| 11:19:20 AM PST - Mon, Feb 11th 2013 |
|
Hi,
We're having an issue with TDDFT calculations on a series of related molecules with D2h symmetry. For the smallest molecule (26 atoms, 18 of which non-hydrogens) all is fine and we find the same excitations when we run the calculation in C1 and when we run the calculation in D2h. For the larger molecules (e.g. a molecule of 66 atoms, 42 of which non-hydrogen) there is however an issue. The calculation run in D2h misses out some of the excitations found in the C1 calculation, including what should be the lowest excitation.
A comparison with CC2 suggested that the lowest excitation for these larger molecules belonged to the au irrep. So we ran a calculation using targetsym au and find not only the au lowest state but also states belonging to irreps b2g, b3g and b1u, all of which are not found for the normal D2h calculation (i.e. without targetsym au), even if we ask for 24 roots.
Now we're aware that the TD-DFT algorithm might struggle if there is no(t sufficient) overlap between the starting vectors and the roots to be found but this seems to be the most extreme example we have ever seen (e.g. we find 24 states lying up to 1.4 eV higher in energy). Also, it seems to be related to the usage of symmetry (in C1 we find the states) and it seems to always the same set of irreps that go missing (au, b2g, b3g and b1u). All and all this makes us wonder if this is more than bad luck, solvable by asking for even more states.
What could be going on here?
Thanks,
Martijn & Cristina
|