Parallel Scheme Research Articles

Parallel Impedance-Reshaping Control Strategy for Suppressing Wide Range High-Frequency Resonance of Shunt Compensated DFIG System

The shunt compensated grid (SCG) connecting with doubly-fed induction generator (DFIG) system easily suffers from high-frequency resonance (HFR). Existing studies reshape the DFIG based on a pre-calculated frequency to suppress the HFR. When facing voltage disturbance, shunt capacitors are added to the grid which greatly changes the resonance frequency, probably resulting functional failure of HFR suppression. In this paper, a parallel impedance-reshaping control (PIRC) strategy for the rotor side converter (RSC) and grid-side converter (GSC) control is proposed to suppress the HFR of the DFIG-SCG system when the shunt capacitor varies in a wide range. A high-order coefficient is combined with the original impedance of the RSC-induction generator (RSC-IG) side, and a damping controller is designed to reshape the impedance of the DFIG system. Based on the parallel structure of the equivalent impedance of the GSC side and the RSC-IG side, the impact of the shunt capacitor to the controller can be eliminated, thus providing effective HFR suppression in a wide frequency range. Case studies demonstrate the advantage of the PIRC in HFR suppression and harmonic suppression.

Optimal edge-cloud collaboration based strategies for minimizing valid latency of railway environment monitoring system

Response speed is vital for the railway environment monitoring system, especially for the sudden-onset disasters. The edge-cloud collaboration scheme is proved efficient to reduce the latency. However, the data characteristics and communication demand of the tasks in the railway environment monitoring system are all different and changeable, and the latency contribution of each task to the system is discrepant. Hence, two valid latency minimization strategies based on the edge-cloud collaboration scheme is developed in this paper. First, the processing resources are allocated to the tasks based on the priorities, and the tasks are processed parallelly with the allocated resources to minimize the system valid latency. Furthermore, considering the differences in the data volume of the tasks, which will induce the waste of the resources for the tasks finished in advance. Thus, the tasks with similar priorities are graded into the same group, and the serial and parallel processing strategies are performed intra-group and inter-group simultaneously. Compared with the other four strategies in four railway monitoring scenarios, the proposed strategies proved latency efficiency to the high-priority tasks, and the system valid latency is reduced synchronously. The performance of the railway environment monitoring system in security and efficiency will be promoted greatly with the proposed scheme and strategies.

Chinese holistic processing: Evidence from cognitive mental architecture using Systems Factorial Technology

Previous research has presented conflicting evidence regarding whether Chinese characters are processed holistically. In past work, we applied Systems Factorial Technology (SFT) and discovered that native Chinese speakers exhibited limited capacity when processing characters and words. To pinpoint the source of this limitation, our current research delved further into the mental architecture involved in processing Chinese characters and English words, taking into consideration information from each component. In our current study, participants were directed to make the same/different judgments on characters/words presented sequentially. Our results indicated that participants utilized a parallel self-terminating strategy when both or neither of the left/right components differed (Experiment 1). Faced with the decisional uncertainty that either the left/right component would also differ, most participants processed with a parallel exhaustive architecture, while a few exhibited the coactive architecture (Experiment 2). Taken together, our work provides evidence that in word/character perception, there is weak holistic processing (parallel self-terminating processing) when partial information is sufficient for the decision; robust holistic processing (coactive or parallel exhaustive processing) occurs under decisional uncertainty. Our findings underscore the significant role that the task and presentation context play in visual word processing.

Impact of multi-GPUs and hyperparameters on parallel algorithms based on pytorch distributed

Neural networks, as powerful models for solving many genuine problems, are suffering from the issue of being too computationally intensive due to the excessive size of the models themselves and the large datasets used for training. This paper’s work focuses on a parallel algorithm, experimented on the classical convolutional neural network LeNet-5, aiming to reduce the time it takes for the neural network to train. This applies a massively parallel GPU structure to take full advantage of its high computing power. Among the approaches to improve the performance of parallel algorithms, many efforts have been made to improve the design of parallel structures, while this work focuses more on the enhancement of parallel algorithms by reasonable settings of neural network hyperparameters. This paper tries to approach (1) reducing the running time by testing parallelism with different numbers of GPUs and (2) further investigating the relationship between parallelism and model accuracy degradation and experimentally deriving a parallel scheme that can reduce the training time while guaranteeing accuracy.

Using Adaptiveness and Causal Superpositions Against Noise in Quantum Metrology.

We derive new bounds on achievable precision in the most general adaptive quantum metrological scenarios. The bounds are proven to be asymptotically saturable and equivalent to the known parallel scheme bounds in the limit of a large number of channel uses. This completely solves a long-standing conjecture in the field of quantum metrology on the asymptotic equivalence between parallel and adaptive strategies. The new bounds also allow us to easily assess the potential benefits of invoking nonstandard causal superposition strategies, for which we prove, similarly to the adaptive case, the lack of asymptotic advantage over the parallel ones.

Singular value decomposition for longitudinal, transverse and mixed ray transforms of 2D tensor fields

The operators of longitudinal, transverse and mixed ray transforms acting on two-dimensional symmetric tensor fields of arbitrary degree m in an unit disk are considered in the article. The singular value decompositions of the operators for a parallel scheme of data acquisition are constructed. Orthogonal bases in original spaces and image spaces are constructed using harmonic, Jacobi and Gegenbauer polynomials. Based on the obtained decompositions the polynomial expressions for the (pseudo)inverse and adjoint operators are obtained.

A column generation approach for the team formation problem

We address a cooperative multi-team formation problem where there are multiple teams in an organization, and teams can be reformed by exchanging members among each other. We formulate this problem as a nonlinear integer programming model, by later reformulating it, which prescribes the optimal exchange decisions for all teams that maximizes the minimum resulting team value across all teams. For this problem, in order to obtain tighter dual bounds, we propose a column generation approach where each column represents specific exchange decisions of a team pair, and at each iteration, a set of attractive exchange patterns for all team pairs are generated by a series of subproblems. We implement a parallel processing scheme to solve all subproblems in order to further accelerate the procedure. In addition, to obtain near-optimal solutions, we propose a column generation-based methodology, where at the last iteration, the restricted master problem is solved as an integer programming model with all generated columns. Our results show that the column generation procedure provides dual bound improvements ranging between 2.44% and 5.31%, on average, over the linear programming relaxation of the original model for our generated instances. In addition, the proposed column generation-based methodology yields near-optimal, and in most cases, optimal solutions by providing CPU savings ranging from 71.6% to 89.2%, on average, over the CPU times of the original integer model.

Digital twin-driven identification of fault situation in distribution networks connected to distributed wind power

The gradual integration of distributed wind power into distribution networks presents significant challenges for identifying faults, as it requires accurate and timely fault identification based on large-scale data. This paper proposes a novel fault situation identification framework driven by digital twins to overcome the existing bottleneck of asynchronous online fault identification and offline post-event analysis. The framework employs a high-precision digital twin avatar with parallel strategies of fully electromagnetic transient calculation to obtain the real-time operation status of distribution networks connected with distributed wind power. Furthermore, the framework employs skip-connected dilated causal convolution to mine meaningful multi-time-scale features from massive data, while using an automatic hyper-parameter tuning strategy. A case study based on IEEE 33-node standard distribution network and a real-world distribution network demonstrates the effectiveness of this framework, achieving ultra-real-time and high-precision simulation and effective fault identification even under noise or missing data, accelerating traditional post-fault analysis to microsecond-level situation identification.

Nonlinear reduced-order modeling for three-dimensional turbulent flow by large-scale machine learning

A large-scale machine learning-based nonlinear reduced-order modeling method was developed for a three-dimensional turbulent flow field (Re=1000) using a neural-network with unsupervised learning. First, a mode decomposition method was applied to three-dimensional flow field data using a convolutional-autoencoder-like neural network. Then, a reduced-order model (ROM) was constructed using long short-term memory neural networks (LSTMs). Consequently, it was demonstrated that the time evolution of the turbulent three-dimensional flow field can be simulated at a significantly lower cost (approximately three orders of magnitude) without a major loss in accuracy. However, neural-network-based mode decomposition for the three-dimensional flow field requires a huge computational cost in terms of calculation and memory usage. Therefore, a distributed machine-learning method was implemented using a hybrid parallelism scheme tailored to the network structure. Thus, it was possible to decompose 1.7 million cells of the three-dimensional flow field data into 64 modes and reproduce those with sufficient accuracy. In this study, a uniform flow around a circular cylinder model was used as a test case. To validate the method, the reduction performance of the proposed mode decomposition method was compared with the proper orthogonal decomposition (POD) method. Furthermore, the target flow field was reproduced using ROM, and the reconstruction accuracy was evaluated in terms of various criteria compared with the accuracy based on POD in conjunction with Galerkin projection method.

Greening as a competitiveness optimizer of servitization

Servitizing firms are increasingly pressured by internal, political and market forces to adopt a parallel green strategy in the hopes of optimizing their competitive response to evolving market demands. But servitization and greening potentially represent divergent shifts in companies' operations and product/service offerings. This study aims to determine whether firms should implement a green investment strategy when already pursuing a complex strategic path of servitization. The study uses the non-parametric ‘benefit-of-the-doubt’ method to compare the competitive efficiency of firms implementing servitized versus non-servitized strategies, and between firms with and without green investment strategies. The study finds that firms pursuing both servitization and a green strategy had higher levels of competitive efficiency compared to those pursuing only one strategy or none at all. The results suggest that the combination of these two strategies generates competitive synergies that optimize overall competitive performance, contradicting the theoretical principles of the resource-based view and complexity perspectives. This result was found to be consistent throughout all analyzed OECD countries.

Comparative analyses of covariance matrix adaptation and iterative ensemble smoother on high-dimensional inverse problems in high-resolution groundwater modeling

Parameter estimation is an inverse problem which is crucial to reliable groundwater model predictions and management. Numerous techniques have been developed to address this challenging problem. This study aims to compare the performance of a stochastic optimization method, covariance matrix adaptation-evolution strategy (CMA-ES), and a data assimilation method, Levenberg-Marquardt based iterative ensemble smoother (ES-LM), on solving inverse problems in groundwater modeling. This study also presents a parallelization strategy to accelerate their implementation. The comparative analysis involves synthetic inverse problems and a real-world case study of calibrating a high-resolution groundwater model. Results from the synthetic problems suggest that both the CMA-ES and the ES-LM are able to achieve desirable data fitting, but the ES-LM generally exhibits greater efficiency than the CMA-ES. Also, with appropriate stopping criteria, both methods can perform well on propagating uncertainty of model predictions. However, both methods can underestimate the uncertainty of model parameters derived from Bayesian posterior. The underestimation is particularly pronounced by the ES-LM when using a small ensemble size, whereas uncertainty quantification by the CMA-ES is less affected by population size. Moreover, calibration of the highly parameterized groundwater model in Louisiana and Southwest Mississippi indicates that the ES-LM can outperform the CMA-ES for high-dimensional inverse problems regarding both data match and computational costs. With proper ensemble size, the two methods can produce comparable uncertainty of model parameters and predictions. This study contributes to the scientific understanding and efficient application of the CMA-ES and the ES-LM.

LPS-Net: Lightweight Parallel Strategy Network for Underwater Image Enhancement

Underwater images are frequently subject to color distortion and loss of details. However, previous enhancement methods did not tackle these mixed degradations by dividing them into sub-problems that could be effectively addressed. Moreover, the parameters and computations required for these methods are usually costly for underwater equipment, which has limited power supply, processing capabilities, and memory capacity. To address these challenges, this work proposes a Lightweight Parallel Strategy Network (LPS-Net). Firstly, a Dual-Attention Enhancement Block and a Mirror Large Receptiveness Block are introduced to, respectively, enhance the color and restore details in degraded images. Secondly, we employed these blocks on parallel branches at each stage of LPS-Net, with the goal of achieving effective image color and detail rendering simultaneously. Thirdly, a Gated Fusion Unit is proposed to merge features from different branches at each stage. Finally, the network utilizes four stages of parallel enhancement, achieving a balanced trade-off between performance and parameters. Extensive experiments demonstrated that LPS-Net achieves optimal color enhancement and superior detail restoration in terms of visual quality. Furthermore, it attains state-of-the-art underwater image enhancement performance on the evaluation metrics, while using only 80.12 k parameters.

Large-scale parallel topology optimization of three-dimensional incompressible fluid flows in a level set, anisotropic mesh adaptation framework

This papers considers the topology optimization of duct flows governed by the three-dimensional steady state Navier–Stokes equations, using anisotropic mesh adaptation to achieve a high-fidelity description of all fluid–solid interfaces. The numerical framework combines an immersed volume method solving stabilized, linear equal-order finite element formulations cast in the Variational Multiscale (VMS) framework, and level set representations of the interface, used as a posteriori anisotropic error estimator to minimize the interpolation error under the constraint of a prescribed number of nodes in the mesh. Both the resolution and remeshing steps are performed in a massively parallel framework allowing for the optimization of large-scale systems. In particular, an original parallelization strategy is used for mesh adaptation, that combines local remeshing performed sequentially and independently on each subdomain with blocked interfaces, and constrained repartitioning to optimally move the interfaces between subdomains in an optimal way, both iterated until a satisfying mesh and partition are obtained. The proposed approach reduces the computational burden related to the call of the finite element solver, compared to classical optimization schemes working on uniform grids with similar mesh refinement. For a given number of nodes, it improves the accuracy in the geometric description of all layouts. Finally, it has the potential to alleviates the end user from most of the post-processing step aiming at extracting the final layout, due to ability of anisotropic adapted meshes to generate intrinsically smooth designs. Numerical results are provided for several three-dimensional problems of power dissipation minimization involving several dozen million state degrees of freedom, for which the optimal designs agree well with reference results from the literature, while providing superior accuracy over prior studies solved on isotropic meshes (in the sense that the flow is better resolved, especially in the near-wall regions, and the layouts are more smooth). The potential of the method for engineering problems of practical interest is eventually exposed by optimizing the distributor section conveying the cold fluid within the plates of a plate fin heat exchanger.

Optimizing Single DGX-A100 System: Overcoming GPU Limitations via Efficient Parallelism and Scheduling for Large Language Models

In this study, we introduce a novel training algorithm specifically designed to overcome the limitations of GPU memory on a single DGX-A100 system. By utilizing the CPU and main memory in the training process and applying a strategy of division and parallelization, our algorithm enhances the size of the trainable language model and the batch size. In addition, we developed a comprehensive management system to effectively manage the execution of the algorithm. This system systematically controls the training process and resource usage, while also enabling the asynchronous deployment of tasks. Finally, we proposed a scheduling technique integrated into the management system, promoting efficient task scheduling in a complex, heterogeneous training environment. These advancements equip researchers with the ability to work with larger models and batch sizes, even when faced with limited GPU memory.

Improved Parallel Implementation of 1D Discrete Wavelet Transform Using CPU-GPU

This work describes a data-level parallelization strategy to accelerate the discrete wavelet transform (DWT) which was implemented and compared in two multi-threaded architectures, both with shared memory. The first considered architecture was a multi-core server and the second one was a graphics processing unit (GPU). The main goal of the research is to improve the computation times for popular DWT algorithms for representative modern GPU architectures. Comparisons were based on performance metrics (i.e., execution time, speedup, efficiency, and cost) for five decomposition levels of the DWT Daubechies db6 over random arrays of lengths 103, 104, 105, 106, 107, 108, and 109. The execution times in our proposed GPU strategy were around 1.2×10−5 s, compared to 3501×10−5 s for the sequential implementation. On the other hand, the maximum achievable speedup and efficiency were reached by our proposed multi-core strategy for a number of assigned threads equal to 32.

Layer-wise partitioning and merging for efficient and scalable deep learning

Deep Neural Network (DNN) models are usually trained sequentially from one layer to another, which causes forward, backward and update locking problems, leading to poor performance in terms of training time. The existing parallel strategies to mitigate these problems provide suboptimal runtime performance. In this work, we have proposed a novel layer-wise partitioning and merging, forward and backward pass parallel framework to provide better training performance. The novelty of the proposed work consists of (1) a layer-wise partition and merging model which can minimise communication overhead between devices without the memory cost of existing strategies during the training process; (2) a forward pass and backward pass parallelisation to address the update locking problem and minimise the total training cost. The experimental evaluation on real use cases shows that the proposed method outperforms the state-of-the-art approaches in terms of training speed; and achieves almost linear speedup without compromising the accuracy performance of the non-parallel approach.

Hemodynamics of different configurations of the left subclavian artery parallel stent graft for thoracic endovascular aortic repair

Background and objectiveParallel (chimney and periscope) graft technique is an effective approach for left subclavian artery (LSA) reconstruction in patients treated by thoracic endovascular aortic repair (TEVAR) for the inadequate landing zone. However, certain stent graft (SG) configurations may promote thrombosis and reduce distal blood flow, increasing risks of cerebral infarction and reintervention. MethodsIn this paper, we first attempt to systematically evaluate the hemodynamic performances of different parallel graft techniques as potential determinants of complication risks. Based on the patient-specific 3D aortic geometry undergoing parallel graft technique, fifteen models in total for five kinds of LSA branched SG configurations (Forward, Backward, Extended, Elliptical and Periscopic) were designed virtually, and the hemodynamic discrepancies between them were analyzed by computational fluid dynamics. ResultsResults show that flow rate of patients undergoing periscope technique reduces by half compared with chimney technique, suggesting that periscope SG may cause more serious flow obstruction to LSA, leading to stroke. For chimney stent structure, the extension length 0has little influence on energy loss and other parameters. Conversely, hemodynamic differences between the retrograde curvature and the antegrade curvature are significant (time average WSS: 47.07%), so the retrograde curvature might prompt SG displacement. Furthermore, the flatter chimney SG induces more aggressive hemodynamic forces, among which the difference of the maximum WSS between the flatter SG and nearly round SG reaches 65.56%, leading to the greater risk of vascular wall damage. ConclusionsResults obtained might provide suggestions for physicians to formulate appropriate parallel graft technique schemes in TEVAR.

Parallel optimization method of train scheduling and shunting at complex high‐speed railway stations

AbstractThe train operations of large stations are critical in determining the efficiency of the railway network. Large high‐speed railway stations often have more than two adjacent stations running in multiple directions and must address highly complex train operation patterns. To develop a more efficient operation plan than the existing route‐based representations for modeling train conflicts, a more systematic track‐based resource network (TRN) representation for a railway station that simulates fixed equipment is used in this paper. Based on this high‐fidelity TRN, an improved operation scheduling model (OSM) is introduced for the integrated scheduling of train operations, including shunting operations. Besides, two approaches based on Lagrangian relaxation (LR) and the Alternating Direction Method of Multipliers (ADMM) are described in detail to address the operation scheduling problem (OSP) for stations, and an efficient heuristic rule‐based algorithm is also designed to generate lower bound solutions. To improve the algorithm efficiency, we designed a parallelization strategy for the proposed approaches. Based on real‐world case studies, the computing results demonstrate that the parallelized ADMM‐based approach can always find high‐quality feasible solutions for complex stations and generate conflict‐free station operation plans with significantly improved efficiency.

Parallelization of a 3D FDTD code and physics studies of electromagnetic wave propagation in fusion plasmas

Numerical codes for electromagnetic wave propagation in magnetized plasmas are mainly based on frequency-domain asymptotic methods, which provide a fast solution and are thus valuable for experiment design and control applications. However, in several cases of practical interest (e.g. mode conversion), these tools run close to their limits of validity and should be compared to full-wave solutions. The code RFFW solves Maxwell's equations with the finite-difference time-domain method in 3D geometry, for scenarios involving high-frequency waves with arbitrary electric field spectrum in plasmas with axisymmetric equilibrium. In fusion-related problems, the code may conduct investigations of wave propagation and absorption relevant to auxiliary plasma heating and current drive, reflectometry and instability control. The code has been parallelized with a hybrid OpenMP-MPI scheme, which has allowed exploiting the much larger processing power and memory of current-day supercomputers. In this work, we present the main aspects of the physics implemented in the code, and also refer shortly to the parallelization scheme. Furthermore, we show results that exhibit the strong scaling performance of the code, and examine cases of electron-cyclotron heating application in medium-sized tokamaks.

A Unified Arbitrary Style Transfer Framework via Adaptive Contrastive Learning

This work presents Unified Contrastive Arbitrary Style Transfer (UCAST), a novel style representation learning and transfer framework, that can fit in most existing arbitrary image style transfer models, such as CNN-based, ViT-based, and flow-based methods. As the key component in image style transfer tasks, a suitable style representation is essential to achieve satisfactory results. Existing approaches based on deep neural networks typically use second-order statistics to generate the output. However, these hand-crafted features computed from a single image cannot leverage style information sufficiently, which leads to artifacts such as local distortions and style inconsistency. To address these issues, we learn style representation directly from a large number of images based on contrastive learning by considering the relationships between specific styles and the holistic style distribution. Specifically, we present an adaptive contrastive learning scheme for style transfer by introducing an input-dependent temperature. Our framework consists of three key components: a parallel contrastive learning scheme for style representation and transfer, a domain enhancement (DE) module for effective learning of style distribution, and a generative network for style transfer. Qualitative and quantitative evaluations show the results of our approach are superior to those obtained via state-of-the-art methods. The code is available at https://github.com/zyxElsa/CAST_pytorch .