“…Exact diagonalization methods [18,[26][27][28]34] suffer from the exponentially growing size of the Hilbert space, which is especially a problem in two or more dimensions, while QMC techniques [35] suffer from the sign problem for frustrated lattices, and projected entangled pair states (PEPSs) [36,37] for this model are complex and computationally costly, even though, in principle, PEPSs have good computational scaling properties in two dimensions. More recently some numerical methods have been developed that are especially useful for frustrated systems and applied to the THM, for example, the large-scale parallel tempering Monte Carlo [38] and some tensor networks methods including entangled-plaquette states [39] and the multiscale entanglement renormalization ansatz (MERA) [40].…”