Based on comprehensive analysis of typical outcrops, latest deep wells drilled and high resolution seismic profiles in the study area, we examined the geologic structure of the Kelameili range, and analyzed the struct...Based on comprehensive analysis of typical outcrops, latest deep wells drilled and high resolution seismic profiles in the study area, we examined the geologic structure of the Kelameili range, and analyzed the structural relationship between the Kelameili range and the Dajing depression, and discussed the tectonic-sedimentary framework in different periods of Carboniferous by using axial surface analysis and balanced section techniques. Understandings in three aspects are achieved:(1) The study area experienced five stages of compressional tectonic movements, the Early Carboniferous, the Late Carboniferous, the Middle-Late Permian, Late Cretaceous and Paleogene, and three stages of extensional tectonic movements, the middle-late Early Carboniferous, the middle-late Late Carboniferous and Early Permian. At the end of the Early Permian and the Mid-Late Cretaceous, the tectonic wedges moved southward respectively.(2) The Kelameili range and Dajing depression had the first basin-range coupling during the early Early Carboniferous, basin-range decoupling in the following middle-late Early Carboniferous to the Early Permian, then basin-range strong recoupling in the Middle Permian, and the basin-range coupling had been inherited in the subsequent Indosinian, Yanshanian and Himalayan movements.(3) During the early Early Carboniferous, the study area was a foreland basin where the Dishuiquan Formation source rock developed;in mid-late Early Carboniferous, a series of NW-and NWW-trending half-garben fault basins developed, where the Songkaersu Formation volcanic reservoir formed. In late Early Carboniferous, the study area entered into depression basin stage after rifting, and the Shuangjingzi Formation source rock developed;in the mid-late Late Carboniferous, Batamayineishan fault basin emerged, and the Upper-Carboniferous volcanic reservoir was formed, affected by the tectonic compression during late Carboniferous and Mid-Permian, the Batamayineishan Formation suffered extensive erosion, and only partially remains in the piedmont depression zone.展开更多
Spark下分布式深度信念网络(Distributed Deep Belief Network,DDBN)存在数据倾斜、缺乏细粒度数据置换、无法自动缓存重用度高的数据等问题,导致了DDBN计算复杂高、运行时效性低的缺陷.为了提高DDBN的时效性,提出一种Spark下DDBN数据...Spark下分布式深度信念网络(Distributed Deep Belief Network,DDBN)存在数据倾斜、缺乏细粒度数据置换、无法自动缓存重用度高的数据等问题,导致了DDBN计算复杂高、运行时效性低的缺陷.为了提高DDBN的时效性,提出一种Spark下DDBN数据并行加速策略,其中包含基于标签集的范围分区(Label Set based on Range Partition,LSRP)算法和基于权重的缓存替换(Cache Replacement based on Weight,CRW)算法.通过LSRP算法解决数据倾斜问题,采用CRW算法解决RDD(Resilient Distributed Datasets)重复利用以及缓存数据过多造成内存空间不足问题.结果表明:与传统DBN相比,DDBN训练速度提高约2.3倍,通过LSRP和CRW大幅提高了DDBN分布式并行度.展开更多
基金Supported by the National Natural Science Foundation of China(41702110)China National Science and Technology Major Project(2017ZX05001-001)National Key Research and Development Project(2017YFC0601405).
文摘Based on comprehensive analysis of typical outcrops, latest deep wells drilled and high resolution seismic profiles in the study area, we examined the geologic structure of the Kelameili range, and analyzed the structural relationship between the Kelameili range and the Dajing depression, and discussed the tectonic-sedimentary framework in different periods of Carboniferous by using axial surface analysis and balanced section techniques. Understandings in three aspects are achieved:(1) The study area experienced five stages of compressional tectonic movements, the Early Carboniferous, the Late Carboniferous, the Middle-Late Permian, Late Cretaceous and Paleogene, and three stages of extensional tectonic movements, the middle-late Early Carboniferous, the middle-late Late Carboniferous and Early Permian. At the end of the Early Permian and the Mid-Late Cretaceous, the tectonic wedges moved southward respectively.(2) The Kelameili range and Dajing depression had the first basin-range coupling during the early Early Carboniferous, basin-range decoupling in the following middle-late Early Carboniferous to the Early Permian, then basin-range strong recoupling in the Middle Permian, and the basin-range coupling had been inherited in the subsequent Indosinian, Yanshanian and Himalayan movements.(3) During the early Early Carboniferous, the study area was a foreland basin where the Dishuiquan Formation source rock developed;in mid-late Early Carboniferous, a series of NW-and NWW-trending half-garben fault basins developed, where the Songkaersu Formation volcanic reservoir formed. In late Early Carboniferous, the study area entered into depression basin stage after rifting, and the Shuangjingzi Formation source rock developed;in the mid-late Late Carboniferous, Batamayineishan fault basin emerged, and the Upper-Carboniferous volcanic reservoir was formed, affected by the tectonic compression during late Carboniferous and Mid-Permian, the Batamayineishan Formation suffered extensive erosion, and only partially remains in the piedmont depression zone.
文摘Spark下分布式深度信念网络(Distributed Deep Belief Network,DDBN)存在数据倾斜、缺乏细粒度数据置换、无法自动缓存重用度高的数据等问题,导致了DDBN计算复杂高、运行时效性低的缺陷.为了提高DDBN的时效性,提出一种Spark下DDBN数据并行加速策略,其中包含基于标签集的范围分区(Label Set based on Range Partition,LSRP)算法和基于权重的缓存替换(Cache Replacement based on Weight,CRW)算法.通过LSRP算法解决数据倾斜问题,采用CRW算法解决RDD(Resilient Distributed Datasets)重复利用以及缓存数据过多造成内存空间不足问题.结果表明:与传统DBN相比,DDBN训练速度提高约2.3倍,通过LSRP和CRW大幅提高了DDBN分布式并行度.
