Setting DFT integration grid based on atom identity

I am trying to reproduce the properties of several molecules that were originally computed with a ‘superfine’ numerical integration grid in Gaussian16. This keyword employs a grid composed of (i) 974 spherical points and 175 radial points for first-row atoms and (ii) 974 spherical points and 250 radial points for subsequent-row atoms. In this case, I would like to use a 175,974 grid for carbon and fluorine atoms and a 250,974 grid for iodine atoms.

Is it possible to set the DFT integration grid via ‘dft_spherical_points’ and ‘dft_radial_points’ (or any other keyword) on a per atom basis?

It is not possible in Psi4.

I can add this as a feature request, but I’m not optimistic for this getting developer attention in the near future.

1 Like

I should point out that this is simply to reproduce the data generated in Gaussian16 before moving to PSI4 for subsequent production level computations.

Thanks for the response and help!

This is actually a good feature to have for heavy atom calculations without ECPs.


Agreed. Maybe the feature request will get the attention it needs to actually make it into a future build.

As far as feature requests go, I would expect this would be a simple one. Psi is just short on people to work on the DFT code.

Normally, I’d offer to take this, but I’m busy with some large code cleanup proejcts at the moment.

1 Like