[lammps-users] Ways to validate a CO2 model

Dear all,

i have created a CO2 model with Lammps. i have run some simulations and i visualized them. The thing is now that ok,i create "nice pictures" but how can i know that all these are physically and chemically correct?How can i use Lammps in order to validate my results?

Regards
Dimitris

Dear all,

i have created a CO2 model with Lammps. i have run some simulations and i visualized them. The thing is now that ok,i create "nice pictures" but how can i know that all these are physically and chemically correct?How can i use Lammps in order to validate my results?

have you had a look into a text book on molecular dynamics simulations?
if not, i recommend to do that before moving on; if you did, perhaps a
second look doesn't hurt. it may provide some enlightenment for you.

you are not the first person trying to simulate CO2 with classical MD.
have you thought about doing a literature search? back in ancient times
when i learned doing MD, my advisor insisted that i did this long before
setting up the first simulation. at that time it sometimes seemed like a
hassle, but it kept me from making many mistakes others had already done.

cheers,
    axel.

p.s.: to save you some suspense. of course you want to match your
simulation with experimental data, especially thermodynamic and
structural information, which in turn you can derive from
the simulation data through statistical mechanical methods.

Thank you for your answer.I have read some things about MD and i have made a literature review around the subject. The thing is that i don't have a chemical backround ,at least until now..So what i am asking is which quantities should i use (in a meaning which are the easiest to compute at least now that i am newbie at Lammps) and where can i find something like a tutorial or some guidelines in order to compute a quantity with the data that i have from Lammps.

Regards
Dimitris

dear dimitris,

Thank you for your answer.I have read some things about MD and i have
made a literature review around the subject. The thing is that i don't
have a chemical backround ,at least until now..So what i am asking is

well... tough luck. there ain't no escape from the blues.

while you can run simulations without knowing what you are doing,
they will be pretty much useless since you cannot interpret them
unless you learn the underlying physics. i was trained as a (wet lab)
chemist myself and can tell you that this has helped me very little.
if you want to learn how to do MD well, you have to learn it from
somebody that practices it.

which quantities should i use (in a meaning which are the easiest to
compute at least now that i am newbie at Lammps) and where can i find
something like a tutorial or some guidelines in order to compute a

the tutorials you are looking for are called "text book" and
"colleague". i did receive good help from my thesis advisor and
the books "understanding molecular simulations" by frenkel and smit,
"statistical mechanics" by mcquarrie and advanced help from "theory
of simple liquids" by hansen and macdonald. the latter should
have some detailed discussion of linear molecules, which might be
of particular interest for you, but i would not recommend looking
at it, before you have gone tete-a-tete with mcquarrie.

the best guideline is to follow what other people have
successfully used to validate their models.
it is as simple (and as difficult) as that.

quantity with the data that i have from Lammps.

please note, that only some of the analysis is automatic
and can be done from within lammps. traditionally, the
majority properties is computed in postprocessing, often
with small analysis programs/tools written specifically
for the system at hand (which tends to be far easier than
writing a generic tool).

cheers,
    axel

How about diffusion coeffs, thermodynamic state variables, etc.

Steve

Ok.the thing is that i am newbie in lammps so i don't know how to compute a quantity using the data taken from lammps

Regards
Dimitris

Ok.the thing is that i am newbie in lammps so i don't know how to
compute a quantity using the data taken from lammps

but this is why you *should* look into the text
books, as they *do* explain how to get properties
from simulation data. if you know for example, that
and how you can compute self-diffusion from the mean
squared displacement or velocity auto-correlations,
then you are most of the way on extracting your
properties of interest.

as i wrote before, lammps is a simulation engine,
it will not automagically provide you with all
answers if you just sprinkle a few magical keywords
into the input. some analysis can be done right
along the simulation, some not. some can be done
either in postprocessing or during the run.
in my experience, running the analysis during
the simulation can be problematic, since you
may not know when your system is equilibrated
well enough to start accumulating data, or find
that you want different settings. other tests
(like coexistance curves) need multiple runs
right away.

i am sorry, if this is a rough awakening for
you, but this is the real life.
no pain, no gain.

cheers,
   axel.