A doubt of QM/MM interface

[taipinghu]

We can perform QM/pol-MM calculations using Gaussian for QM part calculation and Tinker for MM part calculation. I just have a doubt:
How does the MM part respond to the QM part for a self-consistent calculation? Or only self-consistent calculation of MM part is taken into account. I think the following processes must be taken into consideration:
(1)calculate QM part and get electron density;
(2)calculate MM part based on the QM electron density;
(3) loop for (1) and (2) in order to converge.
And can you provide a more detailed description about it in the Garleek interface? Thank you.

[jaimergp]

Hello! Thanks for using our software!

To the best of my knowledge, Gaussian's ONIOM implementation does not provide polarizable embedding options. Since garleek only provides convenience wrappers around Gaussian's external mode, there is not much we can do about it in this current state. (Gaussian is "governing" the calculation and calling garleek on each iteration).

If I understood correctly, implementing a workaround as you suggest would require inverting this relationship: garleek would become the master and control Gaussian on demand (in a fashion similar to EasyMECP. I don't have the time to code something like that, but I would be happy to review a potential pull request if you want to try!