A machine learning approach for predicting computational intensity and domain decomposition in parallel geoprocessing

ABSTRACT

High performance computing is required for fast geoprocessing of geospatial big data. Using spatial domains to represent computational intensity (CIT) and domain decomposition for parallelism are prominent strategies when designing parallel geoprocessing applications. Traditional domain decomposition is limited in evaluating the computational intensity, which often results in load imbalance and poor parallel performance. From the data science perspective, machine learning from Artificial Intelligence (AI) shows promise for better CIT evaluation. This paper proposes a machine learning approach for predicting computational intensity, followed by an optimized domain decomposition, which divides the spatial domain into balanced subdivisions based on the predicted CIT to achieve better parallel performance. The approach provides a reference framework on how various machine learning methods including feature selection and model training can be used in predicting computational intensity and optimizing parallel geoprocessing against different cases. Some comparative experiments between the approach and traditional methods were performed using the two cases, DEM generation from point clouds and spatial intersection on vector data. The results not only demonstrate the advantage of the approach, but also provide hints on how traditional GIS computation can be improved by the AI machine learning.     

空间信息网格中远程空间连接查询的优化

随着空间信息网格的建设,网格平台上管理的空间信息资源越来越丰富,这促进了空间信息网格中空间数据分布式查询的应用需求,而在分布式空间查询中,空间连接查询操作往往成为性能的瓶颈.根据空间信息的特点,通过利用网格计算资源来优化空间连接查询的执行.首先基于网格服务构建网格平台分布式空间数据查询软件结构,通过设计远程空间连接执行服务利用网格平台中的计算资源;根据空间信息的特点.采用基于Kd-Tree空间分区并行连接的方法提高远程空间数据连接操作执行效率,并给出了远程空间连接执行的查询代价模型;然后根据连接代价模型设计了远程空间连接查询执行计划优化生成算法;最后总结了本文工作并探讨了下一步研究方向.     

Efficient GOCE satellite gravity field recovery based on least-squares using QR decomposition

We develop and apply an efficient strategy for Earth gravity field recovery from satellite gravity gradiometry data. Our approach is based upon the Paige-Saunders iterative least-squares method using QR decomposition (LSQR). We modify the original algorithm for space-geodetic applications: firstly, we investigate how convergence can be accelerated by means of both subspace and block-diagonal preconditioning. The efficiency of the latter dominates if the design matrix exhibits block-dominant structure. Secondly, we address Tikhonov-Phillips regularization in general. Thirdly, we demonstrate an effective implementation of the algorithm in a high-performance computing environment. In this context, an important issue is to avoid the twofold computation of the design matrix in each iteration. The computational platform is a 64-processor shared-memory supercomputer. The runtime results prove the successful parallelization of the LSQR solver. The numerical examples are chosen in view of the forthcoming satellite mission GOCE (Gravity field and steady-state Ocean Circulation Explorer). The closed-loop scenario covers 1 month of simulated data with 5 s sampling. We focus exclusively on the analysis of radial components of satellite accelerations and gravity gradients. Our extensions to the basic algorithm enable the method to be competitive with well-established inversion strategies in satellite geodesy, such as conjugate gradient methods or the brute-force approach. In its current development stage, the LSQR method appears ready to deal with real-data applications.     

关于角平分线法建立线状缓冲区算法的改进

分析了用对角平分线法建立缓冲区的算法,提出了一种新的求角平分线和缓冲点的方法,实践证明其在算法实现方面是可行的,而且效果比较理想。     

内蒙古大青山地区寒武系与奥陶系之间的一个重要的层序界面

大青山地区寒武纪－奥陶纪地层分布零星，是由中生代逆冲推覆构造就位于此的一系列近东西向延展的构造岩片组成。下寒武统色麻沟组直接覆盖在太古界变质岩之上，缺失中元古界什那干群，寒武系与奥陶系之间呈假整合接触，标志着区内寒武纪与奥陶纪之间发生过一次抬升构造运动，其形成机制与古亚洲洋内的稳定陆快向华北板块拼贴事件有关，是早奥陶世末期华北板块整体抬升的序幕。这个层序界面的认识，对于恢复华北板块北部边缘构造演化历史具有重要意义。     

一种求气井二项式方程系数的新方法

针对开发早期气田缺乏产能试井资料的实际情况，由于无法用传统的压力降法计算产能方程，故而基于最优化算法，提出了一种利用稳定生产数据和井控储量求取气井产能方程系数的新方法。在只知道稳定生产数据情况下便可进行计算分析。该方法在气田进行了实际应用，并和实际资料以及传统试凑法计算的结果进行了比较，表明方法的计算结果较为可靠。     

生态地理建模中的多尺度问题

本文在分析生态地理建模内涵的基础上,讨论了生态地理建模中的尺度转换问题、跨尺度相互作用问题、空间尺度与时间尺度的关联问题和多尺度数据处理问题.由于生态地理问题的非线性、生态环境的异质性和随机事件,简单的线性尺度转换方法远不能满足生态地理建模的要求.为了从根本上解决生态地理建模中的时空尺度问题,除需要运用微分几何学和等级理论等经典方法外,还需要引入格点生成法和网格计算等现代理论和技术手段.     

Ion acoustic instability of HPT particles,FAC density,anomalous resistivity and parallel electric field in the auroral region

During the magnetic storm of 21st March 1990, the DE-1 spacecraft encountered the auroral region at high invariant latitude at altitudes ranging from a few thousand kilometers in the ionosphere to many earth radii in the magnetosphere. The magnetic field perturbations interpretable as field aligned current (FAC) layers and the electrostatic turbulence possibly due to electrostatic ion acoustic instability driven by these currents are shown. The critical drift velocity of Hot Plasma Torus (HPT) electrons and the growth rate of ion acoustic wave as a function of electron to ion temperature ratio (T _e/T_i) for low and high current densities and energy of HPT electrons are found out. The intense FAC destabilizes the ion acoustic wave and the resultant electrostatic turbulence creates an anomalous resistivity. The current driven resistivity produces parallel electric field and high power dissipation. The anomalous resistivityη, potential difference along the auroral field lines V_t|, intensity of electric field turbulenceE _t| and power produced per unit volumeP are computed. It is found that the change in westward magnetic perturbation increasesJ _t|, η, V_t|, E_t| andP. Hence HPT electrons are heated and accelerated due to power dissipation during magnetically active periods in the auroral region. Concerning, applications, such HPT electrons can be used in particle accelerators like electron ring accelerator, smokatron etc.     

一种在异构系统中实现负载平衡的方法

提出了在异构系统实现负载平衡的区域分解算法和实现负载平衡的计算方法，利用它的负反馈性质解决了异构系统处理机计算速度测量误差造成的负载测量不准问题，并对处理机速度变化，速度测量误差、处理机数量、网格点计算量的分布等因素的影响进行了计算，结果表明本方法具有很强的平衡负载能力和较强的适应性；根据计算结果提出了解决模式网格点计算量不易测量问题的解决方案，并用扩散方程和模拟物理过程进行试验，试验表明这种方法是可行的，平衡负载的效果十分显著。