Journals / Eskişehir Osmangazi Üniversitesi Mühendislik ve Mimarlık Fakültesi Dergisi / 1996 / Cilt: 9 Sayı: 1
A Tutorial On The Singular Value Decomposition
- Pages
- 1–12
- DOI
- —
Abstract
An m y, n real matm A can be factored as ITWV , where V and V areorthonormal, and W is upper left diagonal. Thîs factorization is called Singular VaîueDecomposilion (SVD). The matrlces U, W, and V are usefül in characterizing ıhe malrix A. inthis manuscript geometric characterizations are emphasized. Geometric characîerizations areanaîyzed in terms ofsubspaces, matrix scaling, cmd norms. We also presenî a numerical viewpointfor SVD m orcfer to keep the maîerial setf-contained. in the last section we îreat a special problemwhere action ofîhe matrix A is restncted to a gıven subspace.
Özet
An m y, n real matm A can be factored as ITWV , where V and V are