Problem about EDIP potential

LAMMPS LAMMPS Mailing List Mirror
1

Dear Lammps users,

I want to use EDIP potential to do some simulations about bulk si. It
is very simple model. But the calculated energy using EDIP
potential in LAMMPS is positive. My input file is as follows:

dimension 3
units metal
atom_style atomic
boundary p p p
neighbor 2.0 bin
neigh_modify every 10
read_data data.txt
newton on
mass 1 28.086
pair_style edip
pair_coeff * * Si.edip Si
timestep 0.0001
thermo_style custom step temp etotal vol press pxx pyy pzz lx ly lz xhi yhi zhi
thermo 1000
velocity all create 300 5812775 dist uniform
dump 1 all xyz 1000 dump1.xyz
fix 1 all npt temp 300 300 0.01 x 0 0 0.01 y 0 0 0.01 z 0 0 0.01
couple none drag 0.2
run 10000

results are as follows:
LAMMPS (21 Sep 2012)
Reading data file ...
   orthogonal box = (0 0 0) to (38.4023 37.6264 39.9088)
   2 by 2 by 2 MPI processor grid
   2880 atoms
Setting up run ...
Memory usage per processor = 1.62935 Mbytes
Step Temp TotEng Volume Press Pxx Pyy Pzz Lx Ly Lz Xhi Yhi Zhi
        0 300 -13280.224 57665.741 2291.0518
2268.6894 2301.3285 2303.1375 38.40226 37.626379
39.9088 38.40226 37.626379 39.9088
     1000 230.66921 85.841229 4.5776652e+13 2.0029541e-06
2.002954e-06 2.0029542e-06 2.0029541e-06 35340.545 34992.192
37016.865 17689.474 17514.909 18528.387

So, is there anything wrong with this input file? or if there is
something wrong with the newest EDIP codes?
Thank you very much in advance.

I am looking forward to having your help.

Bests,
Yuhang Jing

2

0 300 -13280.224 57665.741 2291.0518
2268.6894 2301.3285 2303.1375 38.40226 37.626379
39.9088 38.40226 37.626379 39.9088

In this line of output, total energy (including kinetic) is the
3rd column. Why do you say the energy is positive?

Steve

3

Hi Yuhang,

I think you just have a too small pdamp value. On 1000th step, the
volume and pressure blew up, so that it is meaningless to look at the
temperature when dynamics is wrong.

Please see the doc page on how to choose tdamp and pdamp values.

Ray

4

Dear Steve,

Thank you for your reply.
I just ran the same simulation of the same model (bulk silicon) using
the old version of EDIP in Lammps(lammps-21Mar12). The result is good.
So, I think there should be some problems with the newest version of
EDIP in Lammps(lammps-21Sep12).
The result is as follows:

LAMMPS (21 Mar 2012)

dimension 3
units metal
atom_style atomic
boundary p p p

read_data data.txt
  orthogonal box = (0 0 0) to (38.4023 37.6264 39.9088)
  2 by 2 by 2 MPI processor grid
  2880 atoms
newton on

mass 1 28.086
pair_style edip
pair_coeff * * Si.edip Si

timestep 0.0001
thermo_style custom step temp etotal vol press pxx pyy pzz lx ly lz xhi yhi zhi
thermo 1000

velocity all create 300 5812775 dist uniform
dump 1 all xyz 1000 dump1.xyz
fix 1 all npt temp 300 300 0.01 x 0 0 0.01 y 0 0 0.01 z 0 0 0.01
couple none drag 0.2
run 10000
Memory usage per processor = 1.62403 Mbytes
Step Temp TotEng Volume Press Pxx Pyy Pzz Lx Ly Lz Xhi Yhi Zhi
       0 300 -13280.224 57665.741 1844.7425
1816.4729 1866.8848 1850.8699 38.40226 37.626379
39.9088 38.40226 37.626379 39.9088
    1000 203.41694 -13190.993 57763.899 -1259.7896
-1240.0135 -1278.3112 -1261.044 38.419033 37.658499
39.925196 38.410647 37.642439 39.916998
    2000 291.34534 -13167.662 57680.784 845.3026
831.70978 851.89997 852.29805 38.410562 37.618846
39.918574 38.406411 37.622613 39.913687
    3000 306.08822 -13165.607 57716.271 583.27201
576.6317 598.86996 574.31437 38.400576 37.653207
39.91706 38.401418 37.639793 39.91293
    4000 294.89154 -13166.058 57835.579 -1811.9577
-1784.9004 -1845.0892 -1805.8835 38.440532 37.654873
39.956231 38.421396 37.640626 39.932516
    5000 304.72525 -13168.035 57681.012 1180.9126
1159.4317 1184.5043 1198.8018 38.405526 37.641735
39.89969 38.403893 37.634057 39.904245
    6000 297.36381 -13164.89 57696.727 841.59417
841.23381 867.90734 815.64135 38.412756 37.621943
39.924042 38.407508 37.624161 39.916421
    7000 295.92637 -13165.82 57853.844 -1993.7072
-1977.6071 -2024.3153 -1979.1992 38.436713 37.677147
39.949189 38.419487 37.651763 39.928995
    8000 302.82306 -13163.537 57701.592 695.37241
683.09141 700.31771 702.70812 38.409798 37.63017
39.921753 38.406029 37.628274 39.915276
    9000 298.65735 -13167.612 57716.771 412.56778
414.3271 423.23679 400.13945 38.410643 37.64481
39.915846 38.406452 37.635594 39.912323
   10000 298.35608 -13165.972 57775.302 -591.41382
-588.50792 -601.57125 -584.16229 38.424293 37.648805
39.937892 38.413277 37.637592 39.923346
Loop time of 10.6457 on 8 procs for 10000 steps with 2880 atoms

Bests,
Yuhang

5

The old version had a bug, which is now fixed.
As Ray said, you likely have bad dynamics.

Steve