During the molecular dynamics simulations, if we do not know to choose which potential functions (force field), what should we do? How to choose the correct force field (aka, inter-atomic potentials)?
In the website "https://www.ctcms.nist.gov/potentials/", different potentials of the same element are available. There it is clearly mentioned about the details of the potential like for what purpose each potentials are developed.
If you don't find your application, you can either create your own potential or modify the existing potential for your purpose by changing some values.
After you create or modify, you can verify the accuracy of the potential by comparing the basic properties/variables like lattice constant, elastic constants, diffusion coefficient etc. with the literature or experimental values.
Please go through some good literature about force field e.g., Charmm , Amber etc. to get the idea about advantage or disadvantages of a particular force field.
Dear Debasish
Thanks for your kind advise. You mentioned that we have to choose which potential function (aka force field) is more appropriate, what about if we can not find the relevant force field, for example, if I wanna to simulate the mechanical properties of some rocks using MD, in this case, what should I do? Many thanks!
Regards
In the website "https://www.ctcms.nist.gov/potentials/", different potentials of the same element are available. There it is clearly mentioned about the details of the potential like for what purpose each potentials are developed.
If you don't find your application, you can either create your own potential or modify the existing potential for your purpose by changing some values.
After you create or modify, you can verify the accuracy of the potential by comparing the basic properties/variables like lattice constant, elastic constants, diffusion coefficient etc. with the literature or experimental values.
The first step is reading literature about the force fields. If that doesn't help you can search by MD simulation papers about works that are similar to the work that you are interesting in perform.
If you are doing something unique maybe you can try different force fields and see which one provides you better results through comparison with experimental values in the literature to some properties that you can measure in the simulations.
There is no such thing as a correct force field. In a nut-shell they are all wrong. What you want to do is work out what properties do you want to compromise on. As others have mentioned for this literature helps.
You can also work out what you observables you want to measure and then look for force fields which can measure that reliably.
If you are still unsure, go ahead and use one. Simulations by themselves are not helpful. Usually you have experimental data. Try recovering that from your simulation, if you can well keep sticking to that force field. Not all types of results can be recovered, so you may need some approximations.
At the end of the day, here is a bottom line.
If you want detail and accuracy, use CHARMM (or NAMD if you want to use the free version). If you want speed use GROMACS -- but remember the devil is always in the detail.
https://www.atomicrex.org/
https://www.potfit.net/wiki/doku.php
This may help you to create your own potentials.
Dear Zhaoyang,
There are certain parameters required for the use of certain potential functions or force-fields and these parameters are rigorously derived using Ab-initio methods or Density Functional Theory methods.
Now in order to choose relevant potential function/force-field for a specific problem that one wants to study, you have to first know what properties you want to study and then get relevant literature (Use the same forcefields/potential functions). Choose very well accepted field parameters. There will always be discrepancies and ambiguity even to choose the parameters required once you fix on a force-field. So always have a healthy speculation and choose what seems logical.
Hope this helps.
Regards,
Abhishek Peddakotla
- Badges
- Science topic