Target recognition in ghost imaging (GI) systems presents substantial challenges at low sample ratios, when traditional approaches need computationally expensive image reconstruction or have poor accuracy. Deep learning (DL) provides a promising solution by allowing direct feature extraction from GI measurements, however existing methods frequently rely on reconstructed images or are ineffective for natural objects. This study provides an image-free DL architecture for high-accuracy target recognition in GI without intermediate reconstruction, which dramatically improves efficiency and performance. We develop an end-to-end neural network architecture for processing raw GI bucket data (single-pixel measurements) and correlating them to predetermined target classes. The model blends spatial feature encoding with attention techniques to improve discriminative performance in noisy, low-sampling environments. Training uses synthetically augmented GI data to promote generalization, whereas testing is done on experimentally captured natural items. The proposed system outperforms standard GI classification algorithms that rely on reconstructed images, achieving over 90% recognition accuracy at sampling ratios. A comparative investigation reveals a >25% increase in accuracy over conventional procedures at the same sample rate. This study presents a viable DL framework for GI-based target recognition, which eliminates the requirement for image reconstruction while maintaining good accuracy at low sampling rates. The findings highlight the potential for image-free processing in computational imaging, which could enable real-time applications in surveillance, biomedical imaging, and remote sensing. Future work will expand the approach to include dynamic situations and 3D target recovery.

The Ama Fatma oil shale area, located in the Tarfaya-Boujdour basin of southwestern Morocco, was investigated to analyze the content and vertical distribution of rare earth elements (REEs) in shale samples collected from multiple depths. REE concentrations were precisely determined using inductively coupled plasma mass spectrometry (ICP-MS) to better understand their geochemical behavior and origin. The total rare earth element content (ΣREE) ranged from 18 to 50.49 ppm. Average REE concentrations exceeded chondrite values but remained lower than those of the North American Shale Composite (NASC) and the upper continental crust (UCC). Chondrite-normalized patterns showed clear enrichment in light rare earth elements (LREEs) with an average LaN/YbN ratio of 6.4. Negative europium (Eu) anomalies (0.53 < Eu/Eu* < 0.59) and negligible cerium (Ce) anomalies (0.8 < Ce/Ce* < 0.88) were also observed. REE concentrations displayed strong positive correlations with major elements, suggesting a predominantly terrestrial detrital origin. A sequential extraction procedure fractionated REEs into five distinct fractions: exchangeable (F1), carbonate (F2), iron and manganese oxides (F3), organic matter (F4), and residual (F5). The majority of REEs were associated with the iron and manganese oxides fraction (F3), indicating their affinity to these mineral phases. These findings provide insights into the geochemical characteristics and sources of REEs in the Ama Fatma oil shale.

The ATLAS Tile Hadronic Calorimeter, a subsystem of the ATLAS Detector at the Large Hadron Collider, is composed of 256 wedge-shaped modules and measures the energy of hadronic particles. Each module is equipped with a Low Voltage Power Supply (LVPS) consisting of eight transformer-coupled buck converters, which step down 200 Volts DC to the 10 Volts DC required for the front-end electronics. In preparation for the High-Luminosity LHC era, the Phase-II Upgrade must be implemented to accommodate the significantly increased integrated luminosity and radiation levels. Accordingly, analysis of radiation testing is an important focus for evaluating the robust nature of the LVPS electronics.

Due to being located on-detector, the LVPS is subject to the highest radiation exposure among all electronics in the Tile Hadronic Calorimeter. Electronic components of the LVPS, which are manufactured using Complementary Metal–Oxide-Semiconductor (CMOS), bipolar, or bipolar CMOS technologies, are scrutinized when exposed to different radiation sources. Radiation tests performed include Total Ionizing Dose (TID), Single Event Effects, and Non-Ionizing Energy Loss (NIEL). CMOS components are typically the least resilient to TID, while bipolar components are generally prone to displacement damage due to NIEL. Individual and mixed radiation effects on components are explored, and mitigation strategies, including circuit redesign and alternative chip selection, are utilized for a robust final design of the LVPS.

This work successfully demonstrates the application of fundamental electrochemical principles to derive critical physicochemical constants for complex organic molecules. The accurate characterization of mass transfer properties for micropollutants in aqueous systems is a fundamental challenge in applied physical science, bridging condensed matter physics with environmental engineering. This study presents a rigorous determination of the diffusion coefficients for two ubiquitous organic contaminants, Butylparaben and Triclosan, whose environmental persistence necessitates a precise understanding of their transport phenomena. We employ a robust and cost-effective electrolytic conductance method under infinite dilution conditions, systematically investigating the influence of temperature (305.15 K to 319.15 K) and concentration (0.0001 M to 0.0009 M) on both electrolytic and molar conductivity. The resulting data were modeled using both the Kohlrausch equation and a Modified Robinson–Stokes (MRS) quadratic model, with the MRS model demonstrating a superior fit (R² > 0.98) for extrapolating the limiting molar conductivity (Λ°). Using the derived Λ° values, the infinite dilution diffusion coefficients (D°) were calculated via the Nernst–Haskell equation, yielding values of 0.99 x 10⁻⁸ m²/s for Butylparaben and 0.98 x 10⁻⁸ m²/s for Triclosan at 303.15 K. Furthermore, the Nernst–Einstein and Stokes–Einstein equations were utilized to determine the self-diffusion coefficients and corresponding hydrodynamic radii (0.602 x 10⁻¹² m for Butylparaben and 0.740 x 10⁻¹² m for Triclosan), providing deeper insight into single-ion dynamics. The determined transport properties provide essential parameters for developing more accurate computational models of contaminant fate, contributing valuable data to both environmental science and the broader field of soft condensed matter physics.

Microcomputed tomography (mCT) is an advanced technique developed in the medical field and currently used in various areas of society to perform non-destructive testing on various materials, such as continuous fiber-reinforced composites, enabling internal evaluation of the material under loading conditions. Samples of additively manufactured composites reinforced with continuous Kevlar fibers and glass fiber were subjected to compression at two pressure levels (240 and 280 psi) using a pneumatic pressure cell transparent to the microtomograph, designed in the laboratory and optimized for maximum safety within the microtomograph. The results obtained from the mCT tests are digital images with a resolution of 34 μm in .DICOM format, a characteristic of this type of equipment. The digital images are segmented and concatenated to extract the characters necessary to calculate the displacement fields. During asymmetric bending, the (mCT) was a tool that helped observe the behavior of glass and Kevlar fibers and the short carbon fiber matrix (Onyx) under non-uniform stresses, providing important in situ information for the mechanical characterization and design optimization of composites in advanced structural applications. Differences in displacement were observed for three pressure levels (0, 240, and 280 psi), as well as opposite displacements for tension and compression. Finally, it was observed that the short carbon fiber (8 μm in diameter) embedded in the matrix exhibited transparency due to resolution.

Liquid crystal (LC) devices traditionally employ polyimide-coated substrates with unidirectional rubbing to establish a stable nematic director orientation. However, graphene-based surfaces offer an alternative alignment strategy due to their unique molecular interactions with nematic liquid crystals (NLCs). The honeycomb lattice of graphene, with a C–C bond length of 1.42 Å, closely matches the 1.40 Å bond length of benzene rings in typical NLC molecules, enabling epitaxial-like interactions. These interactions are dominated by π–π electron stacking between the aromatic cores of the LC molecules and the graphene lattice, producing uniform planar alignment over large areas. The resulting binding energy, estimated between 0.74 and 0.88 eV per molecule, is associated with partial charge transfer and delocalized π-orbital overlap, further stabilizing the LC orientation.

In this study, we investigated the anchoring properties of nematic 5CB liquid crystal on graphene oxide (GO) thin films using the saturation voltage method (SVM). This approach applies a potential difference to reorient the director from planar to homeotropic, enabling quantitative assessment of anchoring strength. Measurements were performed on sandwich cells with indium–tin oxide electrodes coated with GO and compared with reference cells exhibiting strong (polyimide) and weak (formvar) anchoring. Our results demonstrate that GO substrates significantly influence nematic alignment, highlighting their potential in advanced liquid crystal technologies, including GHz–THz transducers.

The presence of free chlorine ions in industrial wastewater poses significant environmental and operational risks due to their corrosive, oxidative, and toxic nature. Among various dechlorination techniques, activated carbon filtration has proven effective due to its high surface area and strong adsorption capacity. However, optimizing the amount of activated carbon used while ensuring maximum chlorine removal remains a practical challenge. This study presents a MATLAB-based multi-objective optimization approach for modeling and minimizing activated carbon dosage while maximizing dechlorination efficiency. A laboratory-scale experimental setup was developed to simulate the filtration process. A total of 200 experiments were conducted by varying input parameters such as flow rate (10–100 m³/h), initial chlorine concentration (1–10 mg/L), pressure (1.5–5 bar), pH (6.5–8.5), temperature (15–35°C), and activated carbon dose (50–200 kg). These parameters and corresponding residual chlorine concentrations were used to train a neural network in MATLAB using the Levenberg–Marquardt backpropagation algorithm. The network achieved high predictive accuracy with an MSE of 0.00176 and R² = 0.9912. A built-in optimization function in MATLAB was then used to identify the optimal combination of input variables that minimized chlorine levels with the least amount of activated carbon. Results showed that a dose of 84 kg of activated carbon reduced residual chlorine to 0.02 mg/L, maintaining a 98% removal efficiency. This research demonstrates the potential of combining experimental data with intelligent modeling techniques to support sustainable and cost-effective wastewater treatment solutions in industrial applications.

Introduction

Distillation remains the powerhouse of separation processes in the chemical industry, yet its high energy demand presents economic and environmental challenges, especially when dealing with azeotropic mixtures. This study explores process intensification strategies—namely thermal coupling and heat pump techniques—to enhance energy efficiency in pressure swing distillation (PSD) of THF/water and acetone/chloroform azeotropes.

Methods

Four configurations were evaluated: conventional PSD (CPSD), partial heat-integrated PSD (PHIPSD), full heat-integrated PSD (FHIPSD), and heat pump-assisted PSD (HPAPSD). Process design followed a structured four-step approach based on total annual cost (TAC), total energy consumption (TEC), CO₂ emissions, and second law efficiency. PHIPSD and FHIPSD utilized heat recovery between high-pressure and low-pressure columns, reducing steam and cooling utility demands. HPAPSD incorporated vapour recompression.

Results

In the tetrahydrofuran/water system, TAC, TEC, and CO₂ emissions were reduced by up to 50%, 60%, and 83%, respectively, with thermodynamic efficiency reaching 24%. For acetone/chloroform, reductions of up to 71% in TAC, 87% in CO₂ emissions, and efficiencies of 18% were observed. While HPAPSD involved higher capital investment due to compressor systems, it achieved the greatest energy savings and environmental benefits, significantly reducing CO₂ emissions and exergy loss. Thermodynamic efficiency confirmed the HPAPSD's superior performance across both azeotropic systems.

Conclusions

This study illustrates the potential of electrically driven, integrated distillation schemes in minimizing operational costs and environmental emissions. Findings show the need for the broader application of HPAPSD in azeotropic separations, aligning with global sustainability goals and the transition toward renewable energy sources.

Static thumbnails on popular video platforms such as YouTube, Rumble, Crackle, and Netflix often fail to convey the essence of video content, resulting in a mismatch between viewer expectations and the actual material. This paper presents an innovative solution that utilizes Artificial Intelligence (AI) to transform conventional static thumbnails into dynamic highlight clips, providing viewers with an engaging and concise overview of the video. Our proposed model employs convolutional neural networks for effective feature extraction, which is subsequently processed through a Bidirectional Long Short-Term Memory (BiLSTM) network enhanced with an attention mechanism. This combination enables the identification of critical moments within videos, allowing the generation of short video snippets that serve as intelligent thumbnails. To demonstrate the effectiveness of our approach, we conduct experiments using benchmark datasets, including TVSum and SumMe, which feature a diverse range of video content. The model's performance is evaluated using key engagement metrics, such as accuracy, F-Score, and Click-Through Rate (CTR), to assess the impact of dynamic previews compared to traditional static thumbnails. Our results reveal significant enhancements in user interaction, with an F-Score of 0.579 for TVSum and a notable increase in CTR for the dynamic thumbnails. These findings underscore the advantages of AI-generated video highlights in enhancing content discoverability and viewer engagement on video-sharing platforms.

Lower limb amputees, comprising the majority, rely heavily on prosthetic devices for mobility. However, conventional prosthetic sockets often create uneven pressure distributions on the residual limb, leading to localized pressure points that cause skin irritation, discomfort, and long-term tissue damage, ultimately contributing to prosthesis abandonment. This study investigates the application of auxetic meta-structures and polymer nanocomposites to improve prosthetic socket design by enhancing interface pressure distribution, reducing deflection, and improving energy dissipation. Three prosthetic socket designs featuring internal meta-structures, chiral (auxetic), reentrant hexagon (auxetic), and regular honeycomb (non-auxetic), were modeled in SolidWorks and analyzed using finite element analysis (FEA) in Ansys under single-leg stance loading conditions. Each socket was simulated using five materials: polypropylene, ultra-high molecular weight polyethylene (UHMWPE), polypropylene with 5% zinc oxide (ZnO), polypropylene with 5% titanium dioxide (TiO₂), and UHMWPE with 0.5% graphene nanoplatelets. The results demonstrated that auxetic chiral prosthetic sockets consistently exhibited a 44.6% reduction in average interfacial pressure compared to non-auxetic hexagon sockets. The chiral design also showed a more uniform pressure distribution and enhanced energy absorption. While nanoparticle inclusion generally leads to less favorable contact pressure profiles, it offers a promising strategy for tailoring socket stiffness. These findings highlight the benefits of integrating auxetic geometries and nanocomposite materials in socket design to enhance user comfort and minimize prosthesis rejection. Additive manufacturing enables the scalable production of such customized prosthetic sockets, offering a promising pathway to improve access and quality of life for individuals with limb loss.