You can change the algorithm options for several reasons. For example, you can provide starting estimates so that the algorithm converges to a solution. You can also provide estimates and not iterate to evaluate the results calculated using the entered estimates.

- Starting variance value
- To specify starting values different from the Minimum Norm Unbiased Estimates (MINQUE), enter a value for every random term. The value for Error must be greater than 0. At least one of the values for the model terms must be greater than 0. None of the values can be negative.
- Maximum number of iterations
- If the algorithm did not converge, you can increase the maximum number of iterations to try to achieve convergence. Enter 0 to use the values in Starting variance value as the variance components in the analysis. For example, enter 0 when you want to specify the components to evaluate the results calculated using the entered estimates.
- Convergence tolerance for variance estimates
- Usually, the default value works well. The smaller the value, the more stringent the criterion for convergence is. The larger the value, the less stringent the criterion for convergence is.
- Convergence tolerance for likelihood function
- Usually, the default value works well. The smaller the value, the more stringent the criterion for convergence is. The larger the value, the less stringent the criterion for convergence is.