[Pw_forum] Point Charges for Molecular Dynamics

Axel Kohlmeyer akohlmey at gmail.com
Mon Aug 11 13:55:17 CEST 2014

On Sun, Aug 10, 2014 at 11:46 PM, Kevin Schmidt <kevohs at gmail.com> wrote:
> Hello PWSCF Forum,
> First time caller, long time listener.  I am working with a group using
> Quantum Espresso to extract two-body interatomic potentials (IP) for an
> ionic crystal system (metal borides, to be exact).  These potentials will
> then be put to use in molecular dynamics software (DL-Poly).  From my
> understanding with MD, most software nowadays typically employs
> electrostatics as a separate potential from the heteronuclear IP's.  In

it always has been like that for all cases where "full" (i.e. not
screened) electrostatics are assumed.
however, you have to have a good look at methodologies for force field
parameterization. simply computing a point charge distribution from
some QM density and dropping them into an MD force field is highly
unlikely to work. parameterizing a classical force field is a
substantial amount of work and having some initial guess for the
charge distribution is only the beginning. a lot depends on what you
are going to do with your model and whether it will be combined with
other elements from an existing force field. you mention ionic
crystals, so that would typically require a many-body method with
dynamic charges via some charge equilibration system or some bond
order constraint system for achieve good accuracy.
if you stick to a simpler model with fixed charges, you will have to
spend a signficant amount of time validating your choice of
parameters, since very often adjustments need to be made to make the
parameterization able to reproduce some known experimental properties
(and remember, also DFT calculations do not always reproduce those

> order to (1) get the correct heteronuclear IP and (2) simulate a "realistic"
> electrostatic environment in MD, it would be best to have the right point
> charges for the system.  After browsing through the forum and user guides, I

but there is the problem. what are the "right" charges. there are
multiple recipes to map the total density distribution to individual
point charges at the location of the atoms, but this will have errors
since it is often quite ambiguous how to do this projection. for force
field parameterization often methods trying to reproduce the
electrostatic potential are used.

> noticed that PP.x (plot_num = 0) has the ability to output charge density
> distributions.  The output I received had, depending on which plot type
> used, five or six columns, of which I am assuming are related to charge
> values on a grid.  Is this correct, and if so, how do I find the
> corresponding grid?
> My idea was to integrate the charge density in a Wigner-Seitz cell around
> each ion to get point charges for use in electrostatics.  I'm not sure if
> there's a better way to accomplish this goal; this was my attempt.  I

there are multiple methods available. some are even implemented in
Q-E. however, before you continue, please read up on force field
parameterization and do not waste any time on doing Q-E calculations
before you know what you need to do.

> apologize if I have overlooked a thread, please notify me so.  Thanks in
> advance and I look forward to your advice.
> P.S.  The output I received for a model system is given below for a
> specified "filplot", plot_num = 0.
>       72      72      72      72      72      72       4       2
>      1        7.00000000      0.00000000      0.00000000      0.00000000
> 0.00000000      0.00000000
>      1191.5371196339       12.0000000000       80.0000000000     0
>    1   La   11.00
>    2   B     3.00
>    1       0.000000000    0.000000000    0.000000000    1
>    2       0.000000000    0.500000000    0.500000000    2
>    3       0.500000000    0.000000000    0.500000000    2
>    4       0.500000000    0.500000000    0.000000000    2
>   3.434814311E-01  3.431749032E-01  3.422272391E-01  3.405249826E-01
> 3.378906730E-01
>   3.341004832E-01  3.288993448E-01  3.220354913E-01  3.132873382E-01
> 3.024951442E-01
>   2.895821234E-01  2.745777240E-01  2.576219405E-01  2.389737928E-01
> 2.190001080E-01
>   1.981636489E-01  1.769910061E-01  1.560378501E-01  1.358452121E-01
> 1.168975755E-01
>   9.959052347E-02  8.420256729E-02  7.088392771E-02  5.965905586E-02
> 5.043519819E-02
> ..... more values
> --
> Kevin Schmidt
> Chemical Engineering Department
> University of Nevada, Reno, USA
> _______________________________________________
> Pw_forum mailing list
> Pw_forum at pwscf.org
> http://pwscf.org/mailman/listinfo/pw_forum

Dr. Axel Kohlmeyer  akohlmey at gmail.com  http://goo.gl/1wk0
College of Science & Technology, Temple University, Philadelphia PA, USA
International Centre for Theoretical Physics, Trieste. Italy.

More information about the users mailing list