Bug in compute group/group with increasing number of atoms ?

Dear LAMMPS users,

I am interested in calculating the force between the indenter tip and the substrate. I have used lammps_1Nov2013 to obtain the force_displacement curves in my system. However, when I increased the number of atoms, the curves is more fluctuant. I am unsure whether my approach in using compute group/group is the correct way or not ?
In my example system, I have two block copper, one is indenter and one is substrate. There are the LJ potential between two block copper.
You can find the input script in the attached file.

Your response is greatly appreciated.
Regards,

Nghia

MD_v5a.in (2.46 KB)

MD_v5b.in (2.47 KB)

log.MD_v5a.lammps (140 KB)

log.MD_v5b.lammps (140 KB)

MD_v5a.jpg

MD_v5b.jpg

I doubt there is any issue with compute group/group as
a function of system size. You should be able to verify

its result by simply summing the total force of the substrate

(or indenter) atoms.

Steve

Dear LAMMPS users,

I am interested in calculating the force between the indenter tip and the
substrate. I have used lammps_1Nov2013 to obtain the force_displacement
curves in my system. However, when I increased the number of atoms, the
curves is more fluctuant. I am unsure whether my approach in using compute
group/group is the correct way or not ?

for your specific setup, wouldn't it be simpler to use compute/reduce
to determine the net force on the indenter?

axel.

Does your line
dump 1 all xyz 1000 MD_v5b.xyz
show what you expect to see?

Dear Nigel,

The dump file MD_v5b.xyz show what I expect to see.

Thank you,

Nghia

Vào ngày 8:15 Thứ Ba, 7 tháng 1 2014, Nigel <[email protected]…2627…> đã viết:

Does your line
dump 1 all xyz 1000 MD_v5b.xyz
show what you expect to see?

Dear Axel,

I used the command
compute force indenter property/atom fy
compute fatoms indenter reduce sum c_force
And I found that the result is the same with the case I used the “compute group/group”, you can see the data in attached file.
So now I don’t know why the force-displacement curves is more fluctuant with increasing number of atoms.

Thank you,

Vào ngày 1:52 Thứ Ba, 7 tháng 1 2014, Axel Kohlmeyer [email protected] đã viết:

Dear LAMMPS users,

I am interested in calculating the force between the indenter tip and the
substrate. I have used lammps_1Nov2013 to obtain the force_displacement
curves in my system. However, when I increased the number of atoms, the
curves is more fluctuant. I am unsure whether my approach in using compute
group/group is the correct way or not ?

for your specific setup, wouldn’t it be simpler to use compute/reduce
to determine the net force on the indenter?

axel.

log.MD_v5c.lammps (160 KB)

So now I don’t know why the force-displacement curves is more fluctuant with increasing number of atoms.