Page 299 - Linear Models for the Prediction of Animal Breeding Values 3rd Edition

(Continued)
Animal Code Sire or progeny Dam or mate Number of unknown parents
4 1 1 10 1
4 2 7 5 0
5 1 3 2 0
5 2 7 4 0
6 1 1 2 0
6 2 8 3 0
7 1 4 5 0
8 1 3 6 0
9 2 1 10 2
9 2 2 10 2
9 2 3 10 2
10 2 1 9 2
10 2 2 9 2
10 2 3 9 2
10 2 4 1 1

The arrangement of yield data is the same as in Section 17.4.1 in the animal model
analysis without groups.

ITERATIVE STAGE

SOLVING FOR FIXED EFFECTS. This is exactly as described for the animal model with-
out groups in Section 17.4.1, with yield records adjusted for other effects in the model
and solutions for fixed effects computed.

SOLVING FOR ANIMAL SOLUTIONS. Solutions for animals are computed one at a time as
both pedigree and data file sorted by animals are read, as described for the animal
model without groups. Therefore, only the differences in terms of the way diagonals
and ARHSs are accumulated are outlined.
For the kth animal in the pedigree file, calculate:
w = a(4/(2 + no. of unknown parents))
k
For the type 1 record in the pedigree file for the kth animal:
ARHS = ARHS + (uˆ + uˆ )0.5w
k k s d k
DIAG = DIAG + w
k k k
For the type 2 record in the pedigree file for the kth animal:
ARHS = ARHS + (uˆ − 0.5uˆ )0.5w
k k o m k
Accumulation of ARHSs from the data file is as specified in Section 17.4.1 in the
model without groups.
The solution for the kth animal is computed as ARHS /DIAG when all records
k k
for the animal in the pedigree and data file have been read. The solutions in the first

Solving Linear Equations 283