[lammps-users] when simulation, the temperture is "nan". Is there a mistake in my script?

Dear all,
I utilize lammps to simulate heat-transfer process in Crystal. In order to establish temperature gradient, first, I set the whole crystal temperature at 300K, and then I choose a special region and set its temperature at 500K. However, the results is confusing.
Is there anyone encounter the same situation?Or how to solve such problem?
Maosheng

the result:

LAMMPS (14 Oct 2010)
Lattice spacing in x,y,z = 2.87 2.87 2.87
Created orthogonal box = (-14.35 -14.35 -28.7) to (14.35 14.35 28.7)
1 by 1 by 1 processor grid
Created 4000 atoms
100 atoms in group hot
Setting up run …
Memory usage per processor = 2.35413 Mbytes
Step Temp E_pair E_mol TotEng Press
0 299.97314 39331.422 0 39486.482 12468883
100 nan 0 0 nan nan
200 nan 0 0 nan nan
300 nan 0 0 nan nan
400 nan 0 0 nan nan
500 nan 0 0 nan nan
……

the input script:

Initialization

units metal
dimension 3
newton on
boundary p p p
atom_style atomic
neighbor 0.3 bin
neigh_modify check yes
lattice bcc 2.87
region box block -5 5 -5 5 -10 10 units lattice
create_box 1 box
create_atoms 1 box
mass 1 55.85
thermo 100
timestep 0.5
velocity all create 300 123456789 mom yes rot yes dist gaussian units box #300K(T*=T.Kb/epsilon)

EAM potential *********************************************************

pair_style eam
pair_coeff * * Fe_2.eam

sepcial area***********************************************************************************

region hot block INF INF INF INF 9.5 10.0
group hot region hot
compute myTemp all temp/region hot
fix 1 all nvt temp 500.0 500.0 0.5
fix_modify 1 temp myTemp
run 100000
compute 3 all ke/atom
variable temp atom c_3/(1.5*1.0)
fix temp_profile all ave/spatial 1 100000 100000 z lower 0.5 v_temp file temp.profile units lattice
compute 4 all ke/atom
compute 5 all coord/atom 3.0
dump 1 all custom 1 dump.atom id xs ys zs c_3 c_4 c_5
run 600000

2010/11/2 maosheng chai <[email protected]...>:

Dear all,
    I utilize lammps to simulate heat-transfer process in Crystal. In order
to establish temperature gradient, first, I set the whole crystal
temperature at 300K, and then I choose a special region and set its
temperature at 500K. However, the results is confusing.
    Is there anyone encounter the same situation?Or how to solve such
problem?

MD always follows the GI-GO principle.
when you use parameters for your input,
you have to make a _educated_ choice
and not pick some random numbers from
the example sections. remember that the
magnitude of parameters can change a _lot_
for different choices of the units.

e.g, you are using a 500fs time step, but
also a 0.3 angstrom neighbor list "skin".
i don't think that can generate a stable trajectory
in metal units.

axel.

Dear Axel:

Thank you for your letters. I am sorry that I don’t know what the GI-GO principle is. Do you mean that when we choose the applicable parameters for our simulation, we should follow this principle? could you give me more information?
Any correlative literature and suggestions will be highly appreciated. Thank you very much in advance.
Sincerely,
Maosheng

2010/11/4 maosheng chai <[email protected]...>:

Dear Axel:

       Thank you for your letters. I am sorry that I don't know what
the GI-GO principle is. Do you mean that when we choose the applicable
parameters for our simulation, we should follow this principle? could you
give me more information?

GI-GO stands for "garbage in"-"garbage out".

your parameters make no sense and you have
a very high potential energy indicating a bad
input structure.

      Any correlative literature and suggestions will be highly appreciated.

there have been very many discussions about what to do
in a "GI-GO situation" on this mailing list, so have a look at
the mailing list archives. also the manual has some suggestions
for the most common sources of errors.

axel.

The timestep 0.5 ps is too big. Use sth like 0.0005 ps should fix your problem. Remember lammps has different unit system, i.e. Metal, LJ, real … Be careful with this when copy from the files in example dir.

Good luck,

Ajing

2010/11/2 maosheng chai <ms8759144@…24…>