Hi all,
I am running frequency calculations using CBS extrapolation like in the attached example.
x.dat (1.2 KB)
Because of the large basis sets, the latest version I can use is Psi4 1.5 (Git: Rev {HEAD} e9f4d6d). The calculations go through displacements 6 times. During the first four repetitions, SCF Algorithm Type is mostly DF, and the computations are fast. However, in the fifth repetition, SCF Algorithm Type is DISK_DF and it is really slow. This behavior seems to be independent of the available RAM or molecule size.
Is there any way to avoid this slowdown?
Is there any way to check if this behavior occurs in the newer versions?