QChASM/AaronTools.py

int grids ORCA 5

Closed this issue · 2 comments

joegair commented

The theory tools for defining grids are not fully compatible with ORCA 5.0.3

when grids are set to'ultrafine', AaronTools writes ORCA inputs with:

%method
    AngularGrid  Lebedev590
    IntAcc            4.0
end

ORCA 5 uses 6 instead of Lebedev590.

Likewise 'superfinegrid' writes ! Grid7 FinalGrid7 which is not recognized in ORCA 5.0.3. The largest default grid in ORCA 5.0.3 is ! DefGrid3

ajs99778 commented

It looks like they updated the AngularGrid keyword in ORCA 5 to use numbers instead of LebedevX. They also clarified the formula for the number of radial points in the manual. I though the formula in the ORCA 4 manual used the natural log, but it actually uses log10. IntAcc should be much higher across the board. I've fixed these issues.

ORCA can't quite have angular grids as dense as superfinegrid. Gaussian uses a Lebedev-974 grid for superfine, but ORCA's densest Lebedev grid only has 770 points. Of course, both programs prune the grids. There's other reasons why Gaussian and ORCA grids aren't the same aside from pruning, though. I chose 17 for IntAcc as a middle ground for the radial component of superfinegrid.

joegair commented