摘要
Theoptimallyweightedleastsquaresestimateandthelinearminimumvarianceestimatearetwoofthemostpopularestimationmethodsforalinearmodel.Inthispaper,theauthorsmakeacomprehensivediscussionabouttherelationshipbetweenthetwoestimates.Firstly,theauthorsconsidertheclassicallinearmodelinwhichthecoefficientmatrixofthelinearmodelisdeterministic,andthenecessaryandsufficientconditionforequivalenceofthetwoestimatesisderived.Moreover,undercertainconditionsonvariancematrixinvertibility,thetwoestimatescanbeidenticalprovidedthattheyusethesameaprioriinformationoftheparameterbeingestimated.Secondly,theauthorsconsiderthelinearmodelwithrandomcoefficientmatrixwhichiscalledtheextendedlinearmodel;undercertainconditionsonvariancematrixinvertibility,itisprovedthattheformeroutperformsthelatterwhenusingthesameaprioriinformationoftheparameter.
出版日期
2009年01月11日(中国期刊网平台首次上网日期,不代表论文的发表时间)