Dear Axel Kohlmeyer
My purpose is also researching and I have known that the mailing list goal
is validating and developing models to make confidence about them.
​where is that written? i am answering on mailing lists, and particularly
this one for many years, and other people and me have consistently ​told
people, that this mailing list is not to validate and approve other
people's inputs. if a command does not behave as documented or if someone
has a specific question about how something is implemented, that is a topic
for the list.
whether your simulation setup is valid for your research is *your*
responsibility. it would become my responsibility only in the case of you
submitting a paper to a journal and me being asked to review it or if i was
part of your thesis comittee or similar.
I have been investigating on molecular dynamics simulation (exactly on the
reactive force field ReaxFF) since four years ago. In my MSc thesis I spend
the whole time to investigate the ReaxFF potential functions, it
applications. Finally I did some reactive simulations on the Flame
Synthesis processes (with and without imposing the constant external
electric field on my simulation box). As imposing the external electric
field, I have recognized the total energy of the system is *not*
*conserved* in the ReaxFF Fortran source code. I have extensive research
about this phenomenon (e.g. elicit the equation of the electric field from
the source code “*efih=vfieldx*23.02*c1c*ch(i1)*c(i1,1)*”, investigation
of this equation and its constant conversion factors, how the external
electric field effect on the reactive simulations, contact to experts of
this field and also to the developer of the ReaxFF). And in my PhD thesis I
have been deciding to use the Lammps/reax/c for my research since about one
year ago. *So, all of us have similar tasks.*
​what has that to do with the topic at hand? i can list much more
experience than you, did research and collaborated with researchers on much
more complex issues involving electric fields and dielectric properties,
and also most likely have written/modified more lines in the USER-REAXC​
code in LAMMPS than anybody except for the original author. these facts, do
not relieve you to do your due diligence, also do not oblige anybody else
to validate your work, and most certainly do not entitle you to what you
have been asking.
I had learned to install and run simulations on Lammps in my own limit and
I have read the whole document about electric field command, fix addforce
command, compute command, variable command, region command and other
related commands which I thing they are relating to the external electric
field. In my last email I told you “*I know that the external electric
field in my simulations is a classical approximation and it just impose
electric forces on my charged elements*” so *I knew that we are not
exactly applying an electric field!!*
​here you are wrong. the issue i am raising is not about whether this is a
classical approximation or not. electrostatic interactions for any
empirical model in LAMMPS (and that includes ReaxFF) are *always*
classical. however, what fix efield does is a crude approximation even
within that model, compared to, for instance, modeling the same potential
difference through placing explicit charged particles arranged in planes
behind a repulsive wall.​
By the way, I have some technical questions about this subject in my mind:
1- Is my electric field command in the input file correct or not?
​whether it is syntactically correct or not, you can easily gather from the
LAMMPS documentation. the documentation also describes the physics that fix
efield implements. if there is a discrepancy between the documentation and
the actu​al behavior, this would be a topic for this list (or the issue
tracker on github), however, whether this is adequate for *your* research
is *your* job to determine and not anybody elses.
2- Is the total energy conserve in the lammps/reax/c simulations if we
apply an external electric field or not? And what is a solution for this
probably problem?
​again, whether the reax/c pair style sufficiently conserves energy or not,
is something that you need to test and verify. since ReaxFF has more
adjustable parameters, it may take more effort than for a simpler model.
outside of the usual suspects that affect energy conservation (system
setup, time step, neighbor list update settings, system manipulations
through fixes), you also have to tweak multiple reaxff specific parameters,
most importantly the convergence of the charge equilibration.
that you ask about energy​ conservation with an external field (or force)
applied significantly discredits your claim of experience in MD
simulations, as that question can be simply answered from basic
considerations. if you are as seasoned and well trained a person as you
claim, you should know the answer to that question.
3- Is the result of an external electric field in lammps/reax/c runs from
this equation “*efih=vfieldx*23.02*c1c*ch(i1)*c(i1,1)” *in subroutine
efield reax code or* “F = E.Q” *in common lammps command?
​what fix efield does is documented in the LAMMPS manual, it does exactly
that and nothing else. i already stated in my previous criticism, that one
of the problems in using fix efield with reax/c is, that the field
contribution is not considered beyond adding the force term in the fix.
*I think the second one is the answer since when I compare my results from
the serial ReaxFF Fortarn code, the electric field in the lammps run is far
weaker than the electric field in the ReaxFF Fortarn code in the absolutely
similar conditions. *
*4- *If the answer of the last question is the first one, why the
electric field in the lammps run is much weaker?
​you are being redundant here.​
if you want to be certain, set up a simple test system, where you can
compute the outcome by hand or by other external means and compare.
*5- *It is correct that the external electric field cannot include in the
charge equilibration scheme in the ReaxFF force field potential but I think
the resulted forces on the partial charged particles can cleavage or deform
bonds and molecular structures and finally can affect resulted fragments or
products of the chemical reactions. So, as the products of a chemical
reaction are changed, the charge equilibration of the ReaxFF produce other
partial charges for the atoms in molecules due to the change of atomic
chemical environments. On the other words, the external electric field can
indirectly affect the atomic partial charges in the molecules and fragments!
​yes, it can. but ​having an effect at all. does not mean, that the
magnitude is correct.
If you can, please answer to this questions and likely other questions
that I will ask in lammps-mailing-list and if you cannot, you don’t have to
answer my questions.
the mailing list is a volunteer effort and not a service that you are
entitled to. ​if you cannot handle (constructive) criticism, you should
make an ​effort to avoid questions that raise such criticism. please keep
in mind, that criticism actually is a sign that somebody cares about you
doing the right thing. it would be easier to just let your questions
unanswered. as with all volunteer efforts, the quality of responses you get
significantly depends on what and how you ask. having your ego dented a
little bit, when you are crossing the lines of what is considered
appropriate, is a small price to pay for getting help at all.
​axel.​