RESP charges

[anirban mondal]

Dear CP2K users/developers,

  I am trying to compute the RESP charges for some periodic and nonorthorhombic systems (using &RESP). But I found some error while calculating the charges;

CP2K| UNIMPLEMENTED, RESP charges implemented only for orthorhombic cells!

How should I use this keywords combination or adding extra keywords to make it work?

Hope you can help me.

regards;

anirban

--

Anirban Mondal

Molecular Simulations Lab.

Jawaharlal Nehru Centre for Advanced Scientific Research

Bangalore-560064

India

[Dorothea Golze]

Hi,

it seems you are using an old CP2K version. The usage of non-orthorhombic boxes is possible for periodic systems requiring periodic RESP fitting. This option is available since August 2012. Periodic RESP is automatically enabled by using a periodic Poisson solver.

Regards,

Dorothea

2013/8/20 anirban mondal <anirbanbl...@gmail.com>

--
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 post to this group, send email to cp...@googlegroups.com.
Visit this group at http://groups.google.com/group/cp2k.
For more options, visit https://groups.google.com/groups/opt_out.

[Carlos Campana]

Hi,

If you are running "real" periodic systems and not some molecule embedded in a big periodic box you want to be careful as to how to define electrostatic charges from ab-initio calculations. The zero of the electrostatic potential is ill-defined for periodic systems and different ab-intio codes can give you very different ESP charges. I worked on this problem a few years ago a found a simple solution to it. See: J. Chem. Theory Comput., 2009, 5 (10), pp 2866–2878 DOI: 10.1021/ct9003405
My apologies for the personal blurb but it may turn out useful to you after all.

Carlos

[Dorothea Golze]

Hi Carlos,

the problem you mention is taken care of in the CP2K implementation of the periodic RESP fit.

Regards,

Dorothea

2013/8/20 Carlos Campana <campa...@gmail.com>

[Carlos Campana]

Thanks Dorothea. Good to know. I'm certainly more an spectator of this list than an active CP2K user, thus, not very aware of all the up-to-date features of the code.

C.

[anirban mondal]

Thank you all for kind reply. Now I am using the latest version of CP2K. But the RESP computed charges are found to be NAN for all the atoms.

Here I have attached the files. 

Hope you can help me.

[Matthias Krack]

Hi Ani,

you should use a much higher cutoff than 280 Ry. This value is by far not sufficient for your system containing fluorine as indicated by the poor value of 0.00012 for the "Total charge density on r-space grids which should be at least 1.0E-6 or lower. I would suggest a cutoff of 400 or better 600 Ry. You may also reduce the value for EPS_DEFAULT to 1.0E-12. I am not sure, if this will fix your problem with the NAN in the RESP output, but in order to produce reasonable results you should use the proposed settings anyway.

Regards,

Matthias

[Dorothea Golze]

Hi Anirban,

I changed the input structure for the RESP fitting a little bit last year. Example files can be found in the regtest files:
cp2k/cp2k/tests/QS/regtest-properties/resp

The reason why you get NaNs is that you didn't collect any fitting points, i.e. you didn't define where you want to fit the potential. There are two options how to do that

a) fitting the potentials in shells of radii Rmin-Rmax around each atom.  This is the common way for collecting fitting points, i.e. for simple non-periodic systems like small molecules. But it also makes sense to collect the fitting points like  this for periodic systems like MOFs

b) fitting the potential in the region above the surface of a slab-like periodic system

Option a) is initialized by the RESP subsection &NONPERIODIC_SYS and option b) by &PERIODIC_SYS. I admit the names of these subsection are maybe a little bit misleading. In these subsections you only define how to sample the fitting points. Whether a periodic RESP fit is required or not is solely decided on the usage of a periodic or non-periodic Poisson solver.

If you want to visualize which fitting points you actually sampled, there is an option in the RESP print section called &COORD_FIT_POINTS. This option is merely meant for testing since the xyz file with the coordinates of these points can be really large (and you have to adjust number of "atoms", i.e. number of atoms+fitting points in the header of the xyz file.)

Regards,

Dorothea

[Bita Raheb]

Hi,

I am new to CP2K. I want to calculate RESP charge. but I get error ;  for all the atoms=NAN.  what should I do?

Here I have attached my INPUT  file.