Supplementary Appendix of “Simultaneous Covariance Inference for Multimodal Integrative Analysis”
Yin Xia,Lexin Li,Samuel N. Lockhart,William J. Jagust
2019-01-01
Abstract:We further consider the setting when the variables within Xk are correlated. Note that, by construction, the variables within Y k are always correlated. Specifically, we adopt a similar simulation setup as in Section 4.1 of the paper, and add a covariance structure ΣX for the entries of Xk. We set ΣX = c Cov(Xk) = (σX,i,j), where c = Var{N(0, 1)} = 1 for Distribution 1 and c = Var{t(10)} = 1.25 for Distribution 2, and σX,i,i = 1, σX,i,j = 0.5 for 5(k− 1) + 1 ≤ i, j ≤ 5k, with k = 1, . . . , [p/5], and σX,i,j = 0 otherwise. We have chosen this block covariance structure mostly to simplify the computation in the data generation step, since it allows us to efficiently compute the true covariance structure in the simulations. Then we generate n copies of the second modality, {Y k}k=1 in the same way as the paper. In this case, we have Cov(Xk,Y k) = cΣXΣ1,2, with the following three covariance structures similar as those in the paper: