[lammps-users] Temperature Control

I am attempting to model a metallic system using the MEAM package. When using either NPT or NVT, both systems have a severe lag in response to a prescribed temperature when the temperature is changed (see code below). The temperature at the end of this run is approximately 50°C. The only way I have found to adjust for this lag is to make the temperature damping coefficient very small (<10). This is counter-intuitive as such a low coefficient should cause vast fluctuations (as would be seen for pressure). What could be the cause of this need for a small damping coefficient?

#3D Gold

units metal
boundary p p p

atom_style atomic
lattice fcc 4.0782
region box block -3 3 -3 3 -3 3

create_box 2 box
create_atoms 1 box

pair_style meam
pair_coeff * * library.meam Au Au.meam Au Au

neighbor 0.3 bin
neigh_modify delay 5

#fix 1 all nve
timestep .001

minimize 1.0e-10 1.0e-10 1000 10000
velocity all create 100.00 832 rot no dist gaussian
thermo 100
thermo_style custom step temp etotal press vol

##Move 0
fix fnot all npt 100 100 100 xyz -6000 0 300

dump 1 all atom 500 dump_au_meam_p2.meam
dump 2 all xyz 500 dump_au_meam_xp2.xyz

run 10000

restart 60000 Gold.restart

##Stable 0
timestep .001

unfix fnot
fix f0 all npt 100 100 100 xyz 0 0 300

run 50000

##Move 1
timestep .001

unfix f0
fix f1 all npt 100 200 100 xyz 0 0 300

run 20000

##Stable 1
timestep .001

#minimize 1.0e-6 1.0e-8 1000 10000

unfix f1
fix f2 all npt 200 200 100 xyz 0 0 300
run 40000

##Move 2
timestep .001

unfix f2
fix f3 all npt 200 300 100 xyz 0 0 300

run 20000

##Stable 2
timestep .001

unfix f3
fix f4 all npt 300 300 100 xyz 0 0 300

run 40000

Francis,

The behavior that you are describing does not sound counter-intuitive to me, and that damping coefficient that you are using does not sound excessively small, especially considering the fact that you are likely using metal units with time in ps. Ten ps does not sound “very small” to me, and I wouldn’t expect large fluctuations in T, even with a much smaller damping coefficient. If your timestep is 0.005 ps, a damping coefficient as small as 0.5 ps seems reasonable to me.

Pressure tends to fluctuate much more than temperature in MD simulations.

Paul

Hello all

I am attempting to simulate a piece of bulk Indium-Thallium (InTl) using the meam potentials. In looking at the provided “library.meam” file as well as the documentation online, it seems that the library file can only contain fcc, bcc, sc, hcp, diamond, or dimer. Is this correct? Indium is known to have a fct crystal structure. Is there a way to use fct in the library file? Thank you for your help.

Francis