Criticality problems secondary energies

From these calculations, we notice the nonconvergence of the secondary line near 16,168 cm. This line position is only known within 0.5 cm. Usually, such a precision would be very satisfactory because the potential surface used is quite approximate. But if we would like to access the eigenvector, this slow convergence becomes a critical problem. If we compare this Lanczos calculation with an exact one, we conclude that the energy near 16,168 cm 1 converges slowly because it corresponds to two very close eigenenergies equal to 16,168.5 and 16,169.1 cm""1. 

---