This case arises when the “traits” are different measures of the same trait, some with more measurement error than others, or when the variance in the trait due to non-genetic factors differs. Here the Ω matrix has only T rather than T(T+1)/2 unique parameters to be estimated.
