I am trying to build a multinomial logit model using GSEM as I want to incorporate random effects. My dataset is in such a way that some of the alternatives are not available for certain observations. Hence the dataset does not have a constant choice set but varying choice set.

Is there a way to incorporate this in GSEM?