We should have a multigroup fit function built in.
Links[1] https://openmx.ssri.psu.edu/users/mhunter