Hello, this is not a support request, but a clarification that might help others.
I'm optimizing the geometry of excited states.
After multiple failures, I finally found out the weird way of labelling the excited states (roots).
In this example I am using symmetry. The first 5 roots have the following symmetry:
No. Spin Occ Vir Irrep E(Occ) E(Vir) E(Diff)
1 1 197 198 b3u -0.208 -0.039 4.589
2 1 197 199 b2u -0.208 -0.037 4.648
3 1 196 198 b1g -0.273 -0.039 6.365
4 1 195 198 b3u -0.275 -0.039 6.412
5 1 196 199 ag -0.273 -0.037 6.425
So if I want to optimize state 1, I have to request the first root with symmetry b3u:
tddft
rpa
nroots 5
algorithm 1
notriplet
target 1
targetsym b3u
civecs
grad
root 1
end
end
And that's straightforward: target 1 and root 1 in the grad block.
But now I want to optimize state 2. It is the first root with symmetry b2u, so I have to ask for target 1 and targetsym b2u. This is a bit counterintuitive, and not well explained in the documentation, but makes sense. Now, in the grad block I have to ask for root 2, not 1 ! This is very confusing and should be made clear in the documentation. Here's how to request the optimization of root 2 in this particular case:
tddft
rpa
nroots 5
algorithm 1
notriplet
target 1
targetsym b2u
civecs
grad
root 2
end
end
|