MSE vs topology for GMRFs

Closed form expressions for the MSE for stylized topologies such as complete graph, star and cycle seem very doable.. and so are the error exponents