简介:theAlternatingSegmentCrank-Nicolsonschemeforone-dimensionaldiffusionequationhasbeendevelopedin[1],andtheAlternatingBlockCrank-Nicolsonmethodfortwo-dimensionalproblemin[2].Themethodshavetheadvantagesofparallelcomputing,stabilityandgoodaccuracy.Inthispaperforthetwo-dimensionaldiffusionequation,thenetregionisdividedintobands,aspecialkindofblock.ThismethodiscalledthealternatingBandCrank-Nicolsonmethod.
简介:Modelinglog-mortalityratesonO-UtypeprocessesandforecastinglifeexpectanciesareexploredusingU.S.data.IntheclassicLee-Cartermodelofmortality,thetimetrendandtheage-specificpatternofmortalityoveragegrouparelinear,thisisnotthefeatureofmortalitmodel.Toavoidthisdisadvantage,O-Utypeprocesseswillbeusedtomodelthelog-mortalityinthispaper.Infact,thismodelisanAR(1)process,butwithanonlineartimedriftterm.BasedonthemortalitydataofAmericafromHumanMortalitydatabase(HMD),mortalityprojectionconsistentlyindicatesapreferenceformortalitywithO-UtypeprocessesoverthosewiththeclassicalLee-Cartermodel.Bymeansofthismodel,thelowboundsofmortalityratesateveryagearegiven.Therefore,lengtheningofmaximumlifeexpectanciesspanisestimatedinthispaper.
简介:结合偏最小二乘法和支持向量机的优缺点,提出基于偏最小二乘支持向量机的天然气消费量预测模型。首先,利用偏最小二乘法确定影响天然气消费量的新综合变量,建立以新综合变量为输入,天然气消费量为输出的支持向量机模型,对天然气消费量进行了预测;然后,与多元回归、偏最小二乘回归、普通支持向量机做误差检验比较,验证该方法的可行性与正确性。结果表明,此天然气消费量预测模型具有较高的精确度和应用价值。