Parallel Computing Research Articles

Analytical Solution for the Problem of Point Location in Arbitrary Planar Domains

This paper presents a general analytical solution for the problem of locating points in planar regions with an arbitrary geometry at the boundary. The proposed methodology overcomes the traditional solutions used for polygonal regions. The method originated from the explicit evaluation of the contour integral using the Residue and Cauchy theorems, which then evolved toward a technique very similar to the winding number and, finally, simplified into a variant of ray-crossing approach slightly more informed and more universal than the classic approach, which had been used for decades. The very close relation of both techniques also emerges during the derivation of the solution. The resulting algorithm becomes simpler and potentially faster than the current state of the art for point locations in arbitrary polygons because it uses fewer operations. For polygonal regions, it is also applicable without further processing for special cases of degeneracy, and it is possible to use in fully integer arithmetic; it can also be vectorized for parallel computation. The major novelty, however, is the extension of the technique to virtually any shape or segment delimiting a planar domain, be it linear, a circular arc, or a higher order curve.

Accelerated iterative DG finite element solvers for large-scale time-harmonic acoustic problems

Finite element methods are widely used to solve time-harmonic wave propagation problems, but solving large cases can be extremely difficult even with the computational power of parallel computers. In this work, the linear system resulting from the finite element discretization is solved with iterative solution methods, which are efficient in parallel but can require a large number of iterations. In standard discontinuous Galerkin (DG) methods, the numerical solution is discontinuous at the interfaces between the elements. In hybridizable DG methods, additional unknowns are introduced at the interfaces between the finite elements, and the physical unknowns are eliminated from the global system, resulting in a hybridized system. We have recently proposed a new strategy, called CHDG, where the additional unknowns correspond to transmission variables, whereas in the standard approach they are numerical fluxes. This strategy improves the properties of the hybridized system for faster iterative solution procedures. In this talk, we present and study a 3D CHDG implementation with nodal finite element basis functions. The resulting scheme has properties amenable to efficient parallel computing. Numerical results are presented to validate the method, and preliminary 3D computational results are proposed.

Direct aeroacoutic simulation of a ducted axial fan with acoustic liners using a volume penalization lattice Boltzmann solver on heterogeneous massively parallel platforms

Aeroacoustic noise of an axial fan is directly solved to understand its mechanism, using our in-house Lattice Boltzmann Method (LBM) solver. To make the LBM solver industrially successful, we must reduce calculation time while maintaining its accuracy. We adopt Volume Penalization method for moving boundaries to achieve independent parallelism, and to utilize GPU (Graphical Processing Units) for parallel computation. In this study, we present a comparison between numerical and experimental results of an axial fan in a lined duct. The acoustic liner is designed to reduce 2nd BPF noise, and the noise reduction of 9.8dB is observed in the experiment. With our LBM code, the noise reduction was reproduced within an accuracy of 2.5dB, and profound insights are provided through the detailed investigation of the numerical results.

Integrating angular and domain decomposition with space-angle discontinuous Galerkin methods in 2D radiative transfer

A space-angle discontinuous Galerkin (saDG) method is used to solve the steady-state radiative transfer equation (RTE) for 2D problems involving absorption, emission, and scattering for a semitransparent medium. This approach discretizes both spatial and angular domains. Parallel computing is based on angular decomposition (AD), and domain decomposition (DD) techniques. The DD technique directly solves the entire domain using the MUMPS library, whereas the AD technique results in an iterative approach for scattering media. This study proposes a novel hybrid AD-DD method, combining the best aspects of both techniques. Numerical results investigate the scalability, performance, and efficiency of AD and DD techniques. It is shown that a hybrid AD-DD technique is superior to these individual techniques by taking advantage of their strengths. Numerical methods demonstrate the applicability of the method of the best combination of hybrid AD-DD to 2D scattering gray media with complex geometries or enclosures with circular and square obstacles.

Development of high-efficient asphalt pavement modeling software for digital twin of road infrastructure

To develop digital twin (DT) of road infrastructure, one critical element is computation of pavement responses (strains, stresses, and deflections) under traffic and environmental loading. This study aims to develop high-efficient asphalt pavement modeling software based on semi-analytical finite element method (SAFEM) for DT application. The algorithms address important aspects in vehicle-tire-pavement interaction modeling, such as dynamic vehicular loading, three-dimensional (3-D) non-uniform tire contact stress, viscoelastic behavior of asphalt material, and interface bonding condition. The simulation accuracy is verified by comparison with full-scale test and field measurements, and the relative differences are around 5 % to 20 %. Techniques including optimized discrete Fourier transform, parallel computing, graphics processing unit (GPU) acceleration, and sparse matrices are implemented for computation efficiency. As compared to the traditional 3-D FEM, SAFEM shows significant savings in computation time and storage usage. The high efficiency and accuracy make the software full of potential to be applied for DT of roadway infrastructure.

A Novel Method for Aircraft Structural Dynamic Strain Trend Signal Processing via Optimized Parallel Computing

In this study, we investigate the underlying causes of drift in the time history curves of measured parameters obtained through strain electrical measurements and assess their impacts on load measurements. To address the challenge of efficiently processing large volumes of aircraft load data, we propose and analyze a multi-level parallel algorithm specifically designed for the data processing of aircraft load measurements. To achieve this objective, we discuss parallel processing at both medium- and fine-grained levels and develop two distinct parallel processing algorithms: one for coarse- and medium-grained aircraft-type data streams, and another for medium- and fine-grained takeoff and landing data streams. The efficacy of these algorithms is validated through the processing of load data measured on a specific aircraft wing. The results demonstrate that the proposed approach offers a novel technical pathway for large-scale scientific computations and enhances data processing efficiency in the domain of aircraft load spectrum analysis.

The future of multimodal artificial intelligence models for integrating imaging and clinical metadata: a narrative review.

With the ongoing revolution of artificial intelligence (AI) in medicine, the impact of AI in radiology is more pronounced than ever. An increasing number of technical and clinical AI-focused studies are published each day. As these tools inevitably affect patient care and physician practices, it is crucial that radiologists become more familiar with the leading strategies and underlying principles of AI. Multimodal AI models can combine both imaging and clinical metadata and are quickly becoming a popular approach that is being integrated into the medical ecosystem. This narrative review covers major concepts of multimodal AI through the lens of recent literature. We discuss emerging frameworks, including graph neural networks, which allow for explicit learning from non-Euclidean relationships, and transformers, which allow for parallel computation that scales, highlighting existing literature and advocating for a focus on emerging architectures. We also identify key pitfalls in current studies, including issues with taxonomy, data scarcity, and bias. By informing radiologists and biomedical AI experts about existing practices and challenges, we hope to guide the next wave of imaging-based multimodal AI research.

Memristor-based circuit design of BiLSTM network

The bidirectional long short-term memory (BiLSTM) network involves significant amount of parameter computations. This paper proposes the memristor-based bidirectional long short-term memory (MBiLSTM) network, with its capability of in-memory computing and parallel computing, can accelerates the parameter computations speed. The MBiLSTM network circuit is composed of normalization circuit, two memristor-based long short-term memory (LSTM) circuits, memristor-based resnet circuit, memristor-based dense circuitand winner-take-all (WTA) circuit. The voltage signals are scaled to the setting range by normalization circuit, memristor-based LSTM circuit is responsible for extracting features from the dataset, memristor-based resnet circuit can enhance the overall performance of the network, memristor-based dense circuit ensures that the final outputs dimension matches the dimension of the target signals, WTA circuit outputs the maximum voltage of memristor-based dense circuit. The effectiveness of the MBiLSTM network is validated through gait recognition experiment and handwritten digit recognition experiment. The stability, robustness and potential errors in the manufacturing process of memristance are analyzed.

Emulating quantum computing with optical matrix multiplication

Optical computing harnesses the speed of light to perform vector-matrix operations efficiently. It leverages interference, a cornerstone of quantum computing algorithms, to enable parallel computations. In this work, we interweave quantum computing with classical structured light by formulating the process of photonic matrix multiplication using quantum mechanical principles such as state superposition and subsequently demonstrate a well-known algorithm, namely, Deutsch–Jozsa’s algorithm. This is accomplished by elucidating the inherent tensor product structure within the Cartesian transverse degrees of freedom of light, which is the main resource for optical vector-matrix multiplication. To this end, we establish a discrete basis using localized Gaussian modes arranged in a lattice formation and demonstrate the operation of a Hadamard gate. Leveraging the reprogrammable and digital capabilities of spatial light modulators, coupled with Fourier transforms by lenses, our approach proves adaptable to various algorithms. Therefore, our work advances the use of structured light for quantum information processing.

Fast interactive simulations of cardiac electrical activity in anatomically accurate heart structures by compressing sparse uniform cartesian grids

Background and Objective:Numerical simulations are valuable tools for studying cardiac arrhythmias. Not only do they complement experimental studies, but there is also an increasing expectation for their use in clinical applications to guide patient-specific procedures. However, numerical studies that solve the reaction–diffusion equations describing cardiac electrical activity remain challenging to set up, are time-consuming, and in many cases, are prohibitively computationally expensive for long studies. The computational cost of cardiac simulations of complex models on anatomically accurate structures necessitates parallel computing. Graphics processing units (GPUs), which have thousands of cores, have been introduced as a viable technology for carrying out fast cardiac simulations, sometimes including real-time interactivity. Our main objective is to increase the performance and accuracy of such GPU implementations while conserving computational resources. Methods:In this work, we present a compression algorithm that can be used to conserve GPU memory and improve efficiency by managing the sparsity that is inherent in using Cartesian grids to represent cardiac structures directly obtained from high-resolution MRI and mCT scans. Furthermore, we present a discretization scheme that includes the cross-diagonal terms in the computational cell to increase numerical accuracy, which is especially important for simulating thin tissue sections without the need for costly mesh refinement. Results:Interactive WebGL simulations of atrial/ventricular structures (on PCs, laptops, tablets, and phones) demonstrate the algorithm’s ability to reduce memory demand by an order of magnitude and achieve calculations up to 20x faster. We further showcase its superiority in slender tissues and validate results against experiments performed in live explanted human hearts. Conclusions:In this work, we present a compression algorithm that accelerates electrical activity simulations on realistic anatomies by an order of magnitude (up to 20x), thereby allowing the use of finer grid resolutions while conserving GPU memory. Additionally, improved accuracy is achieved through cross-diagonal terms, which are essential for thin tissues, often found in heart structures such as pectinate muscles and trabeculae, as well as Purkinje fibers. Our method enables interactive simulations with even interactive domain boundary manipulation (unlike finite element/volume methods). Finally, agreement with experiments and ease of mesh import into WebGL paves the way for virtual cohorts and digital twins, aiding arrhythmia analysis and personalized therapies.

LEVERAGING DATA-DRIVEN TECHNIQUES FOR EFFICIENT DATA MINING IN CLOUD COMPUTING ENVIRONMENTS

The capacity to efficiently use big data and analytics is becoming a critical differentiator for company growth in today's data-driven environment. Using important trends, obstacles, and best practices as a framework, this article investigates how to promote company growth via the use of big data and analytics. An important issue in cloud computing is deciding on an acceptable amount and location of data. Decisions about resource management are based on data aspects and operations in data-driven infrastructure management (DDIM), a novel solution to this problem. It is critical to have a unified system that can manage various forms of big data and the analysis of that data, as well as common knowledge management functions. The approach stated in this research is DD-DM-CCE, or Data-Driven Methods for Efficient Data Mining in Cloud Computing Environments. Improving data using derived information from maximum frequent correlated pattern mining is the main focus of the work. By considering the centrality factor, the DD-DM-CCE method may help choose the best locations to store data in order to reduce access latency. In order to gain a competitive edge, this study offers a cloud-based conceptual framework that can analyze large data in real time and improve decision making. Efficient big data processing is possible with cloud computing infrastructures that can store and analyze massive amounts of data, as this reduces the upfront cost of the massively parallel computer infrastructure needed for big data analytics. According to simulations run on cloud computing, the DD-DM-CCE approach does better than the status quo regarding hit ratio, effective network utilization, and average response time. According to this study, data mining methods are valuable and successful in predicting how consumers will utilize cloud services.

S-слівна арифметика та високоточні обчислення

The intricacies of using S-word arithmetic, the influence of the value of the parameter S on the estimation of the rounding error are analyzed; what are high-precision calculations and where they are used. The problems of two-key cryptography, computer steganography and the problem of transcomputational complexity are considered as areas of application of S-word arithmetic. For the development of S-word arithmetic algorithms, sequential, parallel, quantum computing models are used, and systems of residual classes are used. The architectural features of the computer system for the implementation of an effective algorithm in various models of calculations are considered. For the parallel computing model, the importance of reducing the connected steps is indicated, which can increase the amount of processed data, but allows to involve a larger number of parallel processors. This approach is in conflict with a method that reduces the amount of processed data, and there is a need to maintain a balance between these two methods in a parallel computing model. For the quantum computing model, the connection of qubits is a key factor in determining the quantum volume. The physical scheme determines which pairs of qubits can be entangled in a quantum computer. Peculiarities of transferring algorithms to another computing model are considered. An analysis of the complexity of implementing S-word arithmetic operations in sequential, parallel, and quantum computing models is carried out. For the parallel computing model, the importance of reducing the connected steps is indicated, which can increase the amount of processed data, but allows to involve a larger number of parallel processors. This approach is in conflict with a method that reduces the amount of processed data, and there is a need to maintain a balance between these two methods in a parallel computing model. For the quantum computing model, the connection of qubits is a key factor in determining the quantum volume. The physical scheme determines which pairs of qubits can be entangled in a quantum computer. Information is provided about the ongoing scientific forum «Calculation optimization issues», the subject of which is closely related to the topic (1969-2023)

Двоїстий алгоритм пошуку простих чисел на відрізках великих розмірностей

A matrix model of subsequence of natural numbers with a multiplicative basis from the first prime numbers is proposed. The matrix model is a square matrix, where vector columns are arithmetic progressions with difference and number of progressions equal to product of base elements. By crossing out arithmetic progressions with first terms which are multiples of base elements, we obtain a symmetric sparse matrix that contains all the prime numbers of subsequence of natural numbers, except for the base ones, which increases the density of prime numbers in sparse matrices. Sparse matrices are not formed clearly, only the vector of first terms of arithmetic progressions is formed. Properties of sparse matrices have been proven. Formulas that speed up the calculation of compound numbers in arithmetic progressions have been derived, the structures of vector elements of first terms of arithmetic progressions have been determined, the connectivity of symmetric parts of sparse matrices has been investigated. With the expansion of the base the number of pairs of elements with the difference equal to the power of two («twins», «fours», etc.) increases in the vector of first terms. This is a necessary condition for the existence of constants for which linear equations of two variables can have an infinite set of solutions in prime numbers. The irregularity of distribution of prime numbers in subsequences of natural numbers is related to the structure of elements of the vector of first terms. An algorithm for finding prime numbers on segments of large dimensions with a parallel calculation process has been built. The proposed algorithm is binary to algorithms for sifting subsequences of natural numbers by prime divisors. In these algorithms, it is not possible to parallelize the calculation process, since screening procedure requires storing the numerical information from the preceding steps (vector model of array processing). The binary algorithm calculates compound numbers in each pair of arithmetic progressions with symmetric first terms simultaneously, using only vector of the first terms of arithmetic progressions, which makes possible processing of large-dimensional arrays

A novel multi-scale gated convolutional neural network based on informer for predicting the remaining useful life of rotating machinery

Abstract Recently, neural networks based on deep learning techniques have been employed for Remaining Useful Life (RUL) prediction of rotating machinery. However, there are some limitations: 1) CNNs primarily use local receptive fields to extract features, which leads to a relative weakness in modeling long-term dependencies on a global scale; 2) Transformers face difficulties when dealing with uncertainties such as anomalies in input sequences, missing data, or variations in sampling frequency. This paper introduces a informer-based multi-scale gated convolutional network (MSGCN-Informer). Initially, a multi-scale gated convolution module is constructed to effectively extract features across various levels, adeptly capturing temporal patterns and long-term dependencies within the dataset. Subsequently, the derived multi-scale degradation features are utilized in predicting the RUL through an informer network, thereby enhancing the efficiency of parallel computing. To validate the effectiveness and superiority of this method, comparative experiments were conducted using two publicly available bearing datasets and various model approaches.

New γ-Ray Sources Identified in All-sky Surveys Based on Fermipy’s Advanced Algorithm

We employ an efficient method for identifying γ-ray sources across the entire sky, leveraging advanced algorithms from Fermipy, and cleverly utilizing the Galactic diffuse background emission model to partition the entire sky into 72 regions, thereby greatly enhancing the efficiency of discovering new sources throughout the sky through multi-threaded parallel computing. After confirming the reliability of the new method, we applied it for the first time to analyze data from the Fermi Large Area Telescope (Fermi-LAT) encompassing approximately 15.41 yr of all-sky surveys. Through this analysis, we successfully identified 1379 new sources with significance levels exceeding 4σ, of which 497 sources exhibited higher significance levels exceeding 5σ. Subsequently, we performed a systematic analysis of the spatial extension, spectra, and light variation characteristics of these newly identified sources. We identified 21 extended sources and 23 sources exhibiting spectral curvature above 10 GeV. Additionally, we identified 44 variable sources above 1 GeV.

Theoretical Basis of Mathematical Apparatus for Parallel Computing Implementation in Computer-Aided Design Systems

Theoretical Basis of Mathematical Apparatus for Parallel Computing Implementation in Computer-Aided Design Systems

On the Geodetic and Strong Geodetic Parameters of Certain Hypercubic Networks

Networks derived from the hypercube are noted for their efficiency in simulating any bounded-degree communication network and hence they play a significant role in parallel machines. For a graph [Formula: see text], a geodetic cover is a subset [Formula: see text] of [Formula: see text], satisfying the condition that every vertex of [Formula: see text] other than the vertices in [Formula: see text] are covered using geodesics connecting any two vertices in [Formula: see text]. The geodetic number of [Formula: see text] is the least order of such covers in [Formula: see text]. This seminal distance-based parameter is well employed in the areas of location theory, convexity theory and game theory. The strong variant and its edge version are recent variations of the problem that were introduced in due of their applications in social networks. The most established and strong hypercube derivative networks are studied in this paper with respect to certain geodetic parameters. Notably, among the networks that we have considered, the butterfly and the Benes networks belong to the class of bipartite graphs for which the geodetic number problem and its strong version are NP-complete. We have determined the geodetic parameters that include the geodetic number, the strong geodetic number and their corresponding edge variants of the networks taken into consideration. In the process, we have identified a new class of graphs, as semi-extreme edge geodesic graphs and that the enhanced butterfly networks belong to this class. The study of such geodetic parameters in the structural models of parallel computation could be useful for the design of efficient algorithms for the structures considered.

A multi-constraint optimal puncture path planning algorithm for percutaneous interventional radiofrequency thermal fusion of the L5/S1 segments

To minimize variations in treatment outcomes of L5/S1 percutaneous intervertebral radiofrequency thermocoagulation (PIRFT) arising from physician proficiency and achieve precise quantitative risk assessment of the puncture paths. We used a self-developed deep neural network DWT-UNet for automatic segmentation of the magnetic resonance (MR) images of the L5/S1 segments into 7 key structures: L5, S1, Ilium, Disc, N5, Dura mater, and Skin, based on which a needle insertion path planning environment was modeled. Six hard constraints and 6 soft constraints were proposed based on clinical criteria for needle insertion, and the physician's experience was quantified into weights using the analytic hierarchy process and incorporated into the risk function for needle insertion paths to enhance individual case adaptability. By leveraging the proposed skin entry point sampling sub-algorithm and Kambin's triangle projection area sub-algorithm in conjunction with the analytic hierarchy process, and employing various technologies such as ray tracing, CPU multi-threading, and GPU parallel computing, a puncture path was calculated that not only met clinical hard constraints but also optimized the overall soft constraints. A surgical team conducted a subjective evaluation of the 21 needle puncture paths planned by the algorithm, and all the paths met the clinical requirements, with 95.24% of them rated excellent or good. Compared with the physician's planning results, the plans generated by the algorithm showed inferior DIlium, DS1, and Depth (P < 0.05) but much better DDura, DL5, DN5, and AKambin (P < 0.05). In the 21 cases, the planning time of the algorithm averaged 7.97±3.73 s, much shorter than that by the physicians (typically beyond 10 min). The multi-constraint optimal puncture path planning algorithm offers an efficient automated solution for PIRFT of the L5/S1 segments with great potentials for clinical application.

Deep neural network based distribution system state estimation using hyperparameter optimization

In the past decade, distribution system state estimation has become a crucial topic in power system research due to the increasing importance of distribution networks amidst the decline of centralized energy production. This paper addresses a gap in the literature regarding the application of modern hyperparameter optimization techniques in low-voltage distribution system state estimation using deep neural networks. In particular, it demonstrates the use of the Tree-structured Parzen Estimator algorithm, which is a Bayesian hyperparameter optimization method, for distribution system state estimation on real Hungarian low-voltage networks. The study uses data from four real-life low-voltage supply areas in Hungary, which were modeled to address the challenges in obtaining network information. Compared to traditional methods like the weighted least squares method, the Tree-structured Parzen Estimator algorithm significantly improves the accuracy of the voltage amplitude and angle estimations, reducing the relative error by 14–73%. Additionally, it is shown that TPE outperforms simpler methods like Random Search in hyperparameter optimization. The results also reveal connections between the distribution system size and optimal hyperparameters, such as batch size, learning rate, and hidden layer configuration. The proposed non-iterative algorithm, combined with the parallel computation capabilities of deep neural networks utilizing GPU, resulted in four orders of magnitude improvement in runtime. These advancements make the proposed approach a valuable tool for renewable energy integration planning and real-time monitoring, highlighting its potential for practical applications in the power industry.

A novel method of high-speed all-optical logic gate based on metalens

This paper introduces a novel approach to constructing all-optical logic gates based on metalens. The designed structure enables the realization of five commonly used logic gates (AND, OR, NOT, XOR, XNOR). Each logic gate is composed of two to three metalenses. The advantages of this approach are as follows: 1. Metalens offer flexible phase control capabilities, overcoming the strict phase requirements in traditional optical logic gate design. 2. The miniaturization of the metalens makes integration possible and provides a potential method for high-speed parallel optical computing. 3. By controlling different types and quantities of metalens, various types of logic gates can be formed, enhancing programmability during use. The contrast values for the designed logic gates are as follows: 27.95 dB (OR), 18.18 dB (AND), 10.83 dB (NOT), 10.29 dB (XNOR), and 13.56 dB (XOR). With similar structures, the bit rates for the five logic gates at a 50% duty cycle range from 1.04 to 1.05 Tb/s. Overall, these results demonstrate a successful balance between contrast and transmission speed when utilizing metalens to realize all-optical logic gates. From the results, the logic gate proposed in this paper has high contrast and transmission speed, which proves the rationality of using metalens to realize all-optical logic gate.