Uso da iteração nos dados para resolução de equações de modelo misto
Loading...
Files
Date
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Revista Brasileira de Zootecnia
Abstract
Valores genéticos foram preditos usando-se a iteração nos dados em um modelo animal reduzido. O peso corporal, aos 35 dias de idade, de duas linhas de aves de corte foi medido em duas gerações, cujo arquivo de dados consistia de 1044 pais (população-base), 829 progênies-pais e 9039 progênies-não-pais. O número de iterações e o tempo de processamento, para obtenção das soluções das equações de modelo misto via iteração nos dados, foram avaliados por sete critérios de convergência (101 a 10-5 ). Seis iterações foram necessárias, com 9 s de tempo, e 158 iterações, com 5 min e 38 s, para se obterem as soluções para os critérios de convergência de 101 e 10-5, respectivamente. As correlações entre os valores genéticos preditos foram perfeitas (r=1,00), e não houve diferenças entre as tendências genéticas estimadas pelos critérios de convergência de 10-1 a 10-5. Com base nestes resultados, conclui-se que a iteração nos dados pode ser eficientemente usada em microcomputadores, na avaliação genética animal, sem grande demanda de tempo e memória computacional.
Breeding values were predicted by iterating on data using reduced animal model. Body weight at 35 days of age from two lines of meat-type chickens was measured in two generations and data file was consisted on 1044 parents (base population), 829 progeny parents and 9039 progeny non-parents. The number of iterative rounds and processing time for mixed model equations solutions via iterating on data were evaluated using seven convergence criteria (101 to 10-5). A total of six rounds of iteration and 9 s of time and 158 rounds and 5 min and 38 s were required to reach the solutions for 101 and 10-5 convergence criteria, respectively. The correlation between predicted breeding values were perfect (r=1,00) and there were no significant differences between estimated genetic trends using 10-1 to 10-5 as a convergence criteria. Based on these results, it was concluded that iterating on data could be efficiently used for animal genetic evaluation in microcomputer without great computational requirements.
Breeding values were predicted by iterating on data using reduced animal model. Body weight at 35 days of age from two lines of meat-type chickens was measured in two generations and data file was consisted on 1044 parents (base population), 829 progeny parents and 9039 progeny non-parents. The number of iterative rounds and processing time for mixed model equations solutions via iterating on data were evaluated using seven convergence criteria (101 to 10-5). A total of six rounds of iteration and 9 s of time and 158 rounds and 5 min and 38 s were required to reach the solutions for 101 and 10-5 convergence criteria, respectively. The correlation between predicted breeding values were perfect (r=1,00) and there were no significant differences between estimated genetic trends using 10-1 to 10-5 as a convergence criteria. Based on these results, it was concluded that iterating on data could be efficiently used for animal genetic evaluation in microcomputer without great computational requirements.
