[QM/MM] The understanding of QM/MM method implemented in CP2K.

[Xiaohui Yang]

Dear all, 

Thank you for reading this post. Here I have a few questions about the 
QM/MM calculation that implemented in the CP2K.

I am planing to embed a charged QM system in the large MM box. The QM box 
include one charged particles, e.g. Cl-, and some water molecules, and the 
MM region is bulk water. 

The first part of questions is about the fully PBC QM/MM method. Thanks to 
Laino and co-workers, we can have this great method in CP2K. After reading 
their articles, I am still confused about the boundary problem between QM 
and MM regions. 

1. The hydrogen bond in the QM region. 

Because the QM box can interact with its images in the PBC, if a water 
(water A) close to the boundary forms a hydrogen bond with another water 
molecule in the image (water B'), the water A should feel the hydrogen bond 
with water B', and has the electrostatic + vdW interaction with classical 
water molecules in the vicinity. However, in a reasonable QM/MM system, the 
force applied on water A isn't include the hydrogen bond with water B'.

2. The artifical "vacuum" between QM and its images. 

Normally in the QM/MM system, the QM box shoud be a little bit larger than 
the "optimal size" (which means the solution have correct density). If I 
want to reproduce the QM region with correct density, and to prevent the QM 
molecules move out of the QM box, I need to enlarge the QM size a little 
and apply the "WALL" in the QM/MM section.This method is convenient and 
useful. However, there is a few angstrom thickness of "wall skin", for 
example, 1 angstrom (0.5 for each side). In the full PBC QM system, there 
will be a kind of "vacuum" betweem QM box and its images. Therefore, the 
interaction between the QM region and its images will be affected.

Because I don't fully understand above situations, I decide to employ the 
isolated QM system in the PBC MM box. Here I have some questions about the 
POISSON SOLVER.

1. The "POISSON" section in the DFT part. 

If this section is explicitly writen in the input file, there is a warning 
in the output said that this part is ignored in the QM/MM calculation, and 
the electrostatic is controlled only by "PERIODIC" in QM/MM section. 
However, based on my test, the poisson solver of the QM box is contorlled 
by the "POISSON" section in the "DFT" section, and the "PERIODIC" in QM/MM 
section does not affect the result at all. 

2. The "MULTIPOLE SOLVER" and "MULTIPOLE".

As the non-PBC QM box is employed, the MT solver is prefered in my 
calculation. However, the MT solver requires large QM box, which results in 
high computational cost. And I don't know if the MT solver could be used 
together with the "MULTIPOLE" decouple/recouple method to reproduce the 
long range electrostatic potential. 
I noticed that there is a "MULTIPOLE" solver, but I have no idea about it. 
Can I choose this solver in the NON-PBC charged box? As far as I know, for 
PBCs, the poisson solver is "periodic"; for non-PBCs, the "MT" or "WAVELET" 
is required. 

Thanks again for your reading, any comments and suggestions are 
appreciated. 

Attached is a very simple example of QM/MM input and output files to point 
out the misunderstanding of "POISSON" section in both "DFT" part and "QMMM" 
part. (I choose totally different periodic and solver in both sections on 
purpose, and the output POISSON part shows the keyword consistent with DFT 
section.)

BTW, is it possible to use OPLS/AA force field in CP2K? :D

Best Regards,

Xiaohui


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20180514/4818dbce/attachment.htm>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: cl.psf
Type: application/octet-stream
Size: 315 bytes
Desc: not available
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20180514/4818dbce/attachment.obj>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: setup.inp
Type: chemical/x-gamess-input
Size: 3898 bytes
Desc: not available
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20180514/4818dbce/attachment.inp>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: out
Type: application/octet-stream
Size: 57554 bytes
Desc: not available
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20180514/4818dbce/attachment-0001.obj>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: input.inp 2
Type: application/octet-stream
Size: 974 bytes
Desc: not available
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20180514/4818dbce/attachment-0002.obj>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: input.ff
Type: application/octet-stream
Size: 3165 bytes
Desc: not available
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20180514/4818dbce/attachment-0003.obj>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: input.coord
Type: application/octet-stream
Size: 34 bytes
Desc: not available
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20180514/4818dbce/attachment-0004.obj>