PBC within QM/MM

[Ludwig]

Dear all,

I have a doubt on how to define periodic boundary conditions (PBC) for a QM/MM system (protein with a small QM region and a large MM region that includes solvent). As far as I have seen, there are several sections where one can define cell information (sections 1-2 beneath) and Poisson solver methods for electrostatic interactions (sections 3-5):

1. /SUBSYS/CELL/PERIODIC (default XYZ): Specify the directions for which periodic boundary conditions (PBC) will be applied. Important notice: This applies to the generation of the pair lists as well as to the application of the PBCs to positions. See the POISSON section to specify the periodicity used for the electrostatics.

2. /QMMM/CELL/PERIODIC (default None): Specify the directions for which periodic boundary conditions (PBC) will be applied. Important notice: This applies to the generation of the pair lists as well as to the application of the PBCs to positions.

3. /DFT/POISSON/PERIODIC (default XYZ): Specify the directions in which PBC apply. Important notice, this only applies to the electrostatics. See the CELL section to specify the periodicity used for e.g. the pair lists.

4. /MM/POISSON/PERIODIC (default XYZ): Specify the directions in which PBC apply. Important notice, this only applies to the electrostatics. See the CELL section to specify the periodicity used for e.g. the pair lists.

5. /QMMM/PERIODIC/POISSON/PERIODIC (default XYZ): Specify the directions in which PBC apply. Important notice, this only applies to the electrostatics. See the CELL section to specify the periodicity used for e.g. the pair lists.

If I understand this correctly, in a QM/MM framework section 1 can be used to specify the periodicity of the large MM cell that encloses the entire system, and section 2 to define the periodicity of the small QM cell that encloses the electron density (using the Blöchl decoupling scheme described in Laino2006?).

What is a bit confusing to me are sections 3, 4, and 5, when section 2 is either None or Periodic. What are these three sections controlling exactly? The electrostatic interactions between (3) QM-QM’, (4) MM-MM’, and (5) QM-MM’ / MM-QM’ atoms in the primary–periodic’ cells? What happens with 3 and 5 if section 2 is set to None (i.e. system without QM' images)?

Thanks in advance,

Ludwig

[Marcella Iannuzzi]

Dear Ludwig, 

Section 1 and 2 control only the periodicity of the coordinates, but not of the electrostatic interactions. 

For a fully periodic QMMM calculation you need PERIODIC XYZ both in CELL as well as in POISSON, which by the way are the default. 

In addition, you might want to specify the section PERIODIC in QMMM, 

CP2K_INPUT / FORCE_EVAL / QMMM / PERIODIC

something like (example taken from regtests)

    &PERIODIC

     GMAX 0.1

      &MULTIPOLE

       RCUT 30.0

       EWALD_PRECISION 0.000001

      &END

    &END

Regards

Marcella

[Ludwig]

Dear Marcella,

Thanks a lot for the reply! What confuses me is the fact that 2. /QMMM/CELL/PERIODIC is None by default, is it not necessary to set it to XYZ for a fully QM/MM periodic calculation? The CELL section is only updating pair lists for the short-ranged real part of Ewald, and also updating coordinates in case they get out of the primary cell? And to neglect all QM interactions with the periodic images (as if the QM region were “isolated”), is it enough to set 3. /DFT/POISSON/PERIODIC and 5. /QMMM/PERIODIC/POISSON/PERIODIC equal to None?

Thanks again for your help,

Ludwig

[Marcella Iannuzzi]

Dear Ludwig,

Yes you are right,  sorry for the confusion. The default for QMMM&CELL is periodic none, because the pair lists are constructed according to the periodicity of the MM cell. 

The PERIODIC&POISSON sections control the electrostatics interactions (all of them). Hence, if you do not want them to be periodic, set them to NONE.

Regards

Marcella

[Ludwig]

Dear Marcella,

Thanks a lot for all your clarifications, it is really helpful! I am not sure to see what is the practical use of the QMMM/CELL/PERIODIC command. If the QM and MM boxes are equal, then this command is just the same as the one in SUBSYS, right? And if the QM cell is smaller than the MM one, activating this option would not include in the pair lists short-range interactions that in reality are not within the cutoff?

I made a quick scheme of the different options, the question above would be represented by setup 2 compared to 1. I would not expect large differences between the two setups, but still to me it makes more sense setup 1, which is the one by default if one explicitly writes the POISSON section in &QMMM (I suppose that missing this section without deactivating the one at &DFT would lead to fictitious periodic QM images within the MM cell).

Thanks again,

Ludwig