Great thanks!
Sorry, I don’t have the probability language to express my model, but i think i have a “truncated mixture model”, where the number of components is bounded at 6 (5 normals and one uniform).
I read through the link you posted, and to be honest, i don’t understand how to extend it to my problem.
I would like to have a sparse w for each observation, where, for example, if it is likely there are only 2 components need to explain the data, then w would likely be = [1 0 0 0 0 0] or [0 1 0 0 0 0].
However, in that example, there is only a single w. Also, is there a way to force w to be sparse?
I’ve drawn a little picture to give some more context on the data:
