Hi Edo,
I decided to look at the symmetry aspect of the calculation. The conformation with the seemingly small CCSD and CCSD(T) correlation energy is in CS symmetry. In the new code it automatically checks this symmetry with autosymm. If I use SYMMETRY CS then as I said before I get the same answer as with out this directive present:
CCSD correlation energy / hartree = -0.720195222066366
Just by chance I decided to rerun the calculation with SYMMETRY C1 as a directice and surprisingly I am now getting a correlation energy that matches up with the other 3 calculations. The job hasn't finished but you can see from the iterations that the correlation energy is much larger:
MICROCYCLE DIIS UPDATE: 480 5
481 0.0004320511988 -1.1118181107609 130.3 130.4 41.5
482 0.0008659036337 -1.1118394392735 129.3 129.8 41.7
483 0.0057894855964 -1.1118555457920 129.3 129.5 41.9
484 0.0510192426928 -1.1118676172230 130.0 130.3 42.0
485 0.4662714641661 -1.1118759942074 130.3 130.7 43.0
MICROCYCLE DIIS UPDATE: 485 5
486 0.0004166784843 -1.1118190111607 130.7 130.8 42.4
487 0.0004444610069 -1.1118401224515 130.6 132.2 41.5
488 0.0008756038419 -1.1118560832914 128.2 130.3 41.6
489 0.0056253402134 -1.1118681776394 129.4 131.0 42.0
|