Support for selecting basis set per atom type.

[yakutovicha]

Gaussian supports selecting different basis sets for different atom types. See how to do it in the following input example:

FeH

0 2
Fe   0.0    0.0   0.00
H    0.0    0.0   1.50

1  F

H  0
cc-pVTZ
****
Fe 0
lanl2dz
****

Fe 0
lanl2dz

It might be the way to go if we encounter SCF convergence problems in the future.

[eimrek]

some ideas on this:

• i think having multiple different basis sets is not a good idea to solve convergence problems. it's cumbersome to explain (e.g. in a paper) and i think the calculation would effectively have to be considered to have the accuracy of the weaker basis set. I think preferable is just to use the weaker basis set for the whole calculation. But perhaps in some cases this still makes sense.
• in my experience, diffuse functions are the main source of convergence problems (e.g. the "+" in pople basis sets) and using the solution of a slightly "less diffuse" basis as the initial guess usually helps
• Good to be mindful for what calculations diffuse basis sets are needed (e.g. a must for anions; but not so important for e.g. the geometry of neutral species).
• but of course this functionality in the plugin would be useful regardless :)