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.     

Discovery of 13 New Variable Stars in the Field of the Open Cluster NGC 2168 （M35）

A wide-field time-series CCD photometric survey of variable stars in the field of the open cluster NGC 2168 was carried out using the BATC Schmidt telescope. In total 13 new variable stars are discovered with three W UMa systems, one EA type and two EB type eclipsing binaries (one of them could be a W UMa system), and seven pulsating stars including three candidates of δScuti stars.     

Multi-wavelength Diagnostics of Bombardment by Non-thermal Particles in Solar Flares

Chromospheric lines, including Ha, Lyα, Lyβand CaⅡK, CaⅡ8542, are systemically and quantitatively investigated with respect to the non-thermal excitation and ionization due to particle beam bombardment for a series of solar semi-empirical atmospheric models. As a result we propose to use the contrast in the integrated intensity of hydrogen lines to estimate the total energy flux of the bombarding beam during the solar flare impulsive phase. Partial frequency redistribution is considered in the Lyαline calculation and a smaller intensity enhancement in the Ha line-centers is found than in the previous results of Fang et al.     

Mobile Agent系统结构及其关键技术

Mobile Agent技术是一门新兴的技术,具有移动性、智能性和异步计算的特点,已成为计算机网络和分布式系统最具活力的发展方向.现对Mobile Agent的概念、系统结构及其关键技术进行了分析和探讨.     

A numerical model for Type II supernovae — The collapse stage

We use the following numerical model for the collapse stage of a Type II supernova of 15 M_⊙. Our electron capture rate includes the effects of the inverse reaction and the neutron-proton mass difference. This decreases the electron density at the collapse stage and led to rather large values of the maximum inward velocity and of the corresponding mass (Umax = 3.06 × 10⁹cm/s, Mmax=0.76 M_⊙). These larger values are more favourable for the propagation of shock after the rebounce and the triggering-off of a Type-II supernova explosion. For neutrino transport, we use a leakage model and an equilibrium diffusion model, respectively, for the thin and thick stages and a grey atmosphere model to assess the effect of neutrino precipitation on the collapse. We found this effect to be small, the energy precipitation to be not more than 10^?5 the neutrino energy loss and the momentum precipitation not more than 10^?6 the gravitational acceleration.     

Observational evidence for solar coronal decimetric radio pulsations with very short periods

Using the decimetric (700–1500 MHz) radio spectrometer and the synchronous observational system with high temporal resolution at four frequencies (1420, 2130, 2840 and 4260 MHz) of Yunnan Observatory, two rare events were observed on 2001 June 24 and 1990 July 30. The former was a small radio burst exhibiting pulsations with short periods (about 29, 40 and 100 ms) in the impulsive phase. The latter was a large radio burst, which at 2840 MHz produced radio pulsations with period of about 30 ms. This paper focuses on pulsations with very short periods in the range of 29–40 ms. The mechanism of generation of such pulsations may be modulation of radio radiation by the periodic trains of whistler packets originating in unstable regions of the corona. Alternatively, these pulsations can be attributed to wave-wave non-linear interactions of electrostatic upper hybrid waves driven by beams of precipitating electrons in flaring loops.     

利用自回归模型检测地球极移重力效应

同时采用4个台站的国际超导重力仪长期连续重力观测资料和国际地球自转服务中心提供的同步地球自转参数，研究了极移引起的地球重力场变化特征。利用自回归模型估计了各序列的功率谱密度和积谱密度，结果表明极移导致的重力效应的主要能量集中在Chandler摆动和周年项附近，叠积后实际重力观测与极移重力信号理论值之间的差异分别为0．4％和3．9％，说明超导重力仪可有效监测极移导致的重力变化。     

桐柏─大别山北麓金矿带地质物化探找矿模型

从桐柏─大别山成矿带区域地质、地球物理、地球化学特征的研究入手，分析与金成矿有关的地层、构造、岩浆岩等控矿因素的地球物理场及地球化学特征；建立区域金矿找矿模型。