基于置换图的便笺存储器分配

在当今的嵌入式系统中,广泛地将片上存储器组织为软件管理的便笺存储器(SPM).Li等研究发现,对于很多嵌入式应用,其相干图中的数组生存期满足包含性.他们证明了满足生存期包含性的数组相干图为超完美图,并提出了一个基于超完美图的SPM分配算法.他们的算法在面向嵌入式应用的SPM分配上获得了当前最好的性能.本文进一步证明满足生存期包含性的数组相干图为置换图.置换图是超完美图的一个子类.在现有技术的情况下,置换图在判定及区间着色方面比超完美图有优势,如存在线性时间的识别算法,存在线性时间的最优区间着色算法.基于此理论结果,我们将Li等的算法在保留原算法逻辑的基础上,改进为基于置换图.实验表明,改进后的算法在很多不满足生存期包含性的相干图上仍能取得最优SPM分配,获得比基于超完美图的分配算法更好的分配结果.     

Relationships between vegetation and stomata, and between vegetation and pollen surface soil in Yunnan, Southwest China

Surface pollen and stomata of 61 samples collected in a study area ranging from tropical seasonal rainforest to oak forest(Quercus spinosa) in the Yulong Snow Mountain region in Yunnan,China,are used to distinguish vegetation communities.The results show that tropical seasonal rainforest(and mountain rainforest),south subtropical evergreen broad-leaved forest,and Quercus shrub are distinguished effectively from other vegetation types by analysis of surface pollen.The south subtropical evergreen broad-leaved forest,Pinus kesiya forest and evergreen broadleaf forest are distinguished effectively from other types of vegetation by pollen analysis.However,P.yunnanensis forest is not distinguished from other vegetation types,and P.armandii,P.densata forest and temperate deciduous conifer mixed forest are not distinguished.The over-representation of Pinus pollen is the main reason that these vegetation communities are not distinguished from each other.Conifer stomata analysis is an effective tool for identifying and distinguishing different types of coniferous forest,and this method performs well even with a small number of sampling points.