Fine Grained Tensor Network Methods.

摘要: We develop a strategy for tensor network algorithms that allows to deal very efficiently with lattices of high connectivity. The basic idea is fine grain the physical degrees freedom, i.e., decompose them into more fundamental units which, after suitable coarse graining, provide original ones. Thanks this procedure, lattice connectivity transformed by an isometry simpler structure, which easier simulate via usual methods. In particular enables use standard schemes contract infinite 2D networks-such as corner transfer matrix renormalization schemes-which are involved on complex structures. prove validity our approach numerically computing ground-state properties ferromagnetic spin-1 transverse-field Ising model triangular and 3D stacked lattice, well hardcore softcore Bose-Hubbard models lattice. Our results benchmarked against those obtained other techniques, such perturbative continuous unitary transformations graph projected entangled pair states, showing excellent agreement also improved performance in several regimes.