In many commonly used models for multivariate traits, the likelihood is specified as a mixture of nested sums of products over the unobserved genotypes of all the family members, in which the familial covariance matrices vary in size and structure for different families, and their sizes can be...