DSpace

King Saud University Repository >
King Saud University >
ACADEMIC PUBLISHING & PRESS >
Journal of the King Saud University - Agricultural Sciences >

Please use this identifier to cite or link to this item: http://hdl.handle.net/123456789/682

Title: Application of F-G diagonalization algorithm to restricted maximum likelihood estimation of variance components
Authors: Ali, Ahmed Kamal Ahmed
Keywords: Restricted maximum likelihood
Algorithm
Issue Date: 1994
Publisher: King Saud University
Citation: Journal of King Saud University, Agricultural Sciences: 6 (2); 221-237
Abstract: The F-G algorithm ofFlury and Gautschi can be used to find an orthogonal matrix B such that: k(B)=T{det[diag(B'CiB)]/det(Ci]}"j is minimum, where Cj is (Z'MZ + ocAI) and nl , nk' are i=l positive weights. The orthogonal matrix B can be interpreted as the matrix which brings matrices Cl , Cksimultaneously as close to diagonality as possible. To reduce the number of operations required by F-G algorithm, Clarkson used a modified algorithm (MF-G) to find an orthogonal matrix B such that B'CjB is nearly diagonal. Both F-G and MF-G algorithm were applied to three sets of mixed model coefficient matrices in animal breeding cases. Close estimate to the exact REML solutions were obtained for traits with low heritability (large oc). One can use equal or unequal weights nI' ,nk to achieve convergence for both algorithms.
URI: http://hdl.handle.net/123456789/682
Appears in Collections:Journal of the King Saud University - Agricultural Sciences

Files in This Item:

File SizeFormat
Application of F-G diagonalization algorithm to restricted maximum likelihood estimation of variance components.pdf311.14 kBAdobe PDFView/Open

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.

 

DSpace Software Copyright © 2002-2007 MIT and Hewlett-Packard - Feedback