Question on defining cylinder around a given region

LAMMPS LAMMPS Mailing List Mirror
1

Dear all,

I would need some advices from you. I am trying to reduce the size of my
simulation box in order to reduce the CPU time.

I have a large cubic box with millions of atoms. Now, I am interested only
in some of the atoms of the box (with some particular kinetic energy). For
the sake of simplicity, let's say these atoms are initially located along a
line that would coincide with the 111 diagonal. Clearly, many of the atoms
in the simulation box are not of interest, but are consuming CPU time. They
are just far from these atoms and are thermalised. Therefore, I would like
to delete them in order to reduce their number.
I was thinking of defining a cylinder (for instance) with a given radius
around the line of atoms of interest, using group command. Then I would
remove the atoms outside of the cylinder with delete_atoms command.

And these are my questions:
Is is possible to define such a cylinder inside an existing box (cubic
here) ? A cylinder that would contain a given region.
If so, once the cylinder would be created and the atoms around are deleted,
would it be possible to set the surface of the cylinder as periodic ?

Many thank in advance for any advice !
Christophe

2

Dear all,

I would need some advices from you. I am trying to reduce the size of my
simulation box in order to reduce the CPU time.

I have a large cubic box with millions of atoms. Now, I am interested only
in some of the atoms of the box (with some particular kinetic energy). For
the sake of simplicity, let's say these atoms are initially located along a
line that would coincide with the 111 diagonal. Clearly, many of the atoms
in the simulation box are not of interest, but are consuming CPU time. They

those atoms may not be of interest, but they contribute to the
properties of the overall system.

are just far from these atoms and are thermalised. Therefore, I would like
to delete them in order to reduce their number.
I was thinking of defining a cylinder (for instance) with a given radius
around the line of atoms of interest, using group command. Then I would
remove the atoms outside of the cylinder with delete_atoms command.

And these are my questions:
Is is possible to define such a cylinder inside an existing box (cubic here)
? A cylinder that would contain a given region.

you can define cylindrical regions in LAMMPS.

If so, once the cylinder would be created and the atoms around are deleted,
would it be possible to set the surface of the cylinder as periodic ?

you lost me at this last statement. what do you mean by this?

axel.

3

Dear Axel,

Thank you for your prompt reply.

> Dear all,
>
> I would need some advices from you. I am trying to reduce the size of my
> simulation box in order to reduce the CPU time.
>
> I have a large cubic box with millions of atoms. Now, I am interested
only
> in some of the atoms of the box (with some particular kinetic energy).
For
> the sake of simplicity, let's say these atoms are initially located
along a
> line that would coincide with the 111 diagonal. Clearly, many of the
atoms
> in the simulation box are not of interest, but are consuming CPU time.
They

those atoms may not be of interest, but they contribute to the
properties of the overall system.

Yes, I do agree with you. Indeed, if you reduce too much the box it could
affect the result. But sometimes it is unnecessary to have very large
boxes. Therefore, I would like to reduce the box (up to some limit).

> are just far from these atoms and are thermalised. Therefore, I would
like
> to delete them in order to reduce their number.
> I was thinking of defining a cylinder (for instance) with a given radius
> around the line of atoms of interest, using group command. Then I would
> remove the atoms outside of the cylinder with delete_atoms command.
>
> And these are my questions:
> Is is possible to define such a cylinder inside an existing box (cubic
here)
> ? A cylinder that would contain a given region.

you can define cylindrical regions in LAMMPS.

Ok, I will have a look at the region command.

> If so, once the cylinder would be created and the atoms around are
deleted,
> would it be possible to set the surface of the cylinder as periodic ?

you lost me at this last statement. what do you mean by this?

I start with a cubic box with periodic boundary conditions in xyz. Then,
out of this cubic box, I would like to define a cylinder, and remove the
atoms around. But then I would like the surface of the cylinder to have
similar properties as the cubic box surface, ie periodic. Is this possible
for a cylinder ?

Christophe

4

> If so, once the cylinder would be created and the atoms around are
> deleted,
> would it be possible to set the surface of the cylinder as periodic ?

you lost me at this last statement. what do you mean by this?

I start with a cubic box with periodic boundary conditions in xyz. Then, out
of this cubic box, I would like to define a cylinder, and remove the atoms
around. But then I would like the surface of the cylinder to have similar
properties as the cubic box surface, ie periodic. Is this possible for a
cylinder ?

no. how could this work geometrically? i.e. how would you identify
which atom from inside the cylinder would correspond to a
"periodically replicated" atom outside?

axel.

5

>> > If so, once the cylinder would be created and the atoms around are
>> > deleted,
>> > would it be possible to set the surface of the cylinder as periodic ?
>>
>> you lost me at this last statement. what do you mean by this?
>
>
>
> I start with a cubic box with periodic boundary conditions in xyz. Then,
out
> of this cubic box, I would like to define a cylinder, and remove the
atoms
> around. But then I would like the surface of the cylinder to have similar
> properties as the cubic box surface, ie periodic. Is this possible for a
> cylinder ?

no. how could this work geometrically? i.e. how would you identify
which atom from inside the cylinder would correspond to a
"periodically replicated" atom outside?

Yeah, that's what I thought...
I already found in the documentation how to define a cylinder with region
command. Very convenient. I think it will be useful for what I want to do.

Thanks a lot for your help Axel.
Christophe

6

> Dear all,
>
> I would need some advices from you. I am trying to reduce the size of my
> simulation box in order to reduce the CPU time.
>
> I have a large cubic box with millions of atoms. Now, I am interested
only
> in some of the atoms of the box (with some particular kinetic energy).
For
> the sake of simplicity, let's say these atoms are initially located
along a
> line that would coincide with the 111 diagonal. Clearly, many of the
atoms
> in the simulation box are not of interest, but are consuming CPU time.
They

those atoms may not be of interest, but they contribute to the
properties of the overall system.

> are just far from these atoms and are thermalised. Therefore, I would
like
> to delete them in order to reduce their number.
> I was thinking of defining a cylinder (for instance) with a given radius
> around the line of atoms of interest, using group command. Then I would
> remove the atoms outside of the cylinder with delete_atoms command.
>
> And these are my questions:
> Is is possible to define such a cylinder inside an existing box (cubic
here)
> ? A cylinder that would contain a given region.

you can define cylindrical regions in LAMMPS.

I found in the documentation how to define a cylinder region with region
command. However, what is not clear to is if the cylinder axis is only x, y
or z. Is it possible to define a cylinder with any arbitrary axis, along
the 111 direction for instance ?

7

X, y, and z are just directions and can correspond to any crystallographic orientations. You can rotate your cubic bulk box (by lattice orient command or the rotate keyword of the fix move command) so that the (111) crystallographic plane correspond to one the x, y, or z planes. Then you can use region command to create a cylinder with an axis that lies on [111].

Ray

8

From: Christophe Ortiz <christophe.ortiz@…4080…>
Date: Wednesday, June 8, 2016 at 8:05 AM
To: Axel Kohlmeyer <[email protected]>
Cc: LAMMPS <lammps-users@…431…ceforge.net>
Subject: Re: [lammps-users] Question on defining cylinder around a given region

X, y, and z are just directions and can correspond to any crystallographic orientations. You can rotate your cubic bulk box (by lattice orient command or the rotate keyword of the fix move command) so that the (111) crystallographic plane correspond to one the x, y, or z planes. Then you can use region command to create a cylinder with an axis that lies on [111].

Thank you Ray for the comment. Seems to be a solution. I was also thinking of the rotate keyword of region command when defining the cylinder. I will try both.

Christophe