Simulating the solid-liquid system without considering the interactions between solid and liquid molecules

Dear professors,

I am simulating the heat transfer between liquid argon and platinum. I do this follow a paper, and the paper method is to set NVT for the argon-platinum system without considering the interactions between argon and platinum at first, and then consider the interactions between argon and platinum to see the heat transfer employing NVE and thermostat.

When I simulate NVT without considering the interactions between argon and platinum, I find that the some of argon atoms enters into the platinum region. It is not reasonable, and there exists an error "atom lost".

How can I avoid this happen and realize the NVT without considering the interactions between two kinds of atoms?

Best wishes,

Xiaohui She
University of Wisconsin-Madison

Dear professors,

I am simulating the heat transfer between liquid argon and platinum. I do
this follow a paper, and the paper method is to set NVT for the
argon-platinum system without considering the interactions between argon
and platinum at first, and then consider the interactions between argon and
platinum to see the heat transfer employing NVE and thermostat.

​LAMMPS already supports multiple well established methods to compute
thermal conductivity. please see the examples/KAPPA directory and the
lecture notes from the tutorials section of the home page.​

When I simulate NVT without considering the interactions between argon and
platinum, I find that the some of argon atoms enters into the platinum
region. It is not reasonable, and there exists an error "atom lost".

​i disagree with your assessment. if you remove interactions, atoms can go
where they please, since there is nothing interfering. so i would consider
LAMMPS' behavior *very* reasonable and rather your expectations
unreasonable. ​from reading between the lines of what you describe, it
casts some doubts on the validity of the method you describe or on your
interpretation of the method.

How can I avoid this happen and realize the NVT without considering the
interactions between two kinds of atoms?

​this simply doesn't make sense on many levels. you cannot have two NVT
ensembles in the same box. you have some other ensemble that is neither.
NVT requires by definition a periodic homogeneous bulk system.​
furthermore, what would contain atoms in their place, if not interactions
with something else?

axel.

If you use neigh_modify exclude you can turn off interactions
between 2 types of atoms (Ar and Pt). You can also use
one fix nvt for the entire system. However when you
turn on Ar/Pt interactions (e.g. 2nd run), they will likely
be highly overlapped.

Steve