Developer of the CRYSTAL code!
Of course, in no case an isolated molecule would have access to neighboring functions. But in the crystal it does. If its own basis is not complete, a molecule in the crystal can get extra-stabilized by the neighboring functions not because of a "physical" or "chemical" interaction with neighboring molecules but because of an artifact due to the incompleteness of its basis. So if you were to compute the cohesive energy just as energy of the crystal - energy of the bare molecule, you would get an extra unphysical interaction. To make sure you get the right cohesive energy you could do one of two things: i) remove the extra interaction from the calculation on the crystal; ii) add it also to the molecular calculation so that it cancels when taking the difference (this is the MOLEBSSE strategy, also known as Boys-Bernardi counterpoise correction).
Hi Jonas,
As you correctly point out, in this case the molecular calculations are not affected by BSSE as you are extracting a single molecule (i.e. with no neighbors).
The calculation that is affected by BSSE is the one on the crystal, where each molecule within the molecular crystal could potentially "benefit" from the basis functions of the neighboring molecules if its own basis is not complete. So we use MOLEBSSE for this, to make sure we compare apples with apples (i.e. the energy of a molecule in the crystal that has access to the basis functions from neighboring molecules, with the energy of an isolated molecule that also has access to the same neighboring basis functions).
Hope this helps clarifying things a little,
Dear Prof. Piquini,
I am sorry but I never received your files. Could you please try again?
Dear Piquini,
thank you for your email. Is there any possibility to have a look at your input-output files? If you prefer you can send them to [email protected].
Hi,
We have checked the implementation. The values printed in the output of the BETAVIB calculation for SHG and Pockels are expressed in atomic units and correspond to \( \frac{1}{2}{\boldsymbol{ \beta}} \). That is, the factor of \( \frac{2\pi}{V} \) seems to be missing to make them the d tensor. In other words, by multiplying the values in the output by \( \frac{2\pi}{V} \) you should get d.
Please, let me know if you think this makes sense based on the values you get for your system.
Sorry for the confusion.
Hope this helps,
