Efficient Representation of Minimally Entangled Typical Thermal States in two dimensions via Projected Entangled Pair States

The Minimally Entangled Typical Thermal States (METTS) are an ensemble of pure states, equivalent to the Gibbs thermal state, that can be efficiently represented by tensor networks. In this article, we use the Projected Entangled Pair States (PEPS) ansatz as to represent METTS on a two-dimensional (2D) lattice. While Matrix Product States (MPS) are less efficient for 2D systems due to their complexity growing exponentially with the lattice size, PEPS provide a more tractable approach. To substantiate the prowess of PEPS in modeling METTS (dubbed as PEPS-METTS), we benchmark it against the purification method for the 2D quantum Ising model at its critical temperature. Our analysis reveals that PEPS-METTS achieves accurate long-range correlations with significantly lower bond dimensions. We further corroborate this finding in the 2D Fermi Hubbard model at half-filling. At a technical level, we introduce an efficient zipper method to obtain PEPS boundary matrix product states needed to compute expectation values. The imaginary time evolution is performed with the neighbourhood tensor update.