{"title":"Improved bisection eigenvalue method for band symmetric Toeplitz matrices","authors":"Y. Eidelman, I. Haimovici","doi":"10.1553/etna_vol58s316","DOIUrl":null,"url":null,"abstract":". We apply a general bisection eigenvalue algorithm, developed for Hermitian matrices with quasisep- arable representations, to the particular case of real band symmetric Toeplitz matrices. We show that every band symmetric Toeplitz matrix T q with bandwidth q admits the representation T q = A q + H q , where the eigendata of A q are obtained explicitly and the matrix H q has nonzero entries only in two diagonal blocks of size ( q − 1) × ( q − 1) . Based on this representation, one obtains an interlacing property of the eigenvalues of the matrix T q and the known eigenvalues of the matrix A q . This allows us to essentially improve the performance of the bisection eigenvalue algorithm. We also present an algorithm to compute the corresponding eigenvectors.","PeriodicalId":282695,"journal":{"name":"ETNA - Electronic Transactions on Numerical Analysis","volume":"33 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"1900-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"1","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"ETNA - Electronic Transactions on Numerical Analysis","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1553/etna_vol58s316","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 1
Abstract
. We apply a general bisection eigenvalue algorithm, developed for Hermitian matrices with quasisep- arable representations, to the particular case of real band symmetric Toeplitz matrices. We show that every band symmetric Toeplitz matrix T q with bandwidth q admits the representation T q = A q + H q , where the eigendata of A q are obtained explicitly and the matrix H q has nonzero entries only in two diagonal blocks of size ( q − 1) × ( q − 1) . Based on this representation, one obtains an interlacing property of the eigenvalues of the matrix T q and the known eigenvalues of the matrix A q . This allows us to essentially improve the performance of the bisection eigenvalue algorithm. We also present an algorithm to compute the corresponding eigenvectors.