with high reliability, achieving F1- scores above 0.83, while the Symptom category remains the most challenging with an F1-score of 0.4367 despite being the most frequent training class. Finally, fine-tuned transformer models outperform the optimal prompting configuration by a factor of 3.76, confirming that prompt-only pipelines remain inadequate for structured clinical entity extraction in low-resource language environments.

This paper investigates model predictive control (MPC) for switched systems subject to control and state constraints. A variable-horizon switched MPC approach is proposed. By steering the system state into a well-designed switching feasible set, the proposed method structurally decouples the dwell-time conditions from the MPC constraints, thereby relaxing the dwell-time requirements to match those of the unconstrained switched systems. Furthermore, algorithms are developed to construct this switching feasible set and characterize the domain of attraction, ensuring both persistent feasibility and closed-loop asymptotic stability. To further decouple the prediction horizon length from strict dwell-time bounds, advanced short-horizon switched MPC schemes are designed, which expand the overall domain of attraction. Simulations illustrate the efficacy of the proposed methods.

Multimodal training relies on image-text corpora collected from external sources, creating opportunities for attackers to poison the data. Stealthy attacks can preserve plausible image-text pairs while concealing the differences used by detectors, so apparently clean data can still redirect the trained model. We therefore ask which properties a poison set must preserve for the attack to remain effective. A small poison set must still exert enough collective influence during training to induce the attacker's target behavior. We analyze this influence in terms of how often an attack pattern occurs and how strongly the examples carrying it jointly affect the model. This analysis motivates six corpus-level features that examine cross-modal neighborhoods, recurring text, and changes after text-span erasure without training the victim model. We introduce TraceGuard, an adaptive rank-based filtering method that uses agreement among complementary feature rankings to identify suspicious examples. It refines the selected set through shared patterns and adapts the removal threshold to each corpus without knowing the attack or poison rate. Across 19 attack configurations spanning image-text learning, generative vision-language model fine-tuning, and encoder-transfer tests, TraceGuard removes an average of 98.4% of poisoned examples and 5.4% of clean examples. After training on the filtered corpora, the residual attack metric is at most 1% in 13 configurations. Matched-removal controls and ablations support the contributions of sample selection and adaptive removal. Stress tests also identify detection failures under adaptive attacks and unnecessary removal on poison-free corpora.

Variational quantum algorithm is a leading approach for quantum chemistry and many-body physics on noisy intermediate-scale quantum devices. The performance depends strongly on the structure of the parameterized quantum circuits. Quantum architecture search (QAS) can automate ansatz design. However, it requires repeated training and evaluation of many candidate circuits, leading to high evaluation cost. In this work, we propose a noise-aware quantum architecture search framework based on ZX-calculus topological reuse (ZX-QAS). The framework encodes the search space with a ternary Gray-code mapping and integrates a noise-aware quantum neural network with a ZX-calculus topological reuse mechanism. The effectiveness of the framework is validated through ground-state energy estimation tasks and one-dimensional transverse-field Ising model tasks under noisy conditions. The results show that the ZX-QAS exhibits stable convergence and remarkably reduces the cost of expensive evaluations under noisy conditions.

Electrically charged regular black holes in general relativity coupled to nonlinear electrodynamics are known to require a double-valued Lagrangian density $L(F)$, with a branch point at the radius where the electromagnetic invariant attains its extremum. We extend the construction of such black holes to an arbitrary number of spacetime dimensions $D\geq4$ and obtain a new family of solutions that unifies the Bardeen and Hayward black holes in a single analytic form, with a single nonlinearity parameter $\gamma$. We show that the Maxwellian weak-field limit is recovered if and only if $\gamma=D-3$, so that the Bardeen and Hayward black holes, which lack this limit in four dimensions, become Maxwellian in five and six dimensions, respectively. The matter sector is written in the dual $P$ framework as a Hamiltonian density $H(P)$ that contains only fixed coupling constants, so that the mass and the charge of the black hole appear as integration constants of a fixed theory; the regular solutions form a one-parameter subfamily in which the mass is tied to the charge. We locate the branch point of $\L(F)$ in closed form and find that it is always hidden inside the event horizon in four dimensions, whereas for $D\geq5$ it emerges from the horizon for near-extremal charges and becomes visible to a distant observer. The null and weak energy conditions are satisfied everywhere, the strong energy condition is violated only inside the horizon, and the dominant energy condition is violated at large distances only for $\gamma>D-3$. Finally, since the theory is fixed, the first law of black hole thermodynamics holds in its standard form; we derive the horizon potential in closed form, the extended first law with the couplings as thermodynamic variables, and the corresponding Smarr formula.

Post-hoc correction adjusts a forecaster that cannot be retrained, such as a foundation model, but a correction fitted where errors are stable can hurt where they shift. We aim for downside control: not much worse than the starting forecast. We combine the frozen forecaster, a static corrector and an online corrector on the simplex, using only losses that mature after the horizon. Across seven benchmarks and four base models, two of them foundation models, the worst deterioration over 28 pairs at the main horizon is 0.15% and gains reach 11.5%. On day-ahead load for seven European bidding zones it lowers mean MSE in all seven zones, while single correctors raise mean MSE by up to 102% where the published forecast is most accurate. Three empirical conditions on expert speed, stream length and outcome alignment, each fixed by a documented failure, delimit its scope. Learning from the provisional outcome improves four zones on the settled one; learning on the settled outcome restores all seven.