Regular grid of gaussians

[Paolo Pegolo]

Dear cp2k community,

In the literature (e.g., here and here) people refer to the possibility of adding to the set of atomic centered gaussians a grid of gaussians functions fixed in space, that helps describing the wavefunction of a system with, e.g., loosely bound electrons. They use cp2k, but I have not been able to find any information about it in the manual.

How can one do this?

Thank you, 

Paolo

[hut...@chem.uzh.ch]

Hi

You should be able to do such a calculation by defining a new
atom type, e.g. AGHOST

&KIND AGHOST
BASIS_SET "my-basis-set"
GHOST
&END KIND

Then add atoms of this type on a regular grid to the coordinate definition.

For a test I would start with a small number of ghost atoms and check
the number of basis functions and the outcome of a geometry optimization
for a simple molecule.

regards

Juerg Hutter

--------------------------------------------------------------
Juerg Hutter Phone : ++41 44 635 4491
Institut für Chemie C FAX : ++41 44 635 6838
Universität Zürich E-mail: hut...@chem.uzh.ch
Winterthurerstrasse 190
CH-8057 Zürich, Switzerland
---------------------------------------------------------------

-----cp...@googlegroups.com wrote: -----
To: "cp2k" <cp...@googlegroups.com>
From: "Paolo Pegolo"
Sent by: cp...@googlegroups.com
Date: 01/03/2021 02:50PM
Subject: [CP2K:14462] Regular grid of gaussians

--
You received this message because you are subscribed to the Google Groups "cp2k" group.
To unsubscribe from this group and stop receiving emails from it, send an email to cp2k+uns...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/09b518d1-6cbf-4b83-ae5e-53a2bee76125n%40googlegroups.com.

[Paolo Pegolo]

Thank you for your answer.

I wrote a file MY-BASIS with the new basis for the element AGHOST with a basis set called my-basis-set. The beginning of such file reads

  AGHOST my-basis-set

Then I added

BASIS_SET_FILE_NAME MY-BASIS

to the &DFT section of the input file, 

&KIND AGHOST
BASIS_SET my-basis-set
GHOST
&END KIND

to the &SUBSYS section, and I put the coordinates of the AGHOST atoms I need to the XYZ file I use to define the topology of the system. I get the error 

Element <AGHOST> provided for KIND <AGHOST> which cannot be mapped

 with any standard element label. Please correct your input file! 

Is there something wrong I am doing?

Regards,

Paolo 

[hut...@chem.uzh.ch]

Hi

You can specify an element type using the keyword ELEMENT XX
in the KIND section.

regards

Juerg Hutter
--------------------------------------------------------------
Juerg Hutter Phone : ++41 44 635 4491
Institut für Chemie C FAX : ++41 44 635 6838
Universität Zürich E-mail: hut...@chem.uzh.ch
Winterthurerstrasse 190
CH-8057 Zürich, Switzerland
---------------------------------------------------------------

-----cp...@googlegroups.com wrote: -----
To: "cp2k" <cp...@googlegroups.com>
From: "Paolo Pegolo"
Sent by: cp...@googlegroups.com

Date: 01/04/2021 02:43PM
Subject: Re: [CP2K:14469] Regular grid of gaussians

To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/e52e835d-8276-4355-8320-c2a8129b03ean%40googlegroups.com.

[Paolo Pegolo]

Ok, thank you. I see that the required ELEMENT keyword affects only the atomic label, in this case.
If I want to use ADMM, will I have to use the same basis set also for the AUX_FIT part? Or is there a way to generate a consistent auxiliary basis?

Regards, 

Paolo

[hut...@chem.uzh.ch]

Hi

you will need to define an AUX_BASIS also for the ghost element.
Your on your own to find a reasonable basis that is compatible
with your ghost basis.

Just a wild guess. If you are using a grid of basis functions, your
basis will be very small (single Gaussian, maybe s and p types).
In that case use the same basis for the AUX_BASIS.

Other idea: try to use no ADMM basis on the grid.
Method 1) try AUX_BASIS NONE
Method 2) Use ADMM METHOD BLOCKED_PROJECTION

I have never tried any of these, you need to experiment.

regards

Juerg Hutter

--------------------------------------------------------------
Juerg Hutter Phone : ++41 44 635 4491
Institut für Chemie C FAX : ++41 44 635 6838
Universität Zürich E-mail: hut...@chem.uzh.ch
Winterthurerstrasse 190
CH-8057 Zürich, Switzerland
---------------------------------------------------------------

-----cp...@googlegroups.com wrote: -----
To: "cp2k" <cp...@googlegroups.com>
From: "Paolo Pegolo"
Sent by: cp...@googlegroups.com

Date: 01/05/2021 04:25PM
Subject: Re: [CP2K:14481] Regular grid of gaussians

Ok, thank you. I see that the required ELEMENT keyword changes only the atomic label, in this case.
If I want to use ADMM, will I have to use the same basis set also for the AUX_BASIS part? Or is there a way to generate a consistent auxiliary basis?

Regards,

Paolo

To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/4b1f4318-adf3-4ea3-99c7-bf08c7063086n%40googlegroups.com.

[Paolo Pegolo]

Thanks a lot for your help and suggestions.

Regards, 

Paolo