Letteratura scientifica selezionata sul tema "DPN synthetic acceleration"

As deep neural network (DNN) models get larger and more complicated, the importance of hardware acceleration becomes more and more apparent. This paper discusses various hardware acceleration strategies for deep learning, especially in the area of computer vision. It explores the use of GPUs, FPGAs, and ASICs, detailing their respective strengths and weaknesses in accelerating DNNs. This paper argues that the future of DNN hardware acceleration lies in hybrid approaches that combine the advantages of different architectures. Software advances such as improved compilers and synthesis tools will also play a critical role in making these techniques more accessible. By utilizing the appropriate hardware technology for a given task and continuing to innovate in both hardware and software, computer vision will make significant advances in performance, efficiency, and scalability. This hybrid approach is key to the future of DNN hardware acceleration, offering a path to overcome the limitations of any single type of hardware.

The gravity method is a geophysical exploration method to measure variations in the acceleration of gravity on the surface of the earth in response to variations in rocks that exist beneath the surface. In gravity exploration requires a preliminary picture as a reference for measurement. This study aims to make forward modeling synthetic OCTAVE based using synthetic data on subsurface rock structures, so as to produce intrusion and fracture models based on differences in the value of the acceleration of gravity from one point to another on the surface of the earth. Synthetic modeling with the geological parameter approach of the study area is based on variations in the price of rock density. The model parameters used in intrusion modeling are the density value of 2.7 g/cm3 and the depth of 850 meters while the fracture modeling uses a density value of 2.7 g/cm3 with a depth of 350 meters and 360 meters and a thickness of 500 meters. From intrusion modeling, the gravity vertical component of attraction force is 0.03 mGal and in the fracture modeling the gravity vertical component of attraction force is 0.0565 mGal. Based on the results of this modeling, distance curve vs. gravity anomaly response is obtained for both cases. In the intrusion rock model obtained by the profile model with an open type down. While the fracture modeling is obtained anomalous profile curve variation which states that in the fracture area with a significant change in the direction of the curve.

This research aimed to know the influence of addition crude product Pseudomonas aeruginosa IA7d biosurfactant from substrate hexadecane and diesel fuel in supporting activity of diesel fuel biodegradation and both potention to replace synthetic surfactant for hydrocarbon remediation. Biodegradation test was done by experimental in laboratory scale by Tween-80 as synthetic surfactant comparator. Growth culture was from seawater of Tanjung Perak Harbor Surabaya include various microbe types in its. 2 percent (w/v) of diesel fuel and three surfactants test (0.70 percent w/v crude product P. aeruginosa IA7d biosurfactant from hexadecane substrate; 60,69 percent w/v crude product P. aeruginosa IA7d biosurfactant from diesel fuel substrate; and 0.10 percent w/v synthetic surfactant Tween 80) at uniform surface tension of surfactant solution (54.03 mN/m) were added into 30 ml seawater sample. Cultures were incubated at room temperature (30 °C) during 14 days with shaking (90 rpm). Result of biodegradation test showed the third addition of surfactants test had an effect on acceleration of diesel fuel biodegradation activity that proved at the increased of microbial count (bacterium, yeast, and mould) and decreased of diesel fuel concentration was shown by chromatograms area of diesel fuel. Highest degradation was successively reached by culture got crude product P. aeruginosa IA7d biosurfactant from diesel fuel substrate (88.034 percent), crude product P. aeruginosa IA7d biosurfactant from hexadecane substrate (80.908 percent), synthetic surfactant Tween-80 (54.732 percent), and control that was without addition surfactant test (0.132 percent). Base on number of microbial count, percentage of decreasing chromatogram peaks of diesel fuel from each surfactant test, and price substrate produce (diesel fuel and hexadecane) among both crude product P. aeruginosa IA7d biosurfactant, known that the crude product P. aeruginosa IA7d biosurfactant from diesel fuel substrate had highest ability acceleration to biodegradation activity on diesel fuel and had a potency to replace synthetic surfactant for remediation hydrocarbon application in environment.

Dalam perencanaan struktur bangunan tahan gempa, tujuan umum dari analisa struktur dinamis adalah memprediksi respons dari struktur terhadap pengaruh ground motion yang memiliki Spectral Acceleration (Sa) pada periode tertentu berdasarkan tingkat kemungkinan terlampui (Probability of Excedence/PE) 10% atau 2% masa layan bangunan 50 tahun. Prediksi respons struktur ditentukan dengan memilih ground motion yang cocok dengan beberapa target spektra dan nantinya ground motion tersebut digunakan sebagai input dalam analisis dinamis struktur. Pada paper ini menyajikan analisis perambatan gelombang gempa dari batuan dasar ke lapisan permukaan (Site Specific Response Analysis/ SSRA). Data-data yang diperlukan adalah data stratifikasi tanah dan parameter kecepatan gelombang geser yang didapatkan dari korelasi empiris terhadap data hasil pemboran dan uji N-SPT. Ground motion synthetic batuan dasar yang digunakan untuk perambatan gelombang gempa diperoleh dari hasil penelitian sebelumnya yang mengadopsi pendekatan statistik Conditional Mean Spectrum (CMS) agar ground motion yang dihasilkan dipermukaan cocok dengan prediksi masalah respon struktur yang sebenarnya. SSRA dilakukan berdasarkan teori perambatan gelombang geser satu dimensi dalam time domain dengan menggunakan program Non-linear Earthquake Response Analysis (NERA). Hasilnya diperoleh percepatan maksimum gempa dipermukaan (peak surface acceleration/ PBA) berbeda-beda di kedua lokasi tergantung pada faktor amplifikasi, karakteristik dan jenis gempa yang terjadi. Direkomendasikan respons spectra desain untuk periode ulang 475 tahun dan 2475 tahun wilayah 4 kelas tanah sedang (SD) untuk kedua lokasi tersebut. Data-data tersebut digunakan sebagai input dalam penentuan beban gempa pada bangunan dalam analisis struktur dinamis.

Bone health is a vital element in the quality of life aspect, especially with the increasing risk of osteoporosis and fractures with age. Extremely Low Frequency (ELF) electromagnetic field therapy offers the potential for bone regeneration through collagen synthesis, osteoblast stimulation, and molecular signal modulation. This study evaluates the potential and efficacy of ELF for bone health based on scientific literature. The results of the analysis show that ELF is effective in accelerating healing, increasing bone density, and is safe at certain intensities. However, optimization of parameters such as duration and intensity of exposure is still needed. This study supports the development of standard protocols for safe and effective ELF applications in bone therapy.

Abstract In the present study we evaluated the effect of complement activation by immune complexes (IC) on the expression of decay-accelerating factor (DAF) on human mesangial cells (MC). MC in culture were incubated with an Ag (DNP-Gelatin) that binds to fibronectin present in the MC matrix. Subsequently, MC were incubated with anti-DNP antibodies in the presence of human serum. By immunoperoxidase staining we showed that these incubations resulted in IC formation and deposition of human C3 and terminal complement components (C5b-9) on the mesangial matrix and on the surface of MC. By immunoperoxidase staining and by RIA we showed that IC formation and complement activation significantly increased DAF expression on the MC plasma membrane. The induction of DAF expression was a consequence of deposition of terminal complement components on the MC because, zymosan-activated serum and IC formation in the presence of C5- or C8-deficient serum failed to increase MC DAF expression. Furthermore, the observed increased DAF expression was the consequence of increased DAF synthesis by MC. Thus, both cycloheximide and actinomycin D blocked the increase on MC DAF observed after incubation with IC and serum. MC DAF had biophysical and functional characteristics similar to DAF in other cells. Thus, 1) MC DAF was resistant to trypsin but was removed from the MC membrane by pronase; 2) phosphatidylinositol-specific phospholipase C removed 48 +/- 4% of MC DAF indicating that MC DAF is anchored in the cell membrane by phosphatidylinositol groups; 3) DAF isolated from MC-inhibited complement-mediated hemolysis and demonstrated a molecular mass of 83 kDa. In conclusion, deposition of terminal complement components on human MC trigger new synthesis and membrane expression of DAF. Because DAF protects cells against complement-mediated lysis, we postulate that DAF may protect glomerular cells during IC and complement-mediated glomerulonephritis.

Haircare products are very important to prevent one of the problems of hair damage, both from synthetic and natural ingredients. One of the natural ingredients of hair care products is Red Palm Oil (RPO) and Extra Virgin Olive Oil (EVOO). This research is an experimental study that aims to see whether Red Palm Oil and Extra Virgin Olive Oil have effectiveness in growing hair. This research method used 4 groups of test animals, namely the red palm oil group, extra virgin olive oil, negative control (aqua dest), and positive control (minoxidil). The hair growth activity test was carried out by applying each formula to the marked rabbit's back and measuring the length of the hair using a caliper on the 5th, 10th, 15th, 20th, 25th, and 30th days. On the 31st day, the hair weight was tested by weighing the hair using an analytical balance. The results showed that each formula had hair growth activity, but the most effective oil preparation in accelerating hair growth was Red Palm Oil (RPO).

The tourism sector is one of the efforts in accelerating the increase in regional original income in Banyuwangi Regency in a sustainable manner during the pandemic and endemic period. This study aims to analyze the priority of rater agreement synthesis with Kendall's coefficient of concordance on objects and tourist attractions in optimizing local revenue in Banyuwangi Regency. This study uses a quantitative approach with the support of a qualitative approach through interviews and questionnaires to a number of experts. The sampling technique in this study was carried out by purposive sampling technique from respondents from the health office, the culture and tourism office, the Banyuwangi Regency Regional Development Planning Agency, the Healthy Banyuwangi Forum, and the Tourism Awareness Group. The analysis technique uses the Analytic Network Process (ANP). The results of the analysis show that the results of the analysis of the value of Kendall's coefficient of concordance W (rater agreement) is 0.95. The diversity of values given by each respondent contributed to the rater agreement score of 95%. The increase in the number of tourists and the length of stay of tourists occupy the highest priority for rater agreement synthesis at 36%. This confirms that optimistic tourism objects and attractions can be a trigger in boosting local revenue in the tourism sector in Banyuwangi Regency

In recent years, deep neural networks (DNNs) have addressed new applications with intelligent autonomy, often achieving higher accuracy than human experts. This capability comes at the expense of the ever-increasing complexity of emerging DNNs, causing enormous challenges while deploying on resource-limited edge devices. Improving the efficiency of DNN hardware accelerators by compression has been explored previously. Existing state-of-the-art studies applied approximate computing to enhance energy efficiency even at the expense of a little accuracy loss. In contrast, bit-serial processing has been used for improving the computational efficiency of neural processing without accuracy loss, exploiting a simple design, dynamic precision adjustment, and computation pruning. This research presents Serial/Parallel Systolic Array (SPSA) and Octet Serial/Parallel Systolic Array (OSPSA) processing elements for edge DNN acceleration, which exploit bit-serial processing on systolic array architecture for improving computational efficiency. For evaluation, all designs were described at the RTL level and synthesized in 28 nm technology. Post-synthesis cycle-accurate simulations of image classification over DNNs illustrated that, on average, a sample 16 × 16 systolic array indicated remarkable improvements of 17.6% and 50.6% in energy efficiency compared to the baseline, with no loss of accuracy.

Permainan sepakbola dicirikan dengan pola aktivitas yang berubah-ubah, olahraga dengan durasi yang panjang dengan intensitas yang tinggi dan aksi yang eksplosif seperti melakukan akselarasi, deakselarasi, merubah arah, melompat, dan menjegal yang tentunya memerlukan kerja yang tinggi dari daya tubuh. Di dalam pertandingan kebutuhan energi sangat tinggi, dipenuhi dengan metabolisme aerob dan anaerob, yang mana perlu dibarengi dengan asupan nutrisi berkualitas dalam bentuk mikronutrisi khususnya asupan/pemaparan vitamin D untuk pemain agar dapat menunjang dan menjaga performa permainan dari seorang pemain, terlebih pada level sepakbola elit. Vitamin D sendiri adalah suatu prekursor hormon yang memiliki peranan penting di dalam menjaga kesehatan tulang dan fungsi imunitas. Penemuan reseptor vitamin D pada sel otot rangka manusia telah membawa para ilmuwan fisiologi otot untuk menemukan peranan penting vitamin D dalam meregulasikan sintesis protein dan fungsi otot. Pada atlit sepakbola elit, terdapat kebutuhan konstan vitamin D di dalam memaksimalkan sintesis protein untuk memenuhi kebutuhan fisik dari latihan harian. Selain dari pemaparan sinar matahari vitamin D pun dapat didapatkan dari sumber makanan yang mengandung nutrisi ini (hewani dan nabati).Soccer was characterized with activity that changed periodically during the game, this sport has long duration and with high intensity dan explosive action like doing acceleration, deacceleration, changing direction, jumping, and tackling. Of course in order to doing that this sport needs high rate of physical performance of the players. In the match, energy requirements is very high, fulfill by aerob metabolism and anaerob, to help the metabolism regulation, micronutrien intake especially vitamin D intake/exposure for the players to support and maintain performance of soccer players, especially in the elite level. Vitamin D is a hormonal precursor that hold an important role to maintain bone health and imunity function. The discovery of vitamin D receptor in human skeletal’s cells made attention to Muscle Physiologyst in order to find a way how vitamin D regulated protein synsthesis and muscle function. For soccer elite athletes, there’s constant needs of vitamin D to built synthesis protein in order to support the physical requirements from daily training. Besides sunlight exposure, vitamin D can be acquired from dietary foods that contain this vitamin (animal and plant source).Soccer was characterized with activity that changed periodically during the game, this sport has long duration and with high intensity dan explosive action like doing acceleration, deacceleration, changing direction, jumping, and tackling. Of course in order to doing that this sport needs high rate of physical performance of the players. In the match, energy requirements is very high, fulfill by aerob metabolism and anaerob, to help the metabolism regulation, micronutrien intake especially vitamin D intake/exposure for the players to support and maintain performance of soccer players, especially in the elite level. Vitamin D is a hormonal precursor that hold an important role to maintain bone health and imunity function. The discovery of vitamin D receptor in human skeletal’s cells made attention to Muscle Physiologyst in order to find a way how vitamin D regulated protein synsthesis and muscle function. For soccer elite athletes, there’s constant needs of vitamin D to built synthesis protein in order to support the physical requirements from daily training. Besides sunlight exposure, vitamin D can be acquired from dietary foods that contain this vitamin (animal and plant source).

Le travail de la thèse se compose de deux parties. La première partie se focalise sur le développement d’un schéma (transport + accélération) aux caractéristiques linéaire surfacique pour la résolution de l’équation du transport sur des géométries 3D extrudées. À présent, seule l’approximation constante peut être utilisée. La différence entre les deux méthodes réside dans le fait que pour l’une la vitesse de convergence au maillage est linéaire (approximation constante) alors qu’elle est quadratique pour l’autre méthode (approximation surfacique linéaire). Pratiquement, la dernière méthode permet de réaliser des calculs avec une même précision qu’avec l’approximation constante tout en réduisant le nombre de mailles, et donc le temps de calcul. Le schéma doit aussi permettre de développer le flux sur une base polynomiale dans la direction axiale, et les sections efficaces ; ce qui importe si l’on veut faire des calculs en évolution. Notez que contrairement à la plupart des approximations linéaires, la source, dans ce schéma, est définie sur les surfaces verticales des régions de calcul. La valeur de la source sur les surfaces horizontales est obtenue à partir d’une interpolation linaire au vol entre les valeurs définies sur les surfaces verticales. L’avantage de cette approche est de pouvoir se débarrasser de l’intégration par le traçage des grandeurs définies sur les surfaces horizontales. Un bilan sur les moments volumiques du flux angulaire est utilisé pour tester la convergence des itérations internes. A cette fin, un opérateur géométrique est défini de manière à construire une source volumique à partir de la source surfacique. La conservation par région est forcée par correction. Concernant l’accélération du transport, le choix a été fait d’implémenter une accélération synthétique de type DPn. La méthode repose sur le développement sur la base des harmoniques sphériques du flux angulaire définie sur les surfaces d’une région de calcul mais peut être vu comme un préconditionnement d’un schéma itératif de type Richardson. Plusieurs arguments motivent ce choix. Le rayon spectral de l’opérateur associé à l’accélération est inférieur à d’autres accélérations comme la diffusion synthetic acceleration (DSA) ou des accélérations non-linéaires de type Coarse mesh finite difference (CMFD), y compris pour des milieux à fort chemins optiques. Par ailleurs, la construction du système d’équation repose sur la même discrétisation spatiale que le transport et limite la nécessité d’une normalisation de certaines grandeurs d’intérêts qui pourraient apparaître avec d’autres méthodes. Enfin, l’utilisation de la forme intégrale du transport rend la méthode attractive pour son utilisation sur des configurations géométriques complexes et des maillages non-structurées. La seconde partie porte sur la correction d’instabilités numériques qui apparaissent lorsque l’on augmente l’ordre de développement spatial du flux. En milieu homogène infini, le terme de fuite issue de l’équation intégro-différentielle doit être nul. Ce n’est pas le cas pour des régions de calcul où la quantité de cordes avec un chemin optique suffisamment faible dépasse une certaine limite. Ce phénomène pénalise la convergence des méthodes du TDT-MOC et la rend impossible s’il est trop important. Ces travaux sont vérifiés sur différentes configurations géométriques
The thesis work consists of two parts. The first part focuses on developing a linear surface characteristics scheme (transport + acceleration) for solving the transport equation on extruded 3D geometries. Currently, only the constant approximation can be used. The difference between the two methods lies in the fact that the convergence speed for the mesh is linear (constant approximation) for one, whereas it is quadratic for the other method (linear surface approximation). Practically, the latter method allows calculations with the same accuracy as the constant approximation while reducing the number of meshes, and thus the computation time. The scheme should also allow the flux to be developed on a polynomial basis in the axial direction, as well as cross sections, which is important for depletion calculations. Note that unlike most linear approximations, the source in this scheme is defined on the vertical surfaces of the calculation regions. The value of the source on the horizontal surfaces is obtained from an on-the-fly linear interpolation between the values defined on the vertical surfaces. The advantage of this approach is to eliminate the need for tracking-based integration of quantities defined on the horizontal surfaces. A balance on the volume moments of the angular flux is used to test the convergence of the inner iterations. To this end, a geometric operator is defined to construct a volume source from the surface source. Conservation per region is enforced by correction. Regarding transport acceleration, the choice was made to implement a DPn-type synthetic acceleration. The method is based on the spherical harmonics expansion of the angular flux defined on the surfaces of a calculation region but can be seen as a preconditioning of a Richardson-type iterative scheme. Several arguments motivate this choice. The spectral radius of the operator associated with the acceleration is lower than other accelerations such as diffusion synthetic acceleration (DSA) or nonlinear accelerations like Coarse Mesh Finite Difference (CMFD), including for media with high optical paths. Furthermore, the construction of the equation system relies on the same spatial discretization as the transport and limits the need for normalization of certain quantities of interest that might appear with other methods. Finally, the use of the integral form of transport makes the method attractive for use in complex geometric configurations and unstructured meshes. The second part deals with correcting numerical instabilities that appear when increasing the spatial development order of the flux. In an infinite homogeneous medium, the leakage term from the integro-differential equation should be zero. This is not the case for calculation regions where the number of chords with sufficiently low optical paths exceeds a certain limit. This phenomenon penalizes the convergence of TDT-MOC methods and makes it impossible if it is too significant. These works are verified and tested on different geometric configurations

A new Hydroprocessed Depolymerized Cellulosic Diesel (HDCD) fuel has been developed using a process which takes biomass feedstock (principally cellulosic wood) to produce a synthetic fuel that has nominally 1/2 cyclo-paraffins and 1/2 aromatic hydrocarbons in content. This HDCD fuel with a low cetane value (Derived Cetane Number from the Ignition Quality Tester, DCN = 27) was blended with naval distillate fuel (NATO symbol F-76) in various quantities and tested in order to determine how much HDCD could be blended before diesel engine operation became problematic. Blends of 20% HDCD (DCN = 45), 30%, 40% (DCN = 41) and 60% HDCD (DCN = 37) by volume were tested with conventional naval distillate fuel (DCN = 49). Engine start performance was evaluated with a conventional mechanically Direct Injected (DI) Yanmar engine and a Waukesha mechanical indirect injected (IDI) CFR diesel engine, and showed that engine start times increased steadily with increasing HDCD content. Longer start times with increasing HDCD content were the result of some engine cycles with poor combustion leading to a slower rate of engine acceleration towards rated speed. A repeating sequence of alternating cycles which combust followed by a non-combustion cycle were common during engine run-up. Additionally, steady state engine testing was also performed using both engines. HDCD has a significantly higher bulk modulus than F76 due to its very high aromatic content, and the engines showed earlier Start of Injection (SOI) timing with increasing HDCD content for equivalent operating conditions. Additionally, due to the lower DCN, the higher HDCD blends showed moderately longer Ignition Delay (IGD) with moderately shorter overall burn durations. Thus, the mid-combustion metric (CA50: 50% burn duration Crank Angle position) was only modestly affected with increasing HDCD content. Increasing HDCD content beyond 40% led to significantly longer start times.

ABSTRACT Three single-cluster hydraulic fracture stages were pumped at the Frontier Observatory for Research in Geothermal Energy (FORGE) site in Milford, Utah. Our goal was to develop a robust model to accurately represent the formation of fracture networks in this naturally fractured geothermal reservoir. To begin this process, we used geological and geophysical data and data from one-dimensional Fullbore Formation MicroImager (FMI) to build a discrete fracture network model for the natural fractures. The natural fracture network (DFN) was built stochastically with areal density, length, and orientation distribution of natural fractures. We then took one-dimensional synthetic cores to ensure that the number and density of fractures per unit length of the core matched with the actual measurements (for each fracture set) until the best statistical description of natural fractures was found. The length distribution of natural fractures was simulated using a power law distribution. INTRODUCTION Utah Frontier Observatory for Research in Geothermal Energy (FORGE) is a dedicated laboratory for developing, testing, and accelerating breakthroughs in EGS technologies to advance the use of geothermal resources (Department of Energy, 2022). Natural fractures have been indicated by outcrop data and Fullbore Formation MicroImager (FMI) log data. Most of the data can be found directly or indirectly in the Geothermal Data Repository (GDR), which includes data from Utah FORGE, as well as all data collected from other researchers funded by the Geothermal Technologies Office (GTO) (Department of Energy, 2022). Fig. 1 shows the locations of the vertical pilot well (58-32), the highly deviated injection well (16A(78)-32), and another deep vertical well (56-32). In this paper, based on the data provided by these wells, a discrete fracture network (DFN) was developed to characterize the natural fracture network. Fractures are explicitly expressed in the form of planes of weakness. A DFN realization (Fig. 2) is built with a specified distribution of fracture orientation, length, and density (Cao et al., 2023). A model that can simulate fracture propagation in naturally fractured reservoirs can be found in Cao and Sharma (2023, 2022a, 2022b).