How to compute the weight matrix for WLS estimation of a multi-group ordinal CFA model

I am attempting to perform a two-group confirmatory factor analysis (CFA) of one continuous factor on six ordinal predictors in OpenMx (R) using (robust) Weighted Least Squares (WLS) estimation. While I am completely new to OpenMx (I only used sem and lavaan ) I think I have most things now: I have a way of estimating polychoric correlations/covariances and estimating the latent means in one group and a good idea on how to implement this model in OpenMx.

I just have one problem, I need a weight matrix in WLS estimation. This should be an asymptotic covariance matrix of the elements of the observed correlations/covariances.

What is this exactly and how can I compute such a matrix?

-