Projects / Programmes
Optimization of algorithms for evaluation of (restricted) maximum likelihood function for large systems with application in animal science
Code |
Science |
Field |
Subfield |
4.02.01 |
Biotechnical sciences |
Animal production |
Genetics and selection |
Code |
Science |
Field |
B110 |
Biomedical sciences |
Bioinformatics, medical informatics, biomathematics biometrics |
B400 |
Biomedical sciences |
Zootechny, animal husbandry, breeding |
livestock production, selection, statistics, mixed model methodology, prediction of breeding value, estimation of dispersion parameters, variance-covariance components, litter size, fattening traits, test-day records, computer science, programming, data structure, finite universe dictionary, sparce matrices
Researchers (11)
Organisations (1)
Abstract
A novel data structure was introduced to store sparse coefficient matrices of mixed model equations. The goal was to minimiye the space for matrix presentation and the time of an access to the matrix element. The siye of our data structure is close to the information theoretic minimum (it differs in the second order term) and permits constant access to the matrix elements and a constant amortiyed time with a high probabilitz for the insertion of a new element in the matrix and a deletion of an existing element.
The algorithm was tested in a setup of mixed model equations that is used for genetic evaluation of breeding animals and estimation of dispersion parameters. As a bz-product we get estimates od dispersion parameters for test-daz records in goats and sheep, fertilitz and fattening traits in pigs.
During our work we also had to do some basic research in datastructures and algorithms. From this research we developed, beside thealgorithms used in prediction of breeding values a novel datadata structure for a succinct representation of Huffman trees. Thisdata structure is space optimal and also permits optimal decoding ofHuffman encoded text. The techniques developed in prediction of breeding values proved to be useful also in various algorithmsused in data communication and in algorithms for the maintenance ofprioritz queues