Spatial Decomposition Technique Research Articles

Study on Socio-Economic Factors of Haze Pollution in the Context of Sustainable Development: Based on Asymmetric Interaction Effects Perspective

Haze pollution has been a global problem plaguing people for nearly two centuries, especially in developing countries, where haze pollution is getting worse and has become the culprit that restricts global sustainable development. As the largest developing country in the world, China’s regional characteristics of haze pollution are increasingly prominent, and the coverage is expanding, which means that there is a strong spatial interaction of haze pollution. The real world is composed of asymmetric spatial relationships, and asymmetric spatial interactions are more consistent with the reality of haze pollution. To this end, this paper identifies the asymmetric spatial interaction effects of haze pollution by using the time series econometric analysis framework and then constructs asymmetric spatial weights. In this paper, the spatial panel Durbin model and spatial effect decomposition technique are used for the empirical investigation to provide new empirical evidence to reveal the socio-economic influences of haze pollution. The study finds that haze pollution in China has a significant asymmetric spatial interaction and the degree of influence varies among cities. During the sample period, there is a significant inverted U-shaped relationship between economic development and haze pollution. However, the vast majority of cities have not passed the top inflection point of the inverted U-curve and are still in the stage where haze pollution is gradually intensifying with economic growth. High population density, high energy consumption, and unreasonable energy consumption structure aggravate haze pollution, while fixed-asset investment and science and technology expenditure help to reduce haze pollution. However, only the spatial spillover effect of the energy consumption structure is significant. Therefore, the improvement and implementation of more effective socio-economic prevention and control policies in the case of asymmetric interaction is the key to effectively dealing with heavy pollution weather, fighting the defense of the blue sky, and achieving sustainable development.

Counterfactual dissimilarity: Can changes in demographics and income explain increased racial integration in US cities?

AbstractUrban areas in the US have experienced important changes in racial/ethnic distributions over the last two decades. In the average urban area today black–white racial integration has increased by 10.6% between 1990 and 2010. Changes in racial and ethnic distributions and gentrification are often associated with changes in residents' demographic characteristics, such as income, education, and age. This paper applies a nonparametric spatial decomposition technique using complete (restricted‐use) microdata files from the 1990 Decennial Long Form Census and 2008–2012 American Community Surveys to assess what portion of the changes in racial distributions can be attributed to changes in individual characteristics. We find that that, on average, a little over a third of the observed increase in integration can be accounted for by changes in observed individual characteristics.

Enabling Resource-Aware Mapping of Spiking Neural Networks via Spatial Decomposition

With growing model complexity, mapping spiking neural network (SNN)-based applications to tile-based neuromorphic hardware is becoming increasingly challenging. This is because the synaptic storage resources on a tile, viz., a crossbar, can accommodate only a fixed number of presynaptic connections per postsynaptic neuron. For complex SNN models that have many presynaptic connections per neuron, some connections may need to be pruned after training to fit onto the tile resources, leading to a loss in the model quality, e.g., accuracy. In this letter, we propose a novel unrolling technique that decomposes a neuron function with many presynaptic connections into a sequence of homogeneous neural units, where each neural unit is a function computation node, with two presynaptic connections. This spatial decomposition technique significantly improves crossbar utilization and retains all presynaptic connections, resulting in no loss of the model quality derived from connection pruning. We integrate the proposed technique within an existing SNN mapping framework and evaluate it using machine learning applications on the DYNAP-SE state-of-the-art neuromorphic hardware. Our results demonstrate an average 60% lower crossbar requirement, 9× higher synapse utilization, 62% lower wasted energy on the hardware, and between 0.8% and 4.6% increase in the model quality.

Modified frequency and spatial domain decomposition method based on maximum likelihood estimation

In this study, a Modified Frequency and Spatial Domain Decomposition (MFSDD) technique is developed for modal parameter identification, using output-only response measurements. According to the presented procedure, the most probable power spectral density matrix of the measured response (output PSD) is updated by a maximum likelihood estimation based on the observed data. Different from the available Frequency Domain Decomposition (FDD) techniques, a prediction error term which is associated with the measurement noise and modelling errors is included in the proposed methodology. In this context, a detailed discussion is provided from various aspects for the effect of measurement noise and modelling errors on the parameter estimation quality. Two numerical and two experimental analysis are conducted in order to demonstrate the effectiveness and accuracy of the proposed methodology under some extreme effects. The obtained results indicate that the proposed method shows very good performance in modal parameter estimation in case of noisy measurements.

Overlapping communication and computation of GPU/CPU heterogeneous parallel spatial domain decomposition MOC method

GPU/CPU heterogeneous parallel calculation has been widely used and great performance has been achieved. Recently it also has been introduced into whole-core transport calculation. Considering the ability to solve the large-scale problem, the spatial domain decomposition (SDD) technique has been applied in the whole-core transport application. However, the communication burden becomes an important bottleneck in heterogeneous parallel SDD calculation and the parallel efficiency decreases dramatically when the number of the calculation domains increasing. GPU brings great acceleration meanwhile the communication time doesn’t change, which makes the parallel efficiency of the heterogeneous parallel calculation worse. In this work, heterogeneous parallel SDD calculation is performed by applying both the spatial domain decomposition and ray parallelization and the parallel performance is analyzed. In addition, a new scheme of the communication is implemented to overlap the MPI communication time and the neutron transport sweep on GPU. In order to accomplish this scheme, both MPI, CUDA protocols are introduced. Numerical results show that the new scheme successfully hide the communication time and the data copy time between GPU and CPU, which brings significant improvement of the parallel efficiency. With the new communication scheme, the large-scale whole-core transport calculation is more practicable.

Implementation of the CPU/GPU hybrid parallel method of characteristics neutron transport calculation using the heterogeneous cluster with dynamic workload assignment

In recent years, graphics processing units (GPUs) have been adopted in many High-Performance Computing (HPC) systems due to their massive computational power and superior energy efficiency. And accelerating CPU-version computational code on heterogeneous clusters with multi-core CPUs and GPUs has attracted a lot of attention. One of the focus on heterogeneous computing is to efficiently take advantage of all computational resources, including both CPU and GPU available on a cluster. In this paper, a heterogeneous MPI + OpenMP/CUDA parallel algorithm for solving the 2D neutron transport equation with the method of characteristic (MOC) is implemented. In this algorithm, the spatial domain decomposition technique provides the coarse-grained parallelism with the MPI protocol while the fine-grained parallelism is exploited through OpenMP (in CPU calculated domain) and CUDA (in GPU calculated domain) based on the ray parallelization. In order to efficiently leverage the computing power of heterogeneous clusters, a dynamic workload assignment scheme is proposed, which is to distribute the workload based on the runtime performance of CPUs and GPUs in the cluster. Moreover, the strong scaling performance of the MPI + CUDA parallelization is studied through a performance analysis model which provides the detailed impact of the degradation in iteration scheme, the load imbalance issue, the data copy between CPUs and GPUs, and the MPI communication in the MPI + CUDA parallel algorithm. And the corresponding conclusion is still tenable for the MPI + OpenMP/CUDA parallelization. The C5G7 2D benchmark and an extended 2D whole-core problem are calculated with MPI + CUDA parallelization, MPI + OpenMP/CUDA parallelization, and the MPI parallelization for comparison. Numerical results demonstrate that the heterogeneous parallel algorithm maintains the desired accuracy. And the dynamic workload assignment scheme can provide the optimal workload assignment which ideally matches the experimental results. In addition, over 11% improvement is observed in MPI + OpenMP/CUDA parallelization compared against the MPI + CUDA parallelization. Moreover, the CPUs/GPUs heterogeneous clusters significantly outperform the CPUs clusters and one heterogeneous node shows basically five times faster than a CPUs node.

Unit-Cell-Based Domain Decomposition Method for Efficient Simulation of a Truncated Electromagnetic Bandgap Structure in High-Speed PCBs

In this paper, we present a unit-cell-based domain decomposition method (UC-DDM) for rapid and accurate simulation of predicting the parallel plate noise (PPN) suppression of a truncated electromagnetic bandgap (EBG) structure in high-speed printed circuit boards (PCBs). The proposed UC-DDM divides the analysis domain of the truncated EBG structure into UCs as sub-domains. Solving a sub-domain is based on a novel UC model, yielding an analytical expression for the impedance parameter (Z-parameter) of the UC. The novel UC model is derived using a spatial decomposition technique, which results in the modal decomposition of quasi-transverse electromagnetic (TEM) and transverse magnetic (TM) modes. In addition, we analytically derive a impedance-parameter recombination method (ZRM) to obtain the analytical solution of a finite EBG array from the sub-domain results. The proposed UC-DDM is verified through comparison with full-wave simulation results for various EBG arrays. Comparison between the UC-DDM and a full-wave simulation of a truncated EBG structure reveals that a substantial improvement in computation time with high accuracy is achieved. It is demonstrated that the simulation time of the proposed method is only 0.1% of that of a full-wave simulation without accuracy degradation.

Towards accelerating smoothed particle hydrodynamics simulations for free-surface flows on multi-GPU clusters

Starting from the single graphics processing unit (GPU) version of the Smoothed Particle Hydrodynamics (SPH) code DualSPHysics, a multi-GPU SPH program is developed for free-surface flows. The approach is based on a spatial decomposition technique, whereby different portions (sub-domains) of the physical system under study are assigned to different GPUs. Communication between devices is achieved with the use of Message Passing Interface (MPI) application programming interface (API) routines. The use of the sorting algorithm radix sort for inter-GPU particle migration and sub-domain “halo” building (which enables interaction between SPH particles of different sub-domains) is described in detail. With the resulting scheme it is possible, on the one hand, to carry out simulations that could also be performed on a single GPU, but they can now be performed even faster than on one of these devices alone. On the other hand, accelerated simulations can be performed with up to 32 million particles on the current architecture, which is beyond the limitations of a single GPU due to memory constraints. A study of weak and strong scaling behaviour, speedups and efficiency of the resulting program is presented including an investigation to elucidate the computational bottlenecks. Last, possibilities for reduction of the effects of overhead on computational efficiency in future versions of our scheme are discussed.

Convergence of certain class of enhanced spatial domain decompositions

The analysis as a stationary method of the spatial domain decomposition (SDD) technique with several enhancements is studied in this article. In particular, we consider two of the most popular modifications of the SDD method: (i) extension of the domains and (ii) introduction of a relaxation parameter. Despite both modifications, we show that the method can still be formulated as a stationary method and, therefore, the convergence study can be derived from the iteration matrix. Several examples for three-dimensional geometries are considered to check the validity of the formulation. © 2011 Wiley Periodicals, Inc. Microwave Opt Technol Lett 53:2388–2393, 2011; View this article online at wileyonlinelibrary.com. DOI 10.1002/mop.26245

Fast Analysis of Multiple Antennae Coupling on Very Electrical Large Objects Via Parallel Technique

A fast parallel algorithm for calculating the coupling of multiple antennae on very electrical large objects is presented. While using this algorithm a electrical large object is divided into a series of sub-zones via spatial decomposition technique (SDT) and then the current on each sub-zone can be calculated by multilevel fast multipole algorithm (MLFMA). By using the sub-zone currents the coupling among antennae can be computed then through a two-port network model. The parallel algorithm presented is executed on a cluster of personal computers and both the numerical and experiment results show that this algorithm is very fast with high precision.

Fast Spatial Decomposition and Closest Pair Computation for Limited Precision Input

In this paper we show that if the input points to the geometric closest pair problem are given with limited precision (each coordinate is specified with O( log n) bits), then we can compute the closest pair in O(n log log n) time. We also apply our spatial decomposition technique to the k -nearest neighbor and n -body problems, achieving similar improvements. To make use of the limited precision of the input points, we use a reasonable machine model that allows ``bit shifting'' in constant time—we believe that this model is realistic, and provides an interesting way of beating the Ω(n log n) lower bound that exists for this problem using the more typical algebraic RAM model.

Variable rate vector quantization scheme based on quadtree segmentation

Quadtree segmentation is a hierarchical spatial decomposition technique that partitions an image into variable-sized blocks based on a quadtree structure. Different coding strategies can be adopted to encode blocks of different perceptual significance. In this paper, an novel variable rate vector quantization scheme based on quadtree segmentation is proposed with which low-detailed blocks are adaptively encoded by the mean-subsampled vector quantization while high-detailed blocks by the traditional vector quantization scheme. The simulation results show that the proposed variable-block-sized vector quantization scheme achieves better subjective image quality with higher PSNR at low bit rates.

Performance of the progressive numerical method and its comparison with the modified spatial decomposition technique in solving large scattering problems

Techniques for solving electromagnetic scattering from electrically large structures are studied. Two techniques, the progressive numerical method (PNM) and the spatial decomposition technique (SDT), are applied to study the two-dimensional problems of perfectly conducting objects with TM and TE plane wave illuminations. Both PNM and SDT are based on the method of moments (MoM) and divide the objects into a number of small regions and hence reduce the matrix size and computer time. The PNM uses the concept of overlapping regions. Accurate solutions can be obtained for the TM case and good agreement with the full matrix MoM can be achieved for the TE case. It is found that the SDT needs modification, by not taking into account the virtual interface currents between the two adjacent regions and the iteration procedure is only useful for the TM case. It is also shown that, the modified SDT is a special case of PNM in which there is no overlapping region. Finally, it is found that the PNM is an efficient way to eliminate the internal resonances of scattering problems.

Spatial Decomposition and Reduced Current Fidelity for the Analysis of Tm Scattering From Large 2d oBjects

An investigation of the spatial decomposition technique (SDT) for the analysis of electrically large objects is presented. We apply SDT to two-dimensional perfectly conducting cylinders under TM excitation. Some of the convergence issues of the SDT as well as its application to finite arrays (finite frequency selective surfaces) are discussed. We also present a new method to obtain a reduction in the operation count of SDT, which is referred to as "reduced current fidelity (RCF)." With RCF the coupling between distant SDT subobjects is determined by use of a coarse grid on which currents are obtained from those on a dense grid. An averaging of the currents on distant objects reduces the operation count involved in determining the coupling to other subobjects. It is shown that for electrically large structures, RCF can provide significant reduction in operation count and good accuracy in the resulting structure currents.

Three-dimensional subharmonic waves during transition in the near-wake region of a cylinder

Detailed quantitative measurements of three-dimensional subharmonic waves have been obtained with a scanning laser anemometer in the near-wake region of a circular cylinder. A spatial decomposition technique has been used to separate the secondary three-dimensional flow from the primary Strouhal vortex flow. The appearance of subharmonic frequencies in the spectrum of the secondary instability field at Reynolds numbers of 300 and above provides experimental evidence consistent with the scenario of a period-doubling route to turbulence. The basic flow structure of the secondary field responsible for producing the subharmonics is shown to be pairs of counter-rotating streamwise vortices, which alternate the direction of rotation every Strouhal period.

Motion Planning for an Autonomous Compactor in a Waste Landfill Site

Abstract: This paper describes the algorithms and detailed path‐planning methodologies developed to enable autonomous spreading and compaction processes in a waste landfilling operation. A given landfill site is spatially decomposed into cells using a recursive spatial decomposition technique. The cell size is determined using a probabilistic model for waste generation. Variation in the amount of waste generated on a given day is handled by a further spatial decomposition of a cell into monominoes. The recursive spatial decomposition processes of a landfill site into cells and each cell into monominoes are accomplished by employing a variant of quadtree data structure. A three‐dimensional path plan for an autonomous compactor that considers a variable working gradient in the range of 3:1 to 2:1 is generated for a pair of monominoes. Automatic location of the working place for a compactor is accomplished by a recursive traversal of the quadtree structure. The algorithms have been implemented using a computer‐graphic functional interface standard with C‐program binding.

Quantitative measurements of three-dim ensional structures in the wake of a circular cylinder

The fine scale three-dimensional structures usually associated with streamwise vortices in the near wake of a circular cylinder have been studied at Reynolds numbers ranging from 170 to 2200. Spatially continuous velocity measurements along lines parallel to the cylinder axis were obtained with a scanning laser anemometer. To detect the streamwise vortices in the amplitude modulated velocity field, it was necessary to develop a spatial decomposition technique to split the total flow into a primary flow component and a secondary flow component. The primary flow is comprised of the mean flow and Strouhal vortices, while the secondary flow is the result of the three-dimensional streamwise vortices that are the essence of transition to turbulence. The three-dimensional flow amplitude increases in the primary vortex formation region, then saturates shortly after the maximum amplitude in the primary flow is reached. In the near-wake region the wavelength decreases approximately like Re−0.5, but increases with downstream distance. A discontinuous increase in wavelength occurs below Re = 300 suggesting a fundamental change in the character of the three-dimensional flow. At downstream distances (x/D = 10-20), the spanwise wavelength decreases from 1.42D to 1.03D as the Reynolds number increases from 300 to 1200.

A domain decomposition parallelization strategy for molecular dynamics simulations on distributed memory machines

An algorithm is described which allows large scale molecular dynamics simulations to be carried out on a distributed memory parallel computer. The method is a development of a spatial decomposition technique already described [S.Y. Liem, D. Brown and J.H.R. Clarke, Comput. Phys. Commun. 67 (1991) 261] but incorporates decomposition of 3-D space in all three dimensions with link-cell and neighbour table techniques for enhanced efficiency. It is shown that communications between processors can be minimized without incurring the penalties associated with redundant force calculations. The method has been tested on a Fujitsu AP1000 parallel machine with up to 512 processor elements.

Numerical analysis of electromagnetic scattering by electrically large objects using spatial decomposition technique

Apparent computational difficulties with the direct integral equation and method of moments have prompted an alternative numerical solution procedure based on the spatial decomposition technique. Using rigorous electromagnetic equivalence, the spatial decomposition technique virtually divides an electrically large object into a multiplicity of subzones. It permits the maximum size of the method of moments system matrix that needs to be inverted to be strictly limited, regardless of the electrical size of the large scattering object being modeled. The requirement on the computer resources is O(N), where N is the number of spatial subzones and each subzone is electrically small, spanning on the order of a few wavelengths. Numerical examples are reported along with comparative data and relative error estimation to expose the applicability and limitations of the spatial decomposition technique for the two-dimensional scattering study of electrically large conducting and dielectric objects.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>

Probe and component set-up planning for coordinate measuring machines

Abstract This article is concerned with that part of the inspection planning process that determines how the probe and component should be set up on a coordinate measuring machine (CmiM) for all the dimensional and geometrical constraints on a part that requires inspection. Each component set-up specifies the orientation in which the part should be held and all the probe set-ups connected with it. The probe set-up specifies the orientation in which the probe should be mounted, its configuration (probe body, styles and extension piece types and dimensions) and all the probing operations connected with it. The method used to determine this information is based principally on the analysis of probe approach directions (PADs) generated by the system for each surface to be inspected. This article describes how PADs can be generated for components defined by constructive solid geometry (CSG) using a method based on the spatial decomposition technique.