Papers

We study the Semi-Inclusive Deep Inelastic Scattering (SIDIS) cross-section at large values of hadron transverse momentum $P_t \gtrsim Q$, with $Q^2$ being the photon virtuality. This kinematic regime of SIDIS allows for a collinear factorization in terms of parton distribution functions (PDFs) and collinear fragmentation functions (FFs). On the other hand, at high energies, i.e., at small Bjorken $x$, the same SIDIS process can also be factorized within the Color Glass Condensate (CGC) framework in terms of eikonal dipole amplitudes. In this work, we perform a systematic comparison of the leading-order (LO) and next-to-leading-order (NLO) collinear and the LO CGC factorizations of this process based on the COMPASS and HERA data. We further provide comparisons of these two factorization frameworks for nucleon and nuclear large-$P_t$ SIDIS in the expected Electron Ion Collider (EIC) kinematics. We find, within the present theoretical uncertainties, that it might be difficult to distinguish between these two factorization schemes for nuclear SIDIS at the EIC. We also observe a significant dependence of the LO eikonal CGC predictions on the choice of longitudinal momentum fraction $x_g$, suggesting significant beyond-eikonal corrections for the EIC kinematics.

A new concept, termed virtual neural networks, is introduced, where the count of trainable parameters is kept constant, and scalability is attained purely through computational resources. This concept is an abstract framework that can be realized using any standard convolutional neural network. It merges siamese neural networks with a deep ensemble technique by generating numerous virtual models that share weights derived from a small set of physical models. The ensemble comprises up to hundreds of trained models simultaneously. All virtual networks take the same input, and their interconnected structure induces an internal distortion that boosts the entire ensemble robustness. The accuracy of the ensemble improves as the number of virtual networks increases, without changing the capacity. Virtual neural networks outperform larger capacity models, typical deep ensembles, and contemporary approaches like SWA and Masksembles. Additionally, the highest-performing individual model from the ensemble surpasses other models trained individually, even those with a greater number of parameters. Code: gitlab.com/EnginCZ/virtual-models-public

A Ševera bicomplex is a bicomplex $(M,d_{0},d_{1})$ whose differentials $d_{i}$ are Grothendieck differential operators of order $\leq i$ with respect to a given $A$-module structure on $M$, where $A$ is a graded commutative algebra. One also requires that $(M,d_{0})$ admits a contracting homotopy $K$ that is a differential operator of the same order as $d_{0}$. Under suitable assumptions, the differential $d_{2}$ of the second page of the spectral sequence associated with $(M,d_{0},d_{1})$ is a Batalin-Vilkovisky operator on $E_{2}$, so that when $E_{2}$ is a free rank 1 $A$-module, the graded commutative algebra structure of $A$ is enhanced to a Gerstenhaber algebra structure, and the choice of a $d_{2}$-closed basis element for $E_{2}$ further enhances this to a Batalin-Vilkovisky algebra structure. The prototypical example of this construction is Ševera's description of the Batalin–Vilkovisky algebra structure on the algebra of smooth functions of an odd symplectic manifold. Throughout the whole article we look at Batalin-Vilkovisky algebras through the lenses of Cartan calculus. Indications of a generalization to derived Cartan calculus are briefly discussed in the concluding section.

In the planted clique problem, one observes either an Erdős–Rényi graph on $n$ vertices or such a graph with a clique added to $k = k(n)$ vertices, and seeks to detect or recover the clique. It is widely believed that $k = \Theta(\sqrt{n})$ is the smallest clique size for which polynomial-time algorithms exist for these tasks. We develop new algorithms in this regime using color-coding to estimate signed subgraph counts, further accelerated with fast matrix multiplication. We first show that, for each $t \geq 1$, for $c(t)$ a constant associated to the order of growth of the number of connected graphs of treewidth at most $t$, cliques of size $k = \lambda\sqrt{n}$ planted in a random location with $\lambda > 1 / \sqrt{c(t)}$ can be detected and recovered in time $n^{t + 1 + o(1)}$. For instance, since $c(1) = e$, this recovers by counting signed trees the performance of the $\widetilde{O}(n^2)$-time message-passing algorithm of Deshpande–Montanari (2015) that succeeds when $\lambda > 1 / \sqrt{e} \approx 0.6066$. For $t \geq 3$, the exact value of $c(t)$ is not known, but lower bounds on it give a hierarchy of slower polynomial-time algorithms that succeed for smaller $\lambda$. We further show that the above algorithm for $t = 2$ can be implemented in time $n^{\omega + o(1)}$ for $\omega$ the constant of square matrix multiplication and succeeds when $\lambda > 0.3320$; under the folklore conjecture that $\omega = 2$, this runs in the nearly-linear time of the algorithm of Deshpande–Montanari while finding smaller cliques. Second, we show that the above algorithm for $t = 1$ can be combined with the boosting scheme of Alon–Krivelevich–Sudakov (1998) using rectangular matrix multiplication, giving improved runtimes for smaller $\lambda$. Taken together, our results achieve the best known tradeoff between runtime and signal strength $\lambda$.

We provide a clearer presentation of the Chair44 monotile, a new strongly aperiodic monotile in three-dimensional Euclidean space $E^3$.

Learning robot manipulation policies from human video demonstrations constitutes a promising avenue for scalable robot learning. However, comparing different human-to-robot (H2R) transfer methods remains challenging, as existing approaches are evaluated under different settings, including differing task suites, scene layouts, object instances, and amounts of robot supervision. To address this challenge, we present H2RBench, a Real2Sim benchmark for evaluating H2R transfer methods. H2RBench provides a standardized protocol built on real human video demonstrations and simulated robot demonstrations, and includes four manipulation tasks spanning diverse interaction requirements. We evaluate multiple representative H2R transfer methods, each adopting a different strategy for bridging the embodiment gap. Using H2RBench, we systematically characterize how each method scales with the amount of human demonstrations, revealing that methods differ substantially in their ability to leverage additional human data. We further show that simulation performance is broadly predictive of real-world robot performance, with an overall Pearson correlation of r = 0.89, Spearman correlation of h̊o} = 0.85 and Mean Maximum Rank Violation (MMRV) of 0.06 across method-task configurations. These results establish H2RBench as a practical and scalable benchmark for comparative H2R evaluation prior to real-world deployment.

We construct free immersions of $P^3 \times S^1$ and $S^3 \times S^1$ into $14$-space. These are the first examples of free immersions in critical dimension of closed manifolds that are not spheres, projective spaces, tori, or surfaces. The idea is to mimic De Leo's construction for $m$-tori with $m \leq 5$. De Leo constructs a map that is symmetric with respect to a representation of the Lie group formed by the $(m - 1)$-torus; we replace this group with $\mathrm{SO}(3)$. In our examples, the osculating determinant is constant.

We develop an adaptive localized orthogonal decomposition (ALOD) method for Helmholtz problems with parametrized loads. Under standard resolution, oversampling, and discrete-stability conditions, we establish, to our knowledge, the first wavenumber-robust residual a posteriori error estimate for a Helmholtz LOD discretization. The estimate is two-sided relative to the fine-grid solution and includes a computable bound for the corrector-localization error. In addition, we design a novel family-aware adaptive algorithm using additive Schwarz corrections to construct a shared regional enrichment space for Helmholtz problems with multiple loads. Energy-weighted proper orthogonal decomposition compresses this space. We establish stability and quasi-optimality of the enriched formulation, with bounds independent of the enrichment space. Numerical experiments demonstrate the effectiveness of ALOD for multiple loads and large wavenumbers.

We prove new bounds for Tverberg's theorem with tolerance. We show that $N = rt+\Theta_{d,r}(t^{1/2-1/(2d)})$, where $N$ is the smallest number such that any set of $N$ points in $\mathbb{R}^d$ has a partition into $r$ parts such that the convex hulls of the parts intersect even if we remove any $t$ of the points. We extend Tverberg's theorem with tolerance to families of hyperplanes in $\mathbb{R}^d$, and show that for any set of $rt + O_{d,r}(t^{1/2-1/(2d)}\sqrt{\log (t+1)})$ hyperplanes in $\mathbb{R}^d$ there exists a partition of them into $r$ parts such that the regression hulls of the parts intersect even if any $t$ hyperplanes are removed. Our bounds follow from establishing a connection between Tverberg-type results and discrepancy theory.

The purpose of this paper is to prove that every finite simple group of type $\mathsf B$ or $\mathsf C$ satisfies the inductive blockwise Alperin weight condition at every prime $\ell$ dividing its order. For ${\rm PSp}_{2n}(q)$, where $q$ is odd and $n\geq3$, a permutation lattice argument based on Conlon's induction theorem removes the unitriangularity assumption from earlier results on the inductive condition at odd nondefining primes. At $\ell=2$ in odd defining characteristic, errors and omissions in the underlying classification of radical $2$-subgroups affect an earlier parametrisation of weights for arbitrary blocks. We verify the required weight parametrisation for principal blocks and use Jordan reduction to establish the inductive condition for all blocks. At odd $\ell$ in even defining characteristic and rank at least four, we prove the inductive condition using generic weights and Jordan reduction, without assuming unitriangularity. For $\Omega_{2n+1}(q)$, where $q$ is odd and $n\geq3$, we remove the unitriangularity assumption at odd nondefining primes using Conlon's induction theorem. At $\ell=2$, we prove the stabiliser and extension property for Brauer characters of $\operatorname{Spin}_{2n+1}(q)$ that was assumed in earlier work on the inductive condition. We also give proofs of the inductive condition for the remaining sporadic groups $J_4$, $Fi'_{24}$, the Baby Monster and the Monster, without making a priority claim. Together with the previously established cases, these results imply that the blockwise Alperin weight conjecture holds at $\ell$ for every finite group each of whose nonabelian simple sections of order divisible by $\ell$ is of type $\mathsf B$, of type $\mathsf C$, or sporadic.

Let $K=\mathbb{F}_q(\theta)$, and let $\mathcal{Z}$ be the $K$-algebra generated by Thakur's multiple zeta values $\zeta_A(\mathfrak{s})$. We prove that $\mathcal{Z}$ is a polynomial algebra and construct an explicit polynomial basis of $\mathcal{Z}$ over $K$. We also determine the transcendence degree of $K[\zeta_A(\mathfrak{s}): \operatorname{wt}(\mathfrak{s}) \leqslant w]$ over $K$ for every integer $w \geqslant 1$, generalizing a result of Ngo Dac–Nguyen Chu–Pham. The proof uses Chang's grading theorem, the linear basis theorem proved independently by Chang–Chen–Mishiba and Im–Kim–Le–Ngo Dac–Pham, and the carry relations of Im–Kim–Ngo Dac. The key step is to show that suitable derivations obtained from deconcatenation and linear functionals descend through the carry relations to $\mathcal{Z}$.

Quantitative measurements of absorption circular dichroism (CD) at the single-particle level are essential for understanding how individual chiral nanostructures dissipate optical energy, yet broadband absorptance measurements remain experimentally challenging. Here, we introduce a wavelength-tunable integrating-sphere microscope that determines the absorptance of individual nanoparticles under right- and left-circularly polarized illumination through direct optical energy balance, enabling broadband measurement of the absorption dissymmetry factor, gabs. Measurements of achiral gold nanospheres and strongly anisotropic gold nanorods establish the apparent absorption CD baseline and demonstrate minimal detectable linear-dichroism-to-circular-dichroism (LD-to-CD) leakage under the experimental conditions. Applying this approach to 87 chiral gold helicoids synthesized using L-glutathione (L-helicoids) and 96 synthesized using D-glutathione (D-helicoids) reveals mean particle-level gabs values of opposite sign, with a statistically significant difference between the two populations. Individual particles nevertheless exhibit pronounced heterogeneity in response sign, magnitude, spectral position, and line shape. More than one-third of the particles in each population display opposite-sign absorption CD responses relative to their population-average tendency. Correlative SEM analysis of the L-helicoid population further shows that opposite-sign responses persist among isolated particles exhibiting the characteristic projected helicoid morphology, indicating that aggregation and gross differences in projected morphology are insufficient to explain the observed heterogeneity. These results establish broadband single-particle absorption CD spectroscopy as a direct probe of absorptive chiroptical heterogeneity and reveal particle-specific responses obscured by ensemble averaging.

Instruction-controlled speech synthesis and paralinguistic understanding are often trained independently, leaving reciprocal feedback between the two tasks underexplored. We introduce CycleSpeech, a framework that connects generation and understanding through a shared, structured voice profile that serves as a common target for supervision and reciprocal feedback. The forward cycle assesses whether synthesized speech expresses the intended attributes by comparing recovered and target profiles. The backward cycle evaluates whether profiles inferred from real speech can guide reconstruction of the source speaking style. To support both directions, we construct a bilingual dataset of 20,046 examples pairing instructions, target speech, speaker references, and structured profiles. Building on joint supervised fine-tuning, CycleGRPO alternates policy updates using reciprocal rewards grounded in profile consistency and speaking-style reconstruction. Fixed target profiles anchor feedback from the evolving counterpart. This procedure requires neither human preference annotations nor an additional preference-trained reward model. Evaluations on Chinese and English benchmarks show improved instruction adherence and profile recovery while maintaining competitive synthesis quality. Compared with Step-Audio-2-mini, CycleSpeech improves instruction-match accuracy by 4.50 and 10.06 percentage points in Chinese and English, respectively. Controlled ablations further support the contribution of cycle feedback to generation control. These results support structured voice profiles as an interface for reciprocal training between speech generation and paralinguistic understanding. An online demo is available at https://cyclespeech.github.io.

Localising artifacts in synthetic speech remains challenging, as most evaluation methods yield only global quality scores. This paper presents XSQ-AST, a framework that combines the SQ-AST speech quality model with WhisperX phoneme alignment and multiple saliency methods to produce temporally localised artifact diagnostics without model retraining. Saliency maps are projected onto continuous distributions via kernel density estimation and onto phoneme boundaries via phoneme-discretised saliency maps. A 40-participant listening test validated the framework across five perceptual dimensions. Attention Rollout, Attention Flow and an adapted GradCAM produced temporal distributions that correlated with listener highlights, with different methods best suited to different artifact types. An AUC-ROC analysis confirmed discrimination above chance.

Brain metastases are the most common intracranial malignancy, occurring in roughly 30% of patients with primary solid tumors and carrying a median survival near 5.9 months. Automated segmentation is critical for treatment planning and volumetric monitoring, but metastases are frequently small, numerous, and heterogeneous in size within a single patient. We compare a plain nnU-Net baseline, a Residual Encoder Large (ResEncL) variant, region-based training, and a Primus transformer model for BraTS-METS 2026 Task 1, using patient-grouped cross-validation to prevent leakage from the longitudinal UCSD subset. Primus (label-based) is our strongest individual model by aggregate DSC/NSD, achieving 0.710/0.761 (ET), 0.742/0.785 (TC), 0.683/0.689 (WT), and 0.531/0.436 (RC). ResEncL trails Primus on aggregate DSC/NSD but achieves substantially higher lesion-wise F1 (e.g. ET: 0.452 vs. 0.052); a probability-averaging ensemble of the two only partially preserves ResEncL's F1 advantage (ET lesion-wise F1: 0.064). We further report three postprocessing and label-reconstruction pitfalls we believe generalize beyond this challenge. Code is available at https://github.com/mahdiislam79/BraTS_METS_2026.

Professional photo finishing relies on both global adjustments and region-specific local edits guided by semantic masks, yet current automated methods handle this workflow only partially. We present PrismGPT, a Vision-Language Model (VLM) framework that produces structured, region-aware editing plans from a single input image without relying on commercial black-box tools. Training a VLM to simultaneously diagnose aesthetic deficiencies at both global and local levels while predicting precise editing parameters is challenging due to the vast combinatorial decision space. We address this through proxy-guided learning: two simpler proxy tasks – operation decomposition and region-aware aesthetic ranking – teach the foundational skills the model needs, while a competence-based dynamic scheduler automatically rebalances the multi-task training ratio, progressively shifting emphasis from the proxy tasks to the primary editing task as each skill is mastered. Crucially, all reasoning traces used for supervised fine-tuning are self-synthesized by the same base model, eliminating the need for a stronger external teacher. Experiments on MIT-Adobe FiveK and SPIRE, a new professionally retouched benchmark we introduce, show that PrismGPT achieves state-of-the-art results while using only ~6% of the training data compared to the previous best method.

A vanishing sequence $V_n$ of divergence-free matrix fields is one that is bounded in $L^1$ and is carried by open sets $A_n$ of vanishing volume. Recently established, Bouchitté's vanishing mass conjecture says that the directions such a sequence can carry are rigidly constrained: their limiting distribution must be a superposition of microstructures whose barycenters are singular matrices. We prove the converse in the non-symmetric setting: every such superposition is attained by a vanishing sequence. In fact, we are able to construct divergence-free fields $V_n$ $supported$ on $A_n$, whose relative boundary is a smooth compact manifold.

We prove that every smooth immersed disk in Euclidean three-space with nonzero constant mean curvature, regular up to the boundary and mapping its boundary diffeomorphically onto a circle, is an embedded spherical cap.

The transition to virtualized Radio Access Networks (vRAN) enables dynamic power control through fine-grained CPU resource management. However, existing approaches treat gNB as a monolithic entity, failing to exploit the heterogeneous computational characteristics of different protocol layers. This paper proposes ALARM, a layer-aware adaptive resource management framework for constrained 5G vRAN. ALARM decomposes the gNB into functional layers representing distinct RAN tasks with heterogeneous computational demands, and controls CPU resources at the layer level based on processing weights. The framework monitors per-layer performance to detect violations and responds by scaling only the affected layer. This targeted adaptation avoids the over-provisioning inherent in traditional uniform scaling approaches. Experimental validation on two constrained platforms demonstrates up to 33% power reduction versus non-optimized baseline and 19% beyond uniform approaches, with dynamic adaptation achieving 9.8% additional savings while preserving strict real-time guarantees.

Dark matter (DM) direct detection searches look for nuclear recoils as a result of weakly interacting massive particles (WIMPs) scattering off nuclei. Calculating the rate associated with this WIMP-nucleus elastic scattering process involves input from several components, which include the direct detection experimental response functions, the high energy physics content employed, the DM halo velocity distribution, and the target nuclear structure information. Uncertainties in any of these components can impact the interpretation of experimental signals, and affect analysis of rate predictions. In this work, we focus on the uncertainties present due to the nuclear structure modelling of silicon and germanium targets, where a comprehensive set of nuclear form factors are employed through a non-relativistic effective field theory (NREFT) formalism. We show that these uncertainties, obtained from large-scale shell model calculations, impact the nuclear form factors and thus experimental exclusion limits for a SuperCDMS-like experiment. The impact of high energy physics content (through particle physics coefficients) on the magnitude of the nuclear uncertainties is also explored for several WIMP-nucleus scattering responses. Non-negligible nuclear uncertainties are present for several direct detection nuclear responses, indicating that nuclear structure must be accounted for in experimental analysis and interpretation.

Recent years have witnessed a growing interest in AI-aided Kalman filters. While emerging methodologies, such as KalmanNet, were shown to facilitate tracking in partially known state-space models, they are not directly applicable when the underlying dynamics is unknown. To overcome this limitation, we extend the KalmanNet philosophy to the unknown-dynamics regime by developing blind Kalman filtering frameworks that learn both the predictor and the correction gain from data, assuming that the state-evolution function and the noise statistics are both unavailable. To this end, we first introduce a task-aware neural Kalman filtering framework, Blind-KalmanNet, which carries the learning principle of the Kalman gain into the prediction step through a two-head neural architecture that jointly learns a state-dependent linear surrogate and the Kalman gain from data. Building on this formulation, we then develop our main framework, Koopman-aided Blind-KalmanNet, which incorporates Koopman operator theory to lift the unknown dynamics into a latent space where the state evolves linearly. The lifted linear predictor integrates seamlessly into the Blind-KalmanNet structure: the pre-trained deep Koopman network serves as a globally structured predictor, augmented by the task-aware residual surrogate and the learned Kalman gain inherited from Blind-KalmanNet. Extensive experiments demonstrate that the proposed frameworks achieve competitive performance against baselines, with Koopman-aided Blind-KalmanNet attaining the best accuracy across all considered settings.

It has long been a puzzle why the nuclear modification factor in relativistic $p$Pb collisions is consistent with unity, while the high-$p_T$ elliptic flow is significantly positive. The latter is traditionally interpreted as path-length-dependent energy loss, whereas the former is consistent with zero energy loss. In this Letter, we propose that even at high $p_T$ a hard parton coalesces with a boosted thermal medium parton whose energy is correlated with the angle-dependent radial flow. We compare this flow for $pp$, $p$Pb, $p$O, OO and PbPb collision systems and conclude in particular that the $p$Pb flow is large and anisotropic. The collective medium pushes out the hard partons, and more so in $p$Pb, or in-plane, than in $pp$ collisions, or out-of-plane. The mechanism qualitatively agrees with experiment and predicts a significantly smaller elliptic flow signal for jets than for hadrons.

Aerial manipulators represent the forefront of aerial robotics. Although potentially capable of complex interaction tasks, controlling aerial manipulators throughout the dynamic transitions occurring during task execution presents significant challenges. Abrupt or discontinuous changes in system dynamics generated by the transitions suggest the use of a switched approach, yet the available aerial manipulation methods are not designed for coping with switched regimes. In addition, most available methods fall short in coping with the tight couplings between the aerial vehicle and the manipulator, as well as in coping with the state-dependent uncertainties arising from the difficulty in modeling such couplings. We propose a switched-based adaptive control framework for aerial manipulators not relying on a priori knowledge of the vehicle-manipulator couplings and of state-dependent uncertainties. To guarantee stable manipulation despite changes in system dynamics, the framework provides a class of switching signals characterizing those transition phases for which the system is guaranteed to remain stable. Comparative experiments further validate the effectiveness of the proposed switched-based framework over the state of the art.

When facing complex problems, humans tend to try various ideas for different issues. Human thinking patterns exhibit remarkable flexibility in adapting to diverse scenarios. GPT-o1, GPT-o3, and DeepSeek-R1 adopt long chain-of-thought models to address complex problems by increasing reasoning depth, which default to a forward reasoning mode. We conducted statistical analysis on the accuracy of different mathematical problem datasets on models of different scales, and found five reasons for errors: Insufficient solution-space coverage, Computational mistakes, Unverified assumptions, Ignoring constraint conditions, Maximum response length limitation. To address the above issues, we proposed a backward reasoning pattern construction method aimed at enhancing the model's reverse thinking ability and dynamic adaptability. First, we constructed an easy-hard two-stage Math dataset for training large models and gradually improving their inference ability at different difficulty levels. The dataset contains forward reasoning paths as well as backward reasoning paths. And a two-stage supervised fine-tuning process is applied to progressively train the model's backward reasoning capability. Furthermore, a fine-grained reward mechanism is developed, employing smoothed reward signals to strengthen the model's ability to autonomously select thinking modes during the reasoning process, thereby avoiding reward hacking. A linear-decay balanced sampling strategy is designed to maintain a balance between forward and backward reasoning path samples during training, enabling the model to converge quickly and stably. Experimental results show that our method significantly improves reasoning efficiency and accuracy in tasks such as mathematical proofs, offering a flexible and efficient reasoning paradigm for solving complex problems.

Altermagnets, characterized by vanishing net magnetization and momentum-dependent spin splitting, provide a promising platform for next-generation Josephson devices. Here, we exploit the Josephson effect in superconductor-altermagnet-superconductor junctions and show how to engineer prescribed current-phase relations by device design. Based on these programmable Josephson potentials utilizing altermagnetism, we propose a new class of superconducting qubits that combine large anharmonicity with enhanced robustness against decoherence via coherent two-Cooper-pair tunneling. We show that in the $2\phi$-junction regime, this kind of qubit is intrinsically protected against both charge and flux noise due to parity protection. Magnetic flux can be used to precisely control the qubit and, under appropriate bias, this architecture further suppresses charge and flux noise. Our results establish altermagnets as a versatile platform for Josephson-potential engineering and open a new route toward high-performance superconducting qubits combining high coherence, large anharmonicity, and broad tunability.

Clinical outcome assessments (COAs), such as patient-reported outcomes (PROs), are faced with an interpretability issue due to a variety of score metrics being employed. Therefore, evidence linking COA scores to interpretable patient experiences is necessary to judge the extent to which the results of COA-based clinical trial endpoints are meaningful. The fourth Patient-Focused Drug Development guidance, published in draft form by the FDA in 2023, proposes the use of meaningful score regions (MSRs) to divide COA scores into more easily interpretable categories (e.g. severity-based MSRs of 'none', 'mild', 'moderate' and 'severe'). A recommendation is to compare the magnitude of estimated treatment effects for continuous COA scores (e.g. the difference in mean change from baseline between two treatment groups) to the maximum MSR width. In this article we explore this recommendation further, through historical evidence and simulations, and conclude that using the maximum MSR width as a benchmark may fail to detect meaningful treatment effects in practice. We suggest alternative approaches to evaluate treatment effects against MSRs, describing the expected outcome within each arm in terms of the probabilities of MSR membership.

This paper presents the design, optimization, implementation, and on-board validation of a neural processing unit (NPU) accelerator for real-time vehicle detection on the resource-constrained Xilinx Zynq XC7Z020 device of the PYNQ-Z1 board. The work follows a hardware/software co-design methodology that combines quantization-aware training (QAT), lightweight YOLO-derived detectors, Brevitas/QONNX model export, FINN dataflow compilation, Vivado implementation, and physical benchmarking on the target board. Four simultaneous engineering requirements define successful deployment: throughput above 30 frames/s (FPS), energy efficiency above 7 FPS/W, programmable-logic (PL) hardware latency below 50 ms, and Pascal VOC detection accuracy above 0.55 mAP@0.5. The design space includes LP-YOLO and LP-YOLO Slim variants, a custom YOLOv3-tiny reference, 4-bit and mixed low-bit quantization, 320$\times$320 and 256$\times$256 inputs, manual and automatic FIFO sizing, and programmable-logic clocks from 100 to 200 MHz. The final LP-YOLO Slim configuration uses a 256$\times$256 input, w2a4 quantization, and a 142.86 MHz PL clock. With batch 100 it reaches 35.66 FPS at 2.91 W, corresponding to 12.25 FPS/W, while measured PL latency is 45.11 ms and VOC mAP@0.5 is 0.594. This is the only evaluated configuration for which the supplied measurements satisfy all four requirements simultaneously. The results show that low-bit QAT, architectural slimming, FINN folding and FIFO optimization, and moderate clock scaling can jointly provide a practical real-time detector on a small Zynq FPGA.

Multi-camera imaging and camera arrays have become ubiquitous in many applications, such as autonomous driving, robot control, or virtual reality, and accurate disparity maps of objects and their environment are essential for reliable operation. Despite recent advances in neural networks, correctly estimating the disparity of flat and textureless objects remains challenging. In particular, we consider a scenario defined by flat, textureless surfaces illuminated by a single fixed light source, resulting in one dominant specular reflection visible on the object surface. Under these conditions, reliable geometric and photometric cues are missing, and the specular reflection often causes mispredictions, especially when using conventional methods that rely on texture information to match corresponding pixels. To address this issue, the novel Specular Reflection Disparity Estimation SRDE algorithm is introduced, which is specifically designed for the constrained scenario of planar, textureless objects and single-source illumination. Unlike conventional stereo matching methods, SRDE ignores texture and instead leverages the geometric properties of specular reflections by incorporating the position information of the reflective region, the light source, and the camera setup. We show that SRDE outperforms existing methods by a notable margin, achieving more than a 52% improvement in End Point Error on synthetic images. Further tests demonstrate superior performance on real-world data. Furthermore, we integrate SRDE into existing neural disparity estimation pipelines by selectively replacing predictions in specular regions without modifying the backbone model. This hybrid strategy enables additional performance gains without requiring network retraining.

Autonomous AI agents are increasingly deployed in areas where wrong decisions are hard to reverse. This paper examines schema mismatch: the condition in which an agent operates within an interpretive frame that no longer applies to the current context. Outputs produced under such a mismatch can appear internally consistent, linguistically plausible, and largely factually correct; output-quality metrics alone therefore capture the underlying loss of validity only partially. The paper introduces Epi-Logic, a conceptual framework for epistemic runtime control. It couples the detection of schema dissonance, a graduated reduction of autonomy, and the auditable switch to a validated schema. A schema is formalised as a tuple of variable space, expectation model, validity conditions, axioms, and metadata. The Epi-Score aggregates seven graded dimensions of epistemic dissonance; the temporal validity dimension D8, violations of the validity conditions G, and axiom violations are carried as separate categorical paths that are not offset against the aggregate. The architecture rests on a checking asymmetry: formalised validity conditions can be checked at runtime, whereas the correctness of many actions is established only ex post. The paper separates two architectural properties, a conditional result from sequential changepoint detection, and an empirical remainder. Eight falsifiable propositions with named baselines describe the transition to empirical validation. All propositions are empirically testable hypotheses, not established results.

We investigate next generation reservoir computing (NGRC) as a data-driven approach for inferring unseen components of dynamical systems. We compare NGRC with traditional reservoir computing (RC) using the Lorenz and Rössler system, where two unknown components are inferred from one given component. For both systems, NGRC achieves accurate results while requiring fewer training data and less computational time than RC. We identified an inverse proportional behavior between the number of time-delayed steps needed for NGRC and the temporal resolution, indicating that the physical time span covered by the delay interval is an important factor in determining the required number of delayed steps. Finally, we apply NGRC to the observational climate data of ENSO (El Niño–Southern Oscillation) and infer one observable from the remaining variables. Despite the noise and complexity of the real-world data, the NGRC shows promising results. Our findings demonstrate the potential of NGRC for efficient inference of unseen components in both controlled dynamical systems and real-world data.