LAMMPS WWW Site - LAMMPS Documentation - LAMMPS Mailing List Archives
Re: [lammps-users] Bad dynamics
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [lammps-users] Bad dynamics

From: Axel Kohlmeyer <akohlmey@...24...>
Date: Tue, 24 Apr 2018 11:51:29 -0400

please note, that what you are asking about is how to set up,
validate, and interpret MD simulations and the corresponding force
fields. none of that is directly linked to LAMMPS, but applies to all
MD codes.
as i already suggested in a previous response, you need to discuss
with your adviser or have you adviser identify a competent
collaborator to help, train, and advise you on these topics.

On Tue, Apr 24, 2018 at 10:51 AM, 徐浩 <xuhao@...6726...> wrote:
> Dear All,
> I recently simulated crosslinked epoxy resins. I encountered problem of
> missing bonded atoms for quite a while. When using LJ potential (dreiding
> forcefield), everything is fine, but when switching to X6 potential (for
> relaxation after the crosslinking process using LJ potential), the run crash
> all the time at a certain step, although I have adjusted the
> timesteps,neighbouring list and so on.
> I asked a researcher who has done relevant simulation. He told me that he
> also has missing atoms when there is atom type of H__A (in dreiding, H__A is
> Hydrogen possible of forming h bond). And then he changed H__A to H_(normal
> hydrogen), the problem was solved. The difference between H__A and H_ is
> that the vdw interaction with a H__A is several orders lower than with H_. I
> followed this way and no missing atoms were found anymore, and the
> simulation can keep going. One question is that is it a really correct way
> to change H__A to be H_ ? Will the results be not realistic ? Then I tried
> to use another way to constrain H__A by fix shake and rattle, using

please note, that if i answer any of these questions with "yes" or
"no", there is not much of a value in it. it is *your* work, and *you*
have to make this decision. you cannot put in your paper a statement
like "i used this model parameters because some dude on a mailing list
told me it was ok". you have to do the validation on your own, since
you are responsible for your work and its correctness.

> fix 101 all shake/rattle 0.0001 20 0 t 7 # the type of H__A is 7 in my data
> file
> but I still encounter missing atoms using either shake or rattle. Could
> anyone help to see if I have provided an effective constraint?

again, you are asking for the kind of help, that is the job of an
adviser. a mailing list is a poor replacement for that.

> Another question is more general. I understand that X6 potential provide
> much softer cores compared with LJ. But why soft cores can cause bad
> dynamics? And since these two potentials are quite different, leading to
> much different densities for example, how can we judge which one is more
> realistic and which one is not ?

a potential is realistic if it reproduces properties of the simulated
compound as good as it can be expected for the kind of model chosen
and for as good as is needed for what you want to learn from a
simulation. there is not simple: this is good, this is bad parameter.
it always boils down to some kind of "it depends".


> Regards
> Hao
> ------------------------------------------------------------------------------
> Check out the vibrant tech community on one of the world's most
> engaging tech sites,!
> _______________________________________________
> lammps-users mailing list

Dr. Axel Kohlmeyer  akohlmey@...24...
College of Science & Technology, Temple University, Philadelphia PA, USA
International Centre for Theoretical Physics, Trieste. Italy.