Search Results (9)

Hypergolic ionic liquid fuels are promising alternatives to the toxic hydrazine-based propellants. The present study investigates the combustion performance of a fuel composed of 1-ethyl-3-methylimidazolium thiocyanate ([EMIM][SCN]) and copper(I) thiocyanate (CuSCN) with 95 wt% hydrogen peroxide (H₂O₂) in a 50 N thruster. Two injectors, DM1 (low pressure drop) and DM2 (high pressure drop), were tested at chamber pressures of 10 and 15 bar. The DM1 injector at 15 bar chamber pressure showed high combustion efficiency but suffered from strong pressure oscillations (>10% instability). Switching to the DM2 injector reduced instability (<5%) by increasing the pressure drop. Combustion stability was also improved as the fuel injector orifice diameter/fuel jet velocity (D/V) decreased. FFT analysis showed an instability frequency of 253 Hz with DM1/15 bar, which was higher than the DM1/10 bar test results. In conclusion, the test results revealed the injector or chamber design and pressure drop are the key factors in improving combustion stability for hypergolic propulsion systems. Full article

Background: Flatfoot deformity is a common condition in children and teenagers that may increase the risk of knee, hip, and back pain. Most of the insoles suggested to treat flatfoot symptoms are not designed to adapt to foot temperature during walking, and they are either too soft to provide support or hard enough to be uncomfortable. Purpose: This study aims to develop an advanced solution to diagnose and treat flexible flatfoot (FFT) using infrared thermography measurements and a hybrid insole reinforced by nitinol (NiTiCu) smart-memory-alloy wires (SMAWs), this super-elastic alloy can return back to its pre-deformed shape when heated, which helps to reduce the local high-temperature points caused by the uneven pressure of FFT. This approach achieves a more uniform thermal distribution across the foot, which makes the hybrid insole more comfortable. Methods: The study involved 16 subjects, divided into two groups of eight flat-footed and eight normal. The procedure includes two parts, namely, designing a prototype insole with SMAW properties based on thermography measurement by using SolidWorks, and evaluating this design using Ansys. Second, a hybrid insole reinforced with SMAWs is customized for flatfoot subjects. The thermography measurement differences between the medial and lateral sides of the metatarsophalangeal line are compared for the normal and flatfoot groups before and after wearing the suggested design. Results: The results show that our approach safely diagnosed FFT and significantly improved the thermal distribution in FFT subjects by more than 80% after wearing the suggested design. A paired t-test reported significant (p-value > 0.001) thermal decreases in the high-temperature points after using the SMAW insole, which was closely approximated to the normal subjects. Conclusions: the SMAW-reinforced insole is comfortable and suitable for treating FFT deformity, and infrared thermography is an effective tool to evaluate FFT deformity. Full article

Convolutional Neural Networks (CNNs) are among the most prevalent deep learning techniques employed across various domains. The computational complexity of CNNs is largely attributed to the convolution operations. These operations are computationally demanding and significantly impact overall model performance. Traditional CNN implementations convert convolutions into matrix operations via the im2col (image to column) technique, facilitating parallelization through advanced BLAS libraries. This study identifies and investigates a significant yet intricate pattern of data redundancy within the matrix-based representation of convolutions, a pattern that, while complex, presents opportunities for optimization. Through meticulous analysis of the redundancy inherent in the im2col approach, this paper introduces a mathematically succinct matrix representation for convolution, leading to the development of an optimized FFT-based convolution with finer FFT granularity. Benchmarking demonstrates that our approach achieves an average speedup of 14 times and a maximum speedup of 17 times compared to the regular FFT convolution. Similarly, it outperforms the Im2col+GEMM approach from NVIDIA’s cuDNN library, achieving an average speedup of three times and a maximum speedup of five times. Our FineGrained FFT convolution approach, when integrated into Caffe, a widely used deep learning framework, leads to significant performance gains. Evaluations using synthetic CNNs designed for real-world applications show an average speedup of 1.67 times. Furthermore, a modified VGG network variant achieves a speedup of 1.25 times. Full article

Improving the performance of the convolution operation has become a key target for High Performance Computing (HPC) developers due to its prevalence in deep learning applied mainly to video processing. The improvement is being pushed by algorithmic and implementation innovations. Algorithmically, the convolution can be solved as it is mathematically enunciated, but other methods allow to transform it into a Fast Fourier Transform (FFT) or a GEneral Matrix Multiplication (GEMM). In this latter group, the Winograd algorithm is a state-of-the-art variant that is specially suitable for smaller convolutions. In this paper, we present openCNN, an optimized CUDA C++ implementation of the Winograd convolution algorithm. Our approach achieves speedups of up to $1.76\times$ on Turing RTX 2080Ti and up to $1.85\times$ on Ampere RTX 3090 with respect to Winograd convolution in cuDNN 8.2.0. OpenCNN is released as open-source software. Full article

Cryogenic Electron Microscopy (Cryo-EM) has been established as one of the key players in Structural Biology. It can reconstruct a 3D model of the sample at the near-atomic resolution, which led to a Method of the year award by Nature, and the Nobel Prize in 2017. With the growing number of facilities, faster microscopes, and new imaging techniques, new algorithms are needed to process the so-called movies data produced by the microscopes in real-time, while preserving a high resolution and maximum of additional information. In this article, we present a new algorithm used for movie alignment, called FlexAlign. FlexAlign is able to correctly compensate for the shift produced during the movie acquisition on-the-fly, using the current generation of hardware. The algorithm performs a global and elastic local registration of the movie frames using Cross-Correlation and B-spline interpolation for high precision. We show that our execution time is compatible with real-time correction and that we preserve the high-resolution information up to high frequency. Full article

An existing hybrid MPI-OpenMP scheme is augmented with a CUDA-based fine grain parallelization approach for multidimensional distributed Fourier transforms, in a well-characterized pseudospectral fluid turbulence code. Basics of the hybrid scheme are reviewed, and heuristics provided to show a potential benefit of the CUDA implementation. The method draws heavily on the CUDA runtime library to handle memory management and on the cuFFT library for computing local FFTs. The manner in which the interfaces to these libraries are constructed, and ISO bindings utilized to facilitate platform portability, are discussed. CUDA streams are implemented to overlap data transfer with cuFFT computation. Testing with a baseline solver demonstrated significant aggregate speed-up over the hybrid MPI-OpenMP solver by offloading to GPUs on an NVLink-based test system. While the batch streamed approach provided little benefit with NVLink, we saw a performance gain of $30\%$ when tuned for the optimal number of streams on a PCIe-based system. It was found that strong GPU scaling is nearly ideal, in all cases. Profiling of the CUDA kernels shows that the transform computation achieves 15% of the attainable peak FlOp-rate based on a roofline model for the system. In addition to speed-up measurements for the fiducial solver, we also considered several other solvers with different numbers of transform operations and found that aggregate speed-ups are nearly constant for all solvers. Full article

Pt(Cu)/C and Pt-Ru(Cu)/C electrocatalysts with core-shell structure supported on Vulcan Carbon XC72R have been synthesized by potentiostatic deposition of Cu nanoparticles on the support, galvanic exchange with Pt and spontaneous deposition of Ru species. The duration of the electrodeposition time of the different species has been modified and the obtained electrocatalysts have been characterized using electrochemical and structural techniques. The High Resolution Transmission Electron Microscopy (HRTEM), Fast Fourier Transform (FFT) and Energy Dispersive X-ray (EDX) microanalyses allowed the determining of the effects of the electrodeposition time on the nanoparticle size and composition. The best conditions identified from Cyclic Voltammetry (CV) corresponded to onset potentials for CO and methanol oxidation on Pt-Ru(Cu)/C of 0.41 and 0.32 V vs. the Reversible Hydrogen Electrode (RHE), respectively, which were smaller by about 0.05 V than those determined for Ru-decorated commercial Pt/C. The CO oxidation peak potentials were about 0.1 V smaller when compared to commercial Pt/C and Pt-Ru/C. The positive effect of Cu was related to its electronic effect on the Pt shells and also to the generation of new active sites for CO oxidation. The synthesis conditions to obtain the best performance for CO and methanol oxidation on the core-shell Pt-Ru(Cu)/C electrocatalysts were identified. When compared to previous results in literature for methanol, ethanol and formic acid oxidation on Pt(Cu)/C catalysts, the present results suggest an additional positive effect of the deposited Ru species due to the introduction of the bifunctional mechanism for CO oxidation. Full article

The synthesis of core-shell Pt(Cu) and Pt-Ru(Cu) electrocatalysts allows for a reduction in the amount of precious metal and, as was previously shown, a better CO oxidation performance can be achieved when compared to the nanoparticulated Pt and Pt-Ru ones. In this paper, the carbon black used as the support was previously submitted to electrochemical oxidation and characterized by XPS. The new catalysts thus prepared were characterized by HRTEM, FFT, EDX, and electrochemical techniques. Cu nanoparticles were generated by electrodeposition and were further transformed into Pt(Cu) and Pt-Ru(Cu) core-shell nanoparticles by successive galvanic exchange with Pt and spontaneous deposition of Ru species, the smallest ones being 3.3 nm in mean size. The onset potential for CO oxidation was as good as that obtained for the untreated carbon, with CO stripping peak potentials about 0.1 and 0.2 V more negative than those corresponding to Pt/C and Ru-decorated Pt/C, respectively. Carbon oxidation yielded an additional improvement in the catalyst performance, because the ECSA values for hydrogen adsorption/desorption were much higher than those obtained for the non-oxidized carbon. This suggested a higher accessibility of the Pt sites in spite of having the same nanoparticle structure and mean size. Full article

The increasing availability of satellite imagery acquired by existing and new sensors allows a wide variety of new applications that depend on the use of diverse spectral and spatial resolution data sets. One of the pre-conditions for the use of hybrid image data sets is a consistent geo-correction capacity. We demonstrate how a novel fast template matching approach implemented on a graphics processing unit (GPU) allows us to accurately and rapidly geo-correct imagery in an automated way. The key difference with existing geo-correction approaches, which do not use a GPU, is the possibility to match large source image segments (8,192 by 8,192 pixels) with relatively large templates (512 by 512 pixels) significantly faster. Our approach is sufficiently robust to allow for the use of various reference data sources. The need for accelerated processing is relevant in our application context, which relates to mapping activities in the European Copernicus emergency management service. Our new method is demonstrated over an area northwest of Valencia (Spain) for a large forest fire event in July 2012. We use the Disaster Monitoring Constellation’s (DMC) DEIMOS-1 and RapidEye imagery for the delineation of burnt scar extent. Automated geo-correction of each full resolution image set takes approximately one minute. The reference templates are taken from the TerraColor data set and the Spanish national ortho-imagery database, through the use of dedicated web map services. Geo-correction results are compared to the vector sets derived in the Copernicus emergency service activation request. Full article