Hi Dr. Axel Kohlmeyer,
Thanks your replies.
For useful help, follow is my complete input script, and simulation results. My queation is Why the temperature of bulk_upper rapidly increased?
Like the last letter said, the "compute temp" command cannot work. If you have any good idea, pls tell me.
this is a bit of a complex input and tries to do too many things in
one go, which makes it difficult to debug.
you should approach your simulation in stages. first your should do a
proper equilibration simulation with your nanoparticle at rest. for
this you should remove any input from your input that is not
essential. best start with an empty file and only copy over what
absolutely necessary. if you cannot get the system equilibrated, you
need to figure out why. that should also solve the "temperature
mystery", or make it easier to find a solution. please find a few more
comments in between the lines:
boundary p p p
atom_style hybrid charge atomic
this is redundant and pointless. atom_style charge is a superset of
atomic and thus hybrid is not needed.
communicate multi vel yes
this is not needed for your system as well.
neighbor 0.3 nsq
such a small neighbor skin only works well in solids and with reduced
neigh_modify delay 0 every 1 check yes page 100000 one 10000
neither setting page nor setting one should be needed here.
orthogonal box = (-40 -40 -400) to (40 40 600)
1 by 1 by 1 MPI processor grid
these are a rather small number of atoms for the dimensions of the box.
## Ti type 1, O type 2
pair_style buck/long/coul/long cut long 10.0
pair_coeff 1 1 7.177e+5 0.154 1.210e+2
pair_coeff 1 2 3.911e+5 0.194 2.904e+2
pair_coeff 2 2 2.717e+5 0.234 6.969e+2
pair_modify table 0
why turn off tabulation? you don't use a very accurate kspace solver,
so this will only make you slower, but not more accurate.
kspace_style ewald 1.0e-4
delete_atoms overlap 0.3 all all
why this? you provide initial input from a data file. if you need to
delete atoms, why add them in the first place?
Ewald initialization ...
G vector (1/distance) = 0.244615
estimated absolute RMS force accuracy = 0.0434719
estimated relative force accuracy = 0.000130914
KSpace vectors: actual max1d max3d = 112466 148 13099036
Deleted 0 atoms, new total = 4335
group upper id >= 3970 #upper particle
366 atoms in group upper
region 1 block INF INF INF INF INF -5 units box
region 2 block INF INF INF INF -5 0 units box
region 3 block INF INF INF INF 0 19 units box
group bulk_down region 1
467 atoms in group bulk_down
group bulk_middle region 2
759 atoms in group bulk_middle
group bulk_upper region 3
2743 atoms in group bulk_upper
# computer energy
compute new all temp
compute 1 all pe/atom
compute 2 upper ke
compute 4 all stress/atom pair
compute T_middle bulk_middle temp
compute bulk_upper bulk_upper temp/com
compute T_upper upper temp/com
variable velocity_x equal vcm(upper,x)
variable velocity_y equal vcm(upper,y)
variable velocity_z equal vcm(upper,z)
variable position_x equal xcm(upper,x)
variable position_y equal xcm(upper,y)
variable position_z equal xcm(upper,z)
#initial velocity and temp
velocity upper create 300.0 123456 temp T_upper
velocity bulk_upper create 300.0 123456 temp bulk_upper
velocity bulk_middle create 300 123456 temp T_middle
why so complicated? why not initialize all atoms to 300k and then set
the x/y/z velocity for the "down" group to 0.0/0.0/0.0
fix 1 all nve
fix 2 bulk_down setforce 0.0 0.0 0.0
#fix 3 upper setforce 0.0 0.0 0.0
fix 3 bulk_middle temp/rescale 10 293.0 293.0 0.001 1.0
ouch. temp/rescale is a very bad option to control temperature,
particularly with such a small margin.
fix_modify 3 temp T_middle
#velocity bulk_upper set 0.0 0.0 0.0 units box
velocity upper set 0.0 0.0 -0.002 units box
before moving your nano particle, you need to make certain, that you
can correctly run a static system. it doesn't look like it. rather
that all your problems were hidden by using temp/rescale.