In my simulation, the workpiece material is Si and the tool’s is C.I divide the workpiece into newton layer, thermostating layer and boudary layer, and the tool is same.I use the tersoff potential between Si atoms, C atoms, Si and C atoms. The following is part of my input file.
pair_coeff * * SiC.tersoff Si Si Si C C C
neighbor 1.5 bin
neighbor_modify delay 5
compute…(computing temperature of different layer)
fix 1 all nve
fix fix_workpiece boundary_workpiece setforce 0.0 0.0 0.0
fix fix_tool boundary_tool setforce 0.0 0.0 0.0
velocity boundary_tool set -2 0 0 units box
fix (rescale the temperature at 293K)
dump cfg_file all cfg 500 sic.*.cfg (id type xs ys zs id type potential kinetic stress…)
I think the velocity that is given to tool boundary make the tool have a center-mass velocity, so the velocity should be removed when the temperature of the tool is computed. I think the compute_temp command isn’t fit to compute the tool temperature and the compute_temp/com command or the compute_temp/partial is better than it. But which is better in the two command.
And the default bin value is 2. For saving computing time, what number value should be selected for the value.
The two days, I found an another qustion. When the timestemp reach to same ten thousand, there is error poped that is
p0_31808:p4_error:interrupt SIGx: 13
p0_31808:(28437.052676) net_send: could not write to fd=4, errno=32.
I searched the error in the net and found the error had been parallel computing error. I do the simulation by the parallel computation. When I output the dumpfile by dump not cfg, there is no error.