Speaker
Description
We investigate the role of non-local electronic correlations
in two-dimensional Hubbard models by combining and benchmarking complementary many-body
approaches. In particular, we assess the performance of the two-particle self-consistent (TPSC)
method and its extension with dynamical mean-field theory (DMFT), using cluster DMFT as a ref-
erence. To this end, we employ both continuous-time quantum Monte Carlo and a tensor-network-
based fork tensor product state (FTPS) solver, the latter enabling direct access to real-frequency
quantities and offering a potential route to cluster calculations without analytic continuation. We
consider square-lattice systems with and without next-nearest-neighbour hopping, as well as the
geometrically frustrated triangular lattice, thereby spanning regimes with varying degrees of nest-
ing and frustration. For the square lattice, we find that the combined DMFT+TPSC approach
reproduces cluster DMFT results with high accuracy at significantly reduced computational cost,
while pure TPSC captures qualitative trends but exhibits deficiencies at low frequencies, particu-
larly in the presence of strong nesting. On the triangular lattice, TPSC and DMFT+TPSC yield
reasonable qualitative agreement with cluster results for non-local self-energies, though quantitative
deviations persist. Our results demonstrate that DMFT+TPSC provides an efficient framework for
incorporating short-range correlations beyond single-site DMFT, and highlight both the strengths
and limitations of TPSC-based approaches across different lattice geometries.