1. maximum number of neighbors
In the manual, it means literally *maximum number of neighbors* any atom
can ever has.
As I calculated for graphene monolayer, I set it as 3 because just an atom
meets three adjacent atoms.
However, when I must consider the VdW interaction between each layer, how
can I estimate it? Just Do I test it?
*If you don't know a safe value for this parameter, just set it to the
maximal number that is allowed in GPUMD: 1024. This is a quite safe number.
When you simulate a very large system, say, a system with one million atoms
or more, this number is crucial for determining the memory consumption of
the code. In short: if this number is too large, it might waste memory; if
this number is too small, it might lead to unexpected errors. *
2. cutoff
In the manual, it means *initial cutoff distance used for building the
neighbor list *
and said "if you really don't know how to choose the value of "cutoff", you
can simply set it to cutoff distance of the maximal force cutoff in your
system plus 1.0 A."
Is it mean that if I use lj potential for graphene-graphene system of which
cutoff distance is about 9.0 A, I can set the "cutoff" in xyz.in to 10.0 A?
*Yes, this is a very good choice.*
*Reminder: you may try to enable neighbor list updating to make your
simulation "safer", but one of our recent studies revealed an artifact for
heat transport calculations in these layered systems (with weak LJ
inter-layer interactions) that can be alleviated by disabling the neighbor
list updating! Sounds counterintuitive? You can check this paper: *
*Thermal conductivity reduction in carbon nanotube by fullerene
encapsulation: A molecular dynamics study*
*https://www.sciencedirect.com/science/article/abs/pii/S0008622320301378?via%3Dihub
<https://www.sciencedirect.com/science/article/abs/pii/S0008622320301378?via%3Dihub>*
How can I get the same results with triclinic system?
*There are no special things about triclinic systems compared to
rectangular ones regarding thermal transport calculations. The only caveat
in GPUMD is that one CANNOT use pressure control (i.e., NPT ensemble) when
using a triclinic box. This is a restriction for the time being and we may
try to improve it in the future. *
Zheyong
On Fri, Feb 26, 2021 at 10:19 AM "소순성" <soonsung2001@xxxxxxxxxx> wrote:
Hi, Fan.
Thank you for GPUMD and today, I have questions.
I'd like to HNEMD simulation for graphene bilayer or graphite multilayer.
So I prepared the xyz.in file with triclinic system and
here I have a question when it comes to what I set in the xyz.in file:
1. maximum number of neighbors
In the manual, it means literally *maximum number of neighbors* any atom
can ever has.
As I calculated for graphene monolayer, I set it as 3 because just an atom
meets three adjacent atoms.
However, when I must consider the VdW interaction between each layer, how
can I estimate it? Just Do I test it?
2. cutoff
In the manual, it means *initial cutoff distance used for building the
neighbor list *
and said "if you really don't know how to choose the value of "cutoff",
you can simply set it to cutoff distance of the maximal force cutoff in
your system plus 1.0 A."
Is it mean that if I use lj potential for graphene-graphene system
of which cutoff distance is about 9.0 A, I can set the "cutoff" in xyz.in
to 10.0 A?
How can I get the same results with triclinic system?
P.S I attached my input files for triclinic system.
