Deep neural network (DNN) inference relies increasingly on specialized hardware for high computational efficiency. This work introduces a field-programmable gate array (FPGA)-based dynamically configurable accelerator featuring systolic arrays, high-bandwidth memory, and UltraRAMs. We present two processing unit (PU) configurations with different computing capabilities using the same interfaces and peripheral blocks. By instantiating multiple PUs and employing a heuristic weight transfer schedule, the architecture achieves notable throughput efficiency over prior works. Moreover, we outline how the architecture can be extended to emulate analog in-memory computing (AIMC) devices to aid next-generation heterogeneous AIMC chip designs and investigate device-level noise behavior. Overall, this brief presents a versatile DNN inference acceleration architecture adaptable to various models and future FPGA designs.

We confirm the direct connection between entanglement entropy and the notion of irreversibility in the renormalization-group flow in the context of a simple theory for which a calculation from first principles is feasible. The change of the entanglement entropy for a spherical entangling surface as its radius grows from zero to infinity corresponds to the flow from the UV to the IR. Through analytical and numerical means, we compute the entanglement entropy for a free massive scalar theory, making use of the method of correlation functions. We deduce a $c$-function in $1+1$ dimensions and an $a$-function in $3+1$ dimensions. Both functions are monotonic and vary continuously between one and zero, as expected for this simple theory.

In quantum mechanics separable states can be characterized as convex combinations of product states whereas non-separable states exhibit entanglement. Quantum entanglement has played a pivotal role in both theoretical investigations and practical applications within quantum information science. In this study, we explore the connection between product states and geometric structures, specifically manifolds and their associated geometric properties such as the first fundamental form (metric). We focus on the manifolds formed by the product states of M systems of N levels, examining the induced metric derived from the Euclidean metric. For elementary cases we will compute the Levi-Civita connection, and, where computationally tractable, the scalar curvature.

We propose a novel point cloud U-Net diffusion architecture for 3D generative modeling capable of generating high-quality and diverse 3D shapes while maintaining fast generation times. Our network employs a dual-branch architecture, combining the high-resolution representations of points with the computational efficiency of sparse voxels. Our fastest variant outperforms all non-diffusion generative approaches on unconditional shape generation, the most popular benchmark for evaluating point cloud generative models, while our largest model achieves state-of-the-art results among diffusion methods, with a runtime approximately 70% of the previously state-of-the-art PVD. Beyond unconditional generation, we perform extensive evaluations, including conditional generation on all categories of ShapeNet, demonstrating the scalability of our model to larger datasets, and implicit generation which allows our network to produce high quality point clouds on fewer timesteps, further decreasing the generation time. Finally, we evaluate the architecture's performance in point cloud completion and super-resolution. Our model excels in all tasks, establishing it as a state-of-the-art diffusion U-Net for point cloud generative modeling. The code is publicly available at this https URL.

The CoRoT (Convection, Rotation, and planetary Transits) mission still holds a large trove of high-quality, underused light curves with excellent signal-to-noise and continuous coverage. This paper, the first in a series, identifies and classifies variable stars in CoRoT fields whose variability has not been analyzed in the main repositories. We combine simulations and real data to test a moving-average scheme that mitigates instrumental jumps and enhances the recovery of short-period signals (<1 day) in roughly 20-day time series. For classification, we adopt a supervised selection built on features extracted from folded light curves using the double period, and we construct template-based models that also act as a new classifier for well-sampled light curves. We report 9,272 variables, of which 6,249 are not listed in SIMBAD or VSX. Our preliminary classes include 309 Beta Cephei, 3,105 Delta Scuti, 599 Algol-type eclipsing binaries, 844 Beta Lyrae eclipsing binaries, 497 W Ursae Majoris eclipsing binaries, 1,443 Gamma Doradus, 63 RR Lyrae, and 32 T Tauri stars. The resulting catalog inserts CoRoT variables into widely used astronomical repositories. Comparing sources in the inner and outer Milky Way, we find significant differences in the occurrence of several classes, consistent with metallicity and age gradients. The ability to recover sub-day periods also points to automated strategies for detecting longer-period variability, which we will develop in subsequent papers of this series.

We present a search for solar axions produced through the axion-electron coupling $(g_{ae})$ using data from a novel 7-GridPix detector installed at the CERN Axion Solar Telescope (CAST). The detector, featuring ultra-thin silicon nitride windows and multiple veto systems, collected approximately 160 hours of solar tracking data between 2017-2018. Using machine learning techniques and the veto systems, we achieved a background rate of $1.06\times 10^{-5}\,\text{keV}^{-1}\text{cm}^{-2}\text{s}^{-1}$ at a signal efficiency of about $80\,\%$ in the $0.2$-$8\,\text{keV}$ range. Analysis of the data yielded no significant excess above background, allowing us to set a new upper limit on the product of the axion-electron and axion-photon couplings of g_{ae}\cdot g_{a\gamma} &lt; 7.35\times 10^{-23}\,\text{GeV}^{-1} at $95\,\%$ confidence level. This result improves upon the previous best helioscope limit and demonstrates the potential of GridPix technology for rare event searches. Additionally, we derived a limit on the axion-photon coupling of g_{a\gamma} &lt; 9.0\times 10^{-11}\,\text{GeV}^{-1} at $95\,\%$ CL, which, while not surpassing CAST's best limit, provides complementary constraints on axion models.

The advent of digital streaming platforms have recently revolutionized the landscape of music industry, with the ensuing digitalization providing structured data collections that open new research avenues for investigating popularity dynamics and mainstream success. The present work explored which determinants hold the strongest predictive influence for a track's inclusion in the Billboard Hot 100 charts, including streaming popularity, measurable audio signal attributes, and probabilistic indicators of human listening. The analysis revealed that popularity was by far the most decisive predictor of Billboard Hot 100 inclusion, with considerable contribution from instrumentalness, valence, duration and speechiness. Logistic Regression achieved 90.0% accuracy, with very high recall for charting singles (0.986) but lower recall for non-charting ones (0.813), yielding balanced F1-scores around 0.90. Random Forest slightly improved performance to 90.4% accuracy, maintaining near-perfect precision for non-charting singles (0.990) and high recall for charting ones (0.992), with F1-scores up to 0.91. Gradient Boosting (XGBoost) reached 90.3% accuracy, delivering a more balanced trade-off by improving recall for non-charting singles (0.837) while sustaining high recall for charting ones (0.969), resulting in F1-scores comparable to the other models.

Code optimization is a challenging task requiring a substantial level of expertise from developers. Nonetheless, this level of human capacity is not sufficient considering the rapid evolution of new hardware architectures and software environments. In light of this, recent research proposes adopting machine learning and artificial intelligence techniques to automate the code optimization process. In this paper, we introduce PerfRL, an innovative framework designed to tackle the problem of code optimization. Our framework leverages the capabilities of small language models (SLMs) and reinforcement learning (RL), facilitating a system where SLMs can assimilate feedback from their environment during the fine-tuning phase, notably through unit tests. When benchmarked against existing models, PerfRL demonstrates superior efficiency in terms of speed and computational resource usage, attributed to its reduced need for training steps and its compatibility with SLMs. Furthermore, it substantially diminishes the risk of logical and syntactical errors. To evaluate our framework, we conduct experiments on the PIE dataset using a lightweight large language model (i.e., CodeT5) and a new reinforcement learning algorithm, namely RRHF. For evaluation purposes, we use a list of evaluation metrics related to optimization quality and speedup. The evaluation results show that our approach achieves similar or better results compared to state-of-the-art models using shorter training times and smaller pre-trained models.

The paper presents a cradle-to-gate sustainability assessment methodology specifically designed to evaluate aircraft components in a robust and systematic manner. This methodology integrates multi-criteria decision-making (MCDM) analysis across ten criteria, categorized under environmental impact, cost, and performance. Environmental impact is analyzed through life cycle assessment and cost through life cycle costing, with both analyses facilitated by SimaPro software. Performance is measured in terms of component mass and specific stiffness. The robustness of this methodology is tested through various MCDM techniques, normalization approaches, and objective weighting methods. To demonstrate the methodology, the paper assesses the sustainability of a fuselage panel, comparing nine variants that differ in materials, joining techniques, and part thicknesses. All approaches consistently identify thermoplastic CFRP panels as the most sustainable option, with the geometric mean aggregation of weights providing balanced criteria consideration across environmental, cost, and performance aspects. The adaptability of this proposed methodology is illustrated, showing its applicability to any aircraft component with the requisite data. This structured approach offers critical insights to support sustainable decision-making in aircraft component design and procurement.

Flexible electronics offer unique advantages for conformable, lightweight, and disposable healthcare wearables. However, their limited gate count, large feature sizes, and high static power consumption make on-body machine learning classification highly challenging. While existing bendable RISC-V systems provide compact solutions, they lack the energy efficiency required. We present a mechanically flexible RISC-V that integrates a bespoke multiply-accumulate co-processor with fixed coefficients to maximize energy efficiency and minimize latency. Our approach formulates a constrained programming problem to jointly determine co-processor constants and optimally map Multi-Layer Perceptron (MLP) inference operations, enabling compact, model-specific hardware by leveraging the low fabrication and non-recurring engineering costs of flexible technologies. Post-layout results demonstrate near-real-time performance across several healthcare datasets, with our circuits operating within the power budget of existing flexible batteries and occupying only 2.42 mm^2, offering a promising path toward accessible, sustainable, and conformable healthcare wearables. Our microprocessors achieve an average 2.35x speedup and 2.15x lower energy consumption compared to the state of the art.

The advent of Generative AI (GenAI) in education presents a transformative approach to traditional teaching methodologies, which often overlook the diverse needs of individual students. This study introduces a GenAI tool, based on advanced natural language processing, designed as a digital assistant for educators, enabling the creation of customized lesson plans. The tool utilizes an innovative feature termed 'interactive mega-prompt,' a comprehensive query system that allows educators to input detailed classroom specifics such as student demographics, learning objectives, and preferred teaching styles. This input is then processed by the GenAI to generate tailored lesson plans. To evaluate the tool's effectiveness, a comprehensive methodology incorporating both quantitative (i.e., % of time savings) and qualitative (i.e., user satisfaction) criteria was implemented, spanning various subjects and educational levels, with continuous feedback collected from educators through a structured evaluation form. Preliminary results show that educators find the GenAI-generated lesson plans effective, significantly reducing lesson planning time and enhancing the learning experience by accommodating diverse student needs. This AI-driven approach signifies a paradigm shift in education, suggesting its potential applicability in broader educational contexts, including special education needs (SEN), where individualized attention and specific learning aids are paramount

It is well known that alternative theories to the Standard Model allow -- and sometimes require -- fundamental constants, such as the fine-structure constant, $\alpha$, to vary in spacetime. We demonstrate that one way to investigate these variations is through the Mass-Radius relation of compact astrophysical objects, which is inherently affected by $\alpha$ variations. We start by considering the model of a polytropic white dwarf, which we perturb by adding the $\alpha$ variations for a generic class of Grand Unified Theories. We then extend our analysis to neutron stars, building upon the polytropic approach to consider more realistic equations of state, discussing the impact of such variations on mass-radius measurements in neutron stars. We present some constraints on these models based on current data and also outline how future observations might distinguish between extensions of the Standard Model.

Dark sectors, consisting of new, light, weakly-coupled particles that do not interact with the known strong, weak, or electromagnetic forces, are a particularly compelling possibility for new physics. Nature may contain numerous dark sectors, each with their own beautiful structure, distinct particles, and forces. This review summarizes the physics motivation for dark sectors and the exciting opportunities for experimental exploration. It is the summary of the Intensity Frontier subgroup "New, Light, Weakly-coupled Particles" of the Community Summer Study 2013 (Snowmass). We discuss axions, which solve the strong CP problem and are an excellent dark matter candidate, and their generalization to axion-like particles. We also review dark photons and other dark-sector particles, including sub-GeV dark matter, which are theoretically natural, provide for dark matter candidates or new dark matter interactions, and could resolve outstanding puzzles in particle and astro-particle physics. In many cases, the exploration of dark sectors can proceed with existing facilities and comparatively modest experiments. A rich, diverse, and low-cost experimental program has been identified that has the potential for one or more game-changing discoveries. These physics opportunities should be vigorously pursued in the US and elsewhere.

The need to repeatedly shuttle around synaptic weight values from memory to processing units has been a key source of energy inefficiency associated with hardware implementation of artificial neural networks. Analog in-memory computing (AIMC) with spatially instantiated synaptic weights holds high promise to overcome this challenge, by performing matrix-vector multiplications (MVMs) directly within the network weights stored on a chip to execute an inference workload. However, to achieve end-to-end improvements in latency and energy consumption, AIMC must be combined with on-chip digital operations and communication to move towards configurations in which a full inference workload is realized entirely on-chip. Moreover, it is highly desirable to achieve high MVM and inference accuracy without application-wise re-tuning of the chip. Here, we present a multi-core AIMC chip designed and fabricated in 14-nm complementary metal-oxide-semiconductor (CMOS) technology with backend-integrated phase-change memory (PCM). The fully-integrated chip features 64 256x256 AIMC cores interconnected via an on-chip communication network. It also implements the digital activation functions and processing involved in ResNet convolutional neural networks and long short-term memory (LSTM) networks. We demonstrate near software-equivalent inference accuracy with ResNet and LSTM networks while implementing all the computations associated with the weight layers and the activation functions on-chip. The chip can achieve a maximal throughput of 63.1 TOPS at an energy efficiency of 9.76 TOPS/W for 8-bit input/output matrix-vector multiplications.

A search for Dark Sectors is performed using the unique M2 beam line at the CERN Super Proton Synchrotron. New particles ($X$) could be produced in the bremsstrahlung-like reaction of high energy 160 GeV muons impinging on an active target, $\mu N\rightarrow\mu NX$, followed by their decays, $X\rightarrow\text{invisible}$. The experimental signature would be a scattered single muon from the target, with about less than half of its initial energy and no activity in the sub-detectors located downstream the interaction point. The full sample of the 2022 run is analyzed through the missing energy/momentum channel, with a total statistics of $(1.98\pm0.02)\times10^{10}$ muons on target. We demonstrate that various muon-philic scenarios involving different types of mediators, such as scalar or vector particles, can be probed simultaneously with such a technique. For the vector-case, besides a $L_\mu-L_\tau$ $Z'$ vector boson, we also consider an invisibly decaying dark photon ($A'\rightarrow\text{invisible}$). This search is complementary to NA64 running with electrons and positrons, thus, opening the possibility to expand the exploration of the thermal light dark matter parameter space by combining the results obtained with the three beams.

Cryo-electron tomography (cryo-ET) is an imaging technique that allows three-dimensional visualization of macro-molecular assemblies under near-native conditions. Cryo-ET comes with a number of challenges, mainly low signal-to-noise and inability to obtain images from all angles. Computational methods are key to analyze cryo-electron tomograms. To promote innovation in computational methods, we generate a novel simulated dataset to benchmark different methods of localization and classification of biological macromolecules in tomograms. Our publicly available dataset contains ten tomographic reconstructions of simulated cell-like volumes. Each volume contains twelve different types of complexes, varying in size, function and structure. In this paper, we have evaluated seven different methods of finding and classifying proteins. Seven research groups present results obtained with learning-based methods and trained on the simulated dataset, as well as a baseline template matching (TM), a traditional method widely used in cryo-ET research. We show that learning-based approaches can achieve notably better localization and classification performance than TM. We also experimentally confirm that there is a negative relationship between particle size and performance for all methods.

The entanglement entropy of a massless scalar field in de Sitter space depends on multiple scales, such as the radius of the entangling surface, the Hubble constant and the UV cutoff. We perform a high-precision numerical calculation using a lattice model in order to determine the dependence on these scales in the Bunch-Davies vacuum. We derive the leading de Sitter corrections to the flat-space entanglement entropy for subhorizon entangling radii. We analyze the structure of the finite-size effects and we show that the contribution to the entanglement entropy of the sector of the theory with vanishing angular momentum depends logarithmically on the size of the overall system, which extends beyond the horizon.

This paper presents an innovative pedagogical framework employing tangible interactive games to enhance artificial intelligence (AI) knowledge and literacy among elementary education students. Recognizing the growing importance of AI competencies in the 21st century, this study addresses the critical need for age-appropriate, experiential learning tools that demystify core AI concepts for young learners. The proposed approach integrates physical role-playing activities that embody fundamental AI principles, including neural networks, decision-making, machine learning, and pattern recognition. Through carefully designed game mechanics, students actively engage in collaborative problem solving, fostering deeper conceptual understanding and critical thinking skills. The framework further supports educators by providing detailed guidance on implementation and pedagogical objectives, thus facilitating effective AI education in early childhood settings. Empirical insights and theoretical grounding demonstrate the potential of tangible interactive games to bridge the gap between abstract AI theories and practical comprehension, ultimately promoting AI literacy at foundational educational levels. The study contributes to the growing discourse on AI education by offering scalable and adaptable strategies that align with contemporary curricular demands and prepare young learners for a technologically driven future.