How to get the pair_style from Pair Coeffs? - from data file to in file

Dear lammps users,

I use Material Studio (MS) to construct the structure and use the tool ‘msi2lmp’ to convert it to data file. In the processes of typing in MS and conversion, I use ‘cvff’ forcefield.

The Pair Coeffs in the data file are like this:

Pair Coeffs

1 0.0380000011 2.4499714540
2 0.0389999952 3.8754094636
3 0.0389999952 3.8754094636

Does anyone can tell me what I should choose for the pair_style in the ‘in’ file? I know there should be some connection between pair_style and Pair Coeffs, but there are so many kinds of pair_styles that have similar Pair Coeffs. I really do not know how to choose.

Does the pair_style only relate to the force field? If so, what the corresponding pair_style of ‘cvff’ force field.

Thanks for your time,

Dongwei

Dear lammps users,

I use Material Studio (MS) to construct the structure and use the tool
'msi2lmp' to convert it to data file. In the processes of typing in MS and
conversion, I use 'cvff' forcefield.

The Pair Coeffs in the data file are like this:

Pair Coeffs

   1 0.0380000011 2.4499714540
   2 0.0389999952 3.8754094636
   3 0.0389999952 3.8754094636

Does anyone can tell me what I should choose for the pair_style in the 'in'
file? I know there should be some connection between pair_style and Pair
Coeffs, but there are so many kinds of pair_styles that have similar Pair
Coeffs. I really do not know how to choose.

Does the pair_style only relate to the force field? If so, what the
corresponding pair_style of 'cvff' force field.

most classical force fields use some variation of the lennard-jones potential,
but there can be subtle differences, thus the recommended strategy
(not restricted to CVFF):

look up the functional form for the bonded/non-bonded term(s) in there
desired force field. there has to be some kind of publication of it.
there also are likely recommended parameters for cutoffs and how to
treat coulomb, if present and other special features. running
simulations without having certainty about this, increases your
chances to produce inaccurate/inconsistent simulation data.

you can compare this information to the pair styles provided by lammps
and thus select the pair style and other styles.

axel.

Stephanie will probably know the answer to this.
For class2 it is pair_style lj/class2. For cvff it
is probably juts pair_style lj/cut.

Steve

Dear Dongwei,

I do not use the cvff pair potential, but I do believe one would use the
lj/cut potential style in LAMMPS. However, the best place to find the
answer to your question is the journal article corresponding to the cvff
force field. Use the pair style for which the force field was developed.

Cheers,
Stephanie

Dear Axel,

Thank you so much for your quick and informative reply.

I do read the detail of the ‘cvff’ forcefield. For the nonbonded terms, it contains the lennard-jones potential and additional Coulombic interaction. By comparing the style with LAMMPS, I think the pair_style 'lj/cut/coul/cut’ should be used.

Thank you again!

Dongwei

Dear Steve and Stephanie,

Really thank you for your help!!! Your replies encourage me to learn LAMMPS and use LAMMPS in my future research. There are great people behind to support the users.

I read carefully about the ‘cvff’ forcefield and have decided to use ‘lj/cut/coul/cut’ pair_style because there are both LJ and Coulombic interactions in the unbounded terms of ‘cvff’ forcefield.

Thank you all!

Dongwei

Dear Axel,

Thank you so much for your quick and informative reply.

I do read the detail of the 'cvff' forcefield. For the nonbonded terms, it
contains the lennard-jones potential and additional Coulombic interaction.
By comparing the style with LAMMPS, I think the pair_style 'lj/cut/coul/cut'
should be used.

hmm... for a pair style using coul/cut, you should check whether it
would not be better to use the corresponding coul/long variant in
combination with kspace style pppm (or ewald for very small systems).

coulomb interactions are pretty long-range and using a cutoff may not
give the best result and a large enough cutoff will have a very bad
performance.

axel.

Dear Axel,

Thank you for your suggestion!

The could/long maybe a better choice. I will try.

Thanks,
Dongwei

Hi All,

cvff is a rather old forcefield, originally intended for protein simulations. As originally implemented, group-based cutoffs were used. In group-based cutoffs, neutral groups of atoms are treated as a unit, with one of the atoms defined as the switching atom.The cutoff is applied based on the distance between the switching atoms. Based on the distance, the interactions between all atoms in the two groups are included or excluded, or, within the switching range are included but smoothly tapered to zero at the cutoff distance. Again, the smoothing is based on the distance between the switching atoms, not the individual pairs between the two groups. LAMMPS does not support group-based cutoffs.

Why this rather complex scheme? Electrostatic interactions are very strong, and die off slowly, so at typical cutoff distances of 10-20 Å the energy and forces are substantial. Arbitrarily cutting of these interactions thus leads to very large errors and artifacts in the calculation. For example, dipolar molecules with artificially align around e.g. an ion such that the attractive part of the dipole is inside the cutoff and the repulsive part is just outside, since this will lower the energy massively and artificially. Thus with cvff or any other forcefield where there are significant charges on the atoms – say 0.2 or more – simple cutoffs are very dangerous. If you can define neutral groups, all these issue vanish if using group-based cutoffs, yet one does need the expense and complexity of e.g. Ewald sums. However, it is neither easy nor possible to automatically generate neutral groups in general, though it is fairly simple for regular structures such as the proteins that cvff targeted. Since LAMMPS does not handle group based cutoffs, you should look to the pair styles that include long-range electrostatic interactions using Ewald or cell methods, like ewald and ewald/n or pppm. As far as I know, of these only ewald/n will work for non-orthorhombic cells, which may be a consideration depending on your systems.

Paul

[email protected]…1795…

Phone: +1.760.495.4924 x 102
Cell: +1.505.603.8182

This email is covered by the Electronic Communications Privacy Act, 18 U.S.C. Section 2510-2521. This message and any attachments hereto may contain confidential information intended only for the use of the individual or entity named above. If you are not the intended recipient(s), or the employee or agent responsible for delivery of this message to the intended recipient(s), you are hereby notified that any dissemination, distribution or copying of this email message is strictly prohibited. If you have received this message in error, please immediately notify the sender and delete this email from your computer. The sender does not waive any privilege in the event this message was inadvertently disseminated.

MedeA® and Materials Design® are registered trademarks of Materials Design, Inc.

Dear Paul,

Really thank you for your comments.

I agree with you about using pair styles that include long-range electrostatic interactions. However, at present, I can not test the result yet. I will try later when the machine and the new installed software are ready.

Thank you again. I learnt a lot from you!

Dongwei