Number of Basis Functions is Mismatched/Inconsistent!

1

Hello,

I have been reproducing some earlier calculations I ran in Gaussian with Psi4 using the same basis set and functional (DZVP/wb97x-d) and geometry coordinates for single point frequency calculations. While the replication of energetics is spot on (great work), the number of basis functions appears to differ. I noticed this especially when I was looking at the time it took for some of my larger molecules to finish and found them taken 90 times longer than when run in Gaussian. Noting of course that the analytical hessian is not available for this theory, this was still significantly slower than my tests on smaller molecules had found (example alanine versus a larger molecule with 58 atoms). Investigating made me realize the basis function number was on the order of ~3400 for a 70 atom molecule when molecules with the same elements and of similar size were in the 500s of basis functions. This would definitely change the speed of the calculation.

Is there any obvious explanation for this?

I’d be happy to provide the molecular input and calculation files for each program for investigation if this does not have an obvious answer.

Thank you.

2

The most likely explanation is that you’re confusing the number of primary basis functions with the number of auxiliary basis functions in the Psi4 output file. Here’s an example from an old file of mine.

  ==> Primary Basis <==

  Basis Set: AUG-CC-PVDZ
    Blend: AUG-CC-PVDZ
    Number of shells: 14
    Number of basis functions: 32
    Number of Cartesian functions: 34
    Spherical Harmonics?: true
    Max angular momentum: 2

...

   => Auxiliary Basis Set <=

  Basis Set: (AUG-CC-PVDZ AUX)
    Blend: AUG-CC-PVDZ-JKFIT
    Number of shells: 40
    Number of basis functions: 118
    Number of Cartesian functions: 136
    Spherical Harmonics?: true
    Max angular momentum: 3

The number of basis functions has increased by roughly a factor of 4 between the two basis sets. This is entirely normal.

If you’re seeing that 3400 number under the primary basis, then yes, a concrete example would be helpful.

3

That was exactly it. Thank you for catching my mistake. The primary basis function number matches with expectations. Phew!

Thank you,
Jessica