In general, is there any theory of MPS states for classical statistical lattice models at finite temperature?
For example, is there any theorem (in the spirit of Hastings, '07) that says that the transfer matrix of, say, the 2D classical Ising model, can be described efficiently using MPS?