Closed corinwagen closed 2 months ago
Note that similar issues have been seen before: #794
Similarly when I compute f(-) for fluorobenzene, I get ortho = -0.049, meta = -0.045, and para = -0.081.
The literature values are ortho = 0.047, meta = 0.026, and para = 0.093. Granted, these data are computed using HF/3-21G so it's not surprising that the numerical values are different, but the total flip in sign makes me think there's something pretty substantially wrong with the xTB values.
Indeed, the sign switch I introduced in #794 seems to be wrong since we are using Mulliken charges and not Mulliken populations. Therefore, the sign of the fukui functions is switched! I will submit a PR to fix this. Thank you for bringing this to our attention.
any time. the values are correct otherwise?
Otherwise they should be correct. They are calculated according to equations 2a, 2b, and 2c of this paper (also shown at the end of the abstract) where the term Mulliken charges can be misleading in my opinion since I believe what they are actually using is what I understand as Mulliken population. The difference with SCM I would attribute mostly to them using the Hirshfeld partitioning (probably a better idea) and secondly on the STOs they use instead of GTOs, although this difference I assume should be very small.
Describe the bug Fukui indices are bizarre and seem wrong.
For carbon monoxide, I get these indices:
This seems wrong. The SCM Fukui tutorial gives very different values:
And negative Fukui indices are at least a bit unexpected (see discussion here).
To Reproduce Steps to reproduce the behaviour:
co.xyz
:xtb co.xyz --verbose --gfn2 --vfukui
.Please provide all input and output file such that we confirm your report.
Expected behaviour I expect Fukui indices more like those from SCM (vide supra).
Additional context xTB 6.6.1 installed from conda into a Docker image.