Chapter Contents |
Previous |
Next |
The INBREED Procedure |
The memory requirement, s, for PROC INBREED depends on the processing mode. For analysis within nonoverlapping generations, the memory requirement in bytes is
where n is the maximum number of individuals in a generation and f is the maximum number of families in a generation. For a general population that is not divided into nonoverlapping generations, the minimum memory requirement in bytes is
where n is the number of individuals in the population.
In the case of a general population, PROC INBREED can use memory beyond the minimum requirement to reduce computation time. PROC INBREED uses several techniques to reduce memory and computational requirements. The covariance coefficients are stored as scaled binary integers, 16 bits in length. Because the covariance formulas use only addition and division by 2, no loss of precision occurs by using the scaled covariance coefficients until after the thirteenth generation. After the thirteenth generation, the loss of precision is in the fourth decimal place.
To reduce memory and time requirements for multiparous populations, PROC INBREED examines the input to see how many families are present, and then only one coefficient per family or between families is computed and perhaps stored. In a general population, a covariance matrix is computed if necessary and if memory permits. If there is insufficient memory to hold the entire covariance matrix, only as much of the upper-left triangle of the matrix as fits into the available memory is stored. Those coefficients outside the upper-left triangular matrix are computed as needed.
Chapter Contents |
Previous |
Next |
Top |
Copyright © 1999 by SAS Institute Inc., Cary, NC, USA. All rights reserved.