simulation error depending on units

LAMMPS LAMMPS Mailing List Mirror
1

Dear Lammps users,

Lammps version: 10Dec12

I cannot update the latest version because simulations run in TACC cluster (Stampede).

I am simulating a cured (crosslinked) polymer structure. First, I need to relax the created (fully crosslinked) structure with NVE and NPT before getting some properties.

The data file does not contain Velocity section after Atoms. It contains “atoms, bonds, …, atom types, … Masses, … Dihedral Coeffs, …Atoms, … Dihedrals”.

When I set “units with real”, NVE and NPT dynamics have no problem with 1 fs time step.

However, when I set “units with metal”, NPT dynamic has a problem with 0.001 ps (=1 fs) time step. (note that no problem with NVE)

All settings are same but only different is the time step because time unit for “metal” is peco second.

I got the following common error message. "Out of range atoms - cannot compute PPPM (pppm.cpp 1715)

I thought that if I set a correct time step for both units (real: 1 fs, metal: 0.001 ps), both simulations might run.

Is there any idea why the second simulation (with metal units) does not work?

Thanks.

Best regards
Changwoon Jang

2

Changing units can affect the coefficients of your FF as well not just
the time units. Some of the paramterizations provided with certain
pair styles only work in a certain unit system. have you checked this
is not your problem?
Carlos

3

Dear Lammps users,

Lammps version: 10Dec12

I cannot update the latest version because simulations run in TACC cluster
(Stampede).

sure you can update. i do it all the time. there is no difference
between compiling lammps on a local or a remote machine.

I am simulating a cured (crosslinked) polymer structure. First, I need to
relax the created (fully crosslinked) structure with NVE and NPT before
getting some properties.

The data file does not contain Velocity section after Atoms. It contains
"atoms, bonds, ..., atom types, ... Masses, ... Dihedral Coeffs, ...Atoms,
... Dihedrals".

When I set "units with real", NVE and NPT dynamics have no problem with 1 fs
time step.

However, when I set "units with metal", NPT dynamic has a problem with 0.001
ps (=1 fs) time step. (note that no problem with NVE)

All settings are same but only different is the time step because time unit
for "metal" is peco second.

I got the following common error message. "Out of range atoms - cannot
compute PPPM (pppm.cpp 1715)

I thought that if I set a correct time step for both units (real: 1 fs,
metal: 0.001 ps), both simulations might run.

Is there any idea why the second simulation (with metal units) does not
work?

please note that also the time constant of the temperature and
pressure control has the unit of a time and thus changes when
switching from real to metal.

furthermore your energy units are different, too, so your force
constants are different, as well.

axel.

4

Did you change tdamp and pdamp accordingly?

Ray

5

Dear All,

I understood that. The energy unit for “real” is kcal/mol and the one for “metal” is eV. (1kcal/mol = 0.043 eV)

Therefore, I need to change the well depth of pair coeff and the force constant for FF coefficient of bond, angle, dihedral in order to convert “unit” in input file.

Thanks.

Best regard,
Changwoon Jang