Search

Abstract Recently, Transformer-based methods have gained prominence in image super-resolution (SR) tasks, addressing the challenge of long-range dependence through the incorporation of cross-layer connectivity and local attention mechanisms. However, the analysis of these networks using local attribution maps has revealed significant limitations in leveraging the spatial extent of input information. To unlock the inherent potential of Transformer in image SR, we propose the Multi-Attention Fusion Transformer (MAFT), a novel model designed to integrate multiple attention mechanisms with the objective of expanding the number and range of pixels activated during image reconstruction. This integration enhances the effective utilization of input information space. At the core of our model lies the Multi-attention Adaptive Integration Groups, which facilitate the transition from dense local attention to sparse global attention through the introduction of Local Attention Aggregation and Global Attention Aggregation blocks with alternating connections, effectively broadening the network's receptive field. The effectiveness of our proposed algorithm has been validated through comprehensive quantitative and qualitative evaluation experiments conducted on benchmark datasets. Compared to state-of-the-art methods (e.g. HAT), the proposed MAFT achieves 0.09 dB gains on Urban100 dataset for × 4 SR task while containing 32.55% and 38.01% fewer parameters and FLOPs, respectively.

Abstract Efficiently capturing radioactive methyl iodide (CH3I), present at low concentrations in the high-temperature off-gas of nuclear facilities, poses a significant challenge. Here we present two strategies for CH3I adsorption at elevated temperatures using a unified azolate-based metal-organic framework, MFU-4l. The primary strategy leverages counter anions in MFU-4l as nucleophiles, engaging in metathesis reactions with CH3I. The results uncover a direct positive correlation between CH3I breakthrough uptakes and the nucleophilicity of the counter anions. Notably, the optimal variant featuring SCN- as the counter anion achieves a CH3I capacity of 0.41 g g−1 at 150 °C under 0.01 bar, surpassing all previously reported adsorbents evaluated under identical conditions. Moreover, this capacity can be easily restored through ion exchange. The secondary strategy incorporates coordinatively unsaturated Cu(I) sites into MFU-4l, enabling non-dissociative chemisorption for CH3I at 150 °C. This modified adsorbent outperforms traditional materials and can be regenerated with polar organic solvents. Beyond achieving a high CH3I adsorption capacity, our study offers profound insights into CH3I capture strategies viable for practically relevant high-temperature scenarios.

Abstract The reconstruction of rutile TiO2 (110) holds significant importance as it profoundly influences the surface chemistry and catalytic properties of this widely used material in various applications, from photocatalysis to solar energy conversion. Here, we directly observe the asymmetric surface reconstruction of rutile TiO2 (110)-(1×2) with atomic-resolution using in situ spherical aberration-corrected scanning transmission electron microscopy. Density functional theory calculations were employed to complement the experimental observations. Our findings highlight the pivotal role played by repulsive electrostatic interaction among the small polarons −formed by excess electrons following the removal of neutral oxygen atoms− and the subsequent surface relaxations induced by these polarons. The emergence and disappearance of these asymmetric structures can be controlled by adjusting the oxygen partial pressure. This research provides a deeper understanding, prediction, and manipulation of the surface reconstructions of rutile TiO2 (110), holding implications for a diverse range of applications and technological advancements involving rutile-based materials.

Abstract Multi-scale representation provides an effective answer to the scale variation of objects and entities in semantic segmentation. The ability to capture long-range pixel dependency facilitates semantic segmentation. In addition, semantic segmentation necessitates the effective use of pixel-to-pixel similarity in the channel direction to enhance pixel areas. By reviewing the characteristics of earlier successful segmentation models, we discover a number of crucial elements that enhance segmentation model performance, including a robust encoder structure, multi-scale interactions, attention mechanisms, and a robust decoder structure. The attention mechanism of the asymmetric non-local neural network (ANNet) is merged with multi-scale pyramidal modules to accelerate model segmentation while maintaining high accuracy. However, ANNet does not account for the similarity between pixels in the feature map channel direction, making the segmentation accuracy unsatisfactory. As a result, we propose EMSNet, a straightforward convolutional network architecture for semantic segmentation that consists of Integration of enhanced regional module (IERM) and Multi-scale convolution module (MSCM). The IERM module generates weights using four or five-stage feature maps, then fuses the input features with the weights and uses more computation. The similarity of the channel direction feature graphs is also calculated using ANNet’s auxiliary loss function. The MSCM module can more accurately describe the interactions between various channels, capture the interdependencies between feature pixels, and capture the multi-scale context. Experiments prove that we perform well in tests using the benchmark dataset. On Cityscapes test data, we get 82.2% segmentation accuracy. The mIoU in the ADE20k and Pascal VOC datasets are, respectively, 45.58% and 85.46%.

Abstract The ability to capture pixels' long-distance interdependence is beneficial to semantic segmentation. In addition, semantic segmentation requires the effective use of pixel-to-pixel similarity in the channel direction to enhance pixel regions. Asymmetric Non-local Neural Networks (ANNet) combine multi-scale spatial pyramidal pooling modules and Non-local blocks to reduce model parameters without sacrificing performance. However, ANNet does not consider pixel similarity in the channel direction in the feature map, so its segmentation effect is not ideal. This article proposes a Mutually Reinforcing Non-local Neural Networks (MRNNet) to improve ANNet. MRNNet consists specifically of the channel enhancement regions module (CERM), and the position-enhanced pixels module (PEPM). In contrast to Asymmetric Fusion Non-local Block (AFNB) in ANNet, CERM does not combine the feature maps of the high and low stages, but rather utilizes the auxiliary loss function of ANNet. Calculating the similarity between feature maps in channel direction improves the category representation of feature maps in the channel aspect and reduces matrix multiplication computation. PEPM enhances pixels in the spatial direction of the feature map by calculating the similarity between pixels in the spatial direction of the feature map. Experiments reveal that our segmentation accuracy for cityscapes test data reaches 81.9%. Compared to ANNet, the model's parameters are reduced by 11.35 (M). Given ten different pictures with a size of 2048 × 1024, the average reasoning time of MRNNet is 0.103(s) faster than that of the ANNet model.

The reconstruction of the polar CeO2 (100) surface has been a subject of long-standing debates due to its complexity and the limited availability of experimental data. Herein, we successfully reveal a CeO2 (100)-(4 × 6) surface reconstruction by combining in situ spherical aberration-corrected scanning transmission electron microscopy, density functional theory calculations, and a particle swarm optimization-based algorithm for structure searching. We have further elucidated the stabilizing mechanism of the reconstructed structure, which involves the splitting of the filled Ce(4f) states and the mixing of the lower-lying ones with the O(2p) orbitals, as evidenced by the projected density of states. We also reveal that the surface chemisorption properties toward water molecules, an important step in numerous heterogeneous catalytic reactions, are enhanced. These insights into the distinct properties of ceria surface pave the way for performance improvements of ceria in a wide range of applications. [ABSTRACT FROM AUTHOR]

Structural and compositional inhomogeneity is common in zeolites and considerably affects their properties. Thickness-limited lateral resolution, lack of depth resolution, and electron dose-constrained focusing limit local structural studies of zeolites in conventional transmission electron microscopy (TEM). We demonstrate that a multislice ptychography method based on four-dimensional scanning TEM (4D-STEM) data can overcome these limitations. Images obtained from a ~40-nanometer-thick MFI zeolite exhibited a lateral resolution of ~0.85 angstrom that enabled the identification of individual framework oxygen (O) atoms and the precise determination of the orientations of adsorbed molecules. Furthermore, a depth resolution of ~6.6 nanometers allowed probing of the three-dimensional distribution of O vacancies, as well as the phase boundaries in intergrown MFI and MEL zeolites. The 4D-STEM ptychography can be generally applied to other materials with similar high electron-beam sensitivity. [ABSTRACT FROM AUTHOR]

Innovation and sustainability are two important determinants to promote the competitiveness of construction projects. In order to improve the production efficiency, environmental regulation is an indispensable element. Nevertheless, few studies have discussed the complex relationships between innovation, sustainability, environmental regulation and competitiveness at the project level. Motivated by this, the paper aims to explore the interaction between innovation and sustainability, and the impacts on competitiveness, while incorporating the mediating role of sustainability and moderating role of environmental regulation. Based on structural equation modeling method, a novel conceptual model was established. By processing and analyzing 184 valid questionnaires, the findings indicate that innovation has a positive effect on sustainability and competitiveness. Sustainability is also significantly associated with competitiveness. Moreover, the relationship between innovation and competitiveness is partially mediated by sustainability. When construction projects carry out innovation activities or implement sustainability practices, the greater the degree of environmental regulation, the higher the competitiveness. These findings provide theoretical contributions to the existing studies, revealing the intertwined relationships between innovation sustainability environmental regulation and competitiveness at the project level. In practice, project managers can improve project sustainability and competitiveness by adopting sustainability-rooted practice and innovation-based activities.