[lammps-users] about cutoff of short-range interact

Thanks for your reply.

Just as Joanne said, this should be satisfied in conventional MD code. How does it avoid the error in LAMMPS ?

Thanks.

2008/10/9 Adam Crusse <[email protected]…24…>

Not clear on your Q. How is what error avoided?

Steve

if the cut-off is larger than L/2, the periodic boundary conditions will count some interactions twice because the cut-off will include two images of neighbors that are nearly L/2 apart. Thence some error would be included.

2008/10/9 Steve Plimpton <[email protected]>

I wouldn't say that's an error. I'd say that's what a model
with PBC and a long cutoff is supposed to do. If you specify
a cutoff > L, then an atom will interact with its own image.
If your cutoff is 10L then an atom will interact with 1000s of
its own image.

Steve

Steve is right that it’s not an error if that’s how the model was parameterized. However, some models, including many standard Lennard-Jones liquids, are not parameterized with the idea of interacting with multiple images and may give spurious results if the cut-off is large enough to include multiple images. That’s not a LAMMPS problem; that’s a matter of setting up a system that is consistent with the model parameterization.

Joanne

you are right.
Could you please explain how a large cut off
influences the Ewald sum, such as when it > L or >L/2 ?

AC.

2008/10/9 Budzien, Joanne Louise <[email protected]…3…>

No. :slight_smile:

The effect of cut-off on Ewald sum is system-dependent. Read page 159 in Allen and Tildesley and all the pages around it that are necessary for you to answer your own question.

If you want to know about the LAMMPS commands kspace_style ewald/n and pair_style lj/coul, read the paper In’t Veld, Ismail, Grest “Application of Ewald summations to long-range dispersion forces” J. Chem. Phys. 127, 144711 (2007).

Hi,
I mean the regular 3D case,

If I am using pair_style lj/cut/coul/long
for a liqud water system.my cubic box has a L = 30
suppose here is the setting,

pair_style lj/cut/coul/long 10 20
kspace ewald

does the long-cutoff induce some error ?
Thanks.
AC.

2008/10/15 Budzien, Joanne Louise <[email protected]…3…>

All simulations contain some numerical error. For your particular case, I don’t know how big the error will be. Your question cannot be answered in general because it is force field specific. You need to read the literature on the particular force field that you are using and do some experimentation to determine the acceptable cut-off.

Joanne

hi all,

I think Ewald sum allows any value of cuoff (<=L). Right ?

Liu

2008/10/16 Budzien, Joanne Louise <[email protected]…3…>

Hi Liu,

Yes, but permissible and optimal are two different things. Failing to crash is not the same as giving reasonably accurate answers with high calculation efficiency.

Joanne

Hi,

I want to fix the positions of a group of atoms, i.e., the coordinates of these atoms don't change at all in the simulation. Can anyone tell me which command in LAMMPS should I use? Thanks in advance.

Glacier

Hi Glacier,

You need to do three things to achieve this.
1. Use the "group" command to group the atoms.
2. Use the "velocity" command to set the group of atoms to have an initial
velocity of zero
3. Assign a "fix setforce" of zero to all components(fx,fy,fz) to the group.

Jan-Michael

Hi, all,

Firstly thank Jan-Michael for the advice. I set up a simulation like he suggested (please see below the input script). However, I found that this strategy only works for NVT ensemble. Under NPT ensemble, as the unit box expands or shrinks, the coordinates (x, y, z) of the "frozen" atoms also expand or shrink with the same ratio. So, the chosen atoms are not really frozen.

Any comments or suggestion?

Glacier

Glacier,

You need to put "dilate partial" at the end of your npt fix command. This will keep the system from scaling the coordinates of your frozen atoms

Chris.

Glacier,

See the "dilate partial" option for fix NPT:

http://lammps.sandia.gov/doc/fix_npt.html

Paul

From: Iceriver Forever [mailto:[email protected]…8…]
Sent: Thursday, October 16, 2008 3:20 PM
To: Jan-Michael Carrillo
Cc: [email protected]
Subject: Re: [lammps-users] fix the positions of a group of atoms

Hi, all,

Firstly thank Jan-Michael for the advice. I set up a
simulation like he suggested (please see below the input
script). However, I found that this strategy only works for
NVT ensemble. Under NPT ensemble, as the unit box expands or
shrinks, the coordinates (x, y, z) of the "frozen" atoms also
expand or shrink with the same ratio. So, the chosen atoms
are not really frozen.

Any comments or suggestion?

Glacier

-----------------------------------
units real
dimension 3
atom_style molecular
read_data DATA.FILE
include PARM.FILE
neighbor 3.0 bin
neigh_modify delay 1
timestep 0.1
group tube1 molecule 2
group rest subtract all tube1
neigh_modify exclude molecule tube1
delete_bonds tube1 multi remove
velocity tube1 set 0.0 0.0 0.0 sum no units box
#fix 1 rest nvt 298.0 298.0 100.0
fix 1 rest npt 298.0 298.0 100.0 xyz 5.0 5.0
1000.0 drag 0.01
fix 3 tube1 setforce 0.0 0.0 0.0
dump 1 all xtc 1000 mono.xtc
thermo_style multi
thermo 500
restart 1000 restart.a restart.b
run 2000

----------------------------------------
> From: [email protected]
> Date: Thu, 16 Oct 2008 11:01:30 -0400
> Subject: Re: [lammps-users] fix the positions of a group of atoms
> To: [email protected]
> CC: [email protected]
>
> Hi Glacier,
>
> You need to do three things to achieve this.
> 1. Use the "group" command to group the atoms.
> 2. Use the "velocity" command to set the group of atoms to have an
> initial velocity of zero 3. Assign a "fix setforce" of zero to all
> components(fx,fy,fz) to the group.
>
> Jan-Michael
>
> >
> >
> >
> > Hi,
> >
> > I want to fix the positions of a group of atoms, i.e., the
> > coordinates of these atoms don't change at all in the simulation.
> > Can anyone tell me which command in LAMMPS should I use?
Thanks in advance.
> >
> > Glacier
> > _________________________________________________________________
> > When your life is on the go-take your life with you.
> > http://clk.atdmt.com/MRT/go/115298558/direct/01/
> >
--------------------------------------------------------------------
> > ----- This SF.Net email is sponsored by the Moblin Your Move
> > Developer's challenge Build the coolest Linux based applications
> > with Moblin SDK & win great prizes Grand prize is a trip
for two to
> > an Open Source event anywhere in the world
> > http://moblin-contest.org/redirect.php?banner_id=100&url=/
> > _______________________________________________
> > lammps-users mailing list
> > [email protected]
> > https://lists.sourceforge.net/lists/listinfo/lammps-users
> >
>

_________________________________________________________________
Stay organized with simple drag and drop from Windows Live Hotmail.
http://windowslive.com/Explore/hotmail?ocid=TXT_TAGLM_WL_hotma

il_102008

To first order, the total energy of the system is
independent of what Coulomb cutoff you choose for Ewald (or PPPM).
All it does is shift part of the computation from the short-range
pairwise to the long-range or vice versa. The sum of the
2 contributions should stay the same, to roughly the specified
long-range tolerance.

However there can be a big effect on computational efficiency if
you choose a too-short or too-long cutoff. 8-10 Angs is usually good.

Steve