The accuracy of a simple crossing-symmetric approximation for the fully reducible vertex is tested by comparisons of the spin, charge, and pairing correlations with those obtained by Monte Carlo simulations of the two-dimensional Hubbard model. The approximation under study consists in assuming that for parallel spins the fully reducible vertex vanishes, while for anti-parallel spins it is equal to the T-matrix. Up to quarter-filling, accuracies better than 10% are obtained.

ER -