[lammps-users] pair distance < table inner cutoff

Our units are correct. I’ve reduced the simulation down to two atoms with the non-skewed dimensions of the simulation box set very high. What I am now seeing is that energy goes crazy in the simulations. Here are all the files I am using now:

in.trial:

3-d 2 metal surfaces coming in contact and then sliding

dimension 3
boundary p p p
units metal
atom_style atomic

read_data data

#pair potentials

pair_style table spline 299

pair_coeff 1 1 some_pots NiNi

define groups

group qcfix type 1

fix fix_nve all nve
#fix fix_lang all langevin 0 0 1 239482 axes 0 1 0

Run

timestep 0.0001
thermo 100

dump i all xyz 100 out*.xyz

dump 1 all custom 100 comp.trial tag type x y z fx fy fz
run 3000

data:

Atomic coordinate file for a qc

2 atoms

1 atom types

0.0 100 xlo xhi
0.0 7.1856 ylo yhi
0.0 100 zlo zhi

-2.3347444 0.0 0.0 xy xz yz

Masses

1 58.693

Atoms

1 1 9.97992 0.801383 23.459

2 1 8.54497 5.21772 23.459

and some_pots:

NiNi
N 299

1 1.11127 6.37274e+06 1.47418e+08
2 1.16419 2.37116e+06 5.15007e+07
3 1.21711 921911 1.88778e+07
4 1.27002 373147 7.23101e+06
5 1.32294 156716 2.88218e+06
6 1.37586 68096.7 1.19218e+06
7 1.42878 30535.5 510192
8 1.4817 14098 225321
9 1.53461 6688.03 102473
10 1.58753 3254.03 47873.3
11 1.64045 1621.12 22943.5
12 1.69337 825.686 11260.3
13 1.74628 429.363 5651.16
14 1.7992 227.66 2895.22
15 1.85212 122.933 1513.06
16 1.90504 67.5175 805.689
17 1.95795 37.6621 436.684
18 2.01087 21.3058 240.57
19 2.06379 12.2002 134.687
20 2.11671 7.05057 76.5369
21 2.16963 4.09953 44.0425
22 2.22254 2.38932 25.6034
23 2.27546 1.39006 14.9972
24 2.32838 0.802021 8.82468
25 2.3813 0.456056 5.16479
26 2.43421 0.255404 2.96176
27 2.48713 0.14263 1.62444
28 2.54005 0.0834735 0.808231
29 2.59297 0.0570868 0.307309
30 2.64588 0.0509472 -0.000441937
31 2.6988 0.0571359 -0.185214
32 2.75172 0.0705503 -0.289591
33 2.80464 0.0877862 -0.339689
34 2.85756 0.106503 -0.351968
35 2.91047 0.125035 -0.337024
36 2.96339 0.14217 -0.302277
37 3.01631 0.157028 -0.253193
38 3.06923 0.168968 -0.193994
39 3.12214 0.177559 -0.12841
40 3.17506 0.182557 -0.0595049
41 3.22798 0.183857 0.00996788
.
.
. (and so forth)
288 16.2987 0.00225236 -0.00104649
289 16.3516 0.00226902 0.000414178
290 16.4045 0.00220854 0.0018397
291 16.4574 0.00207438 0.00317344
292 16.5103 0.00187279 0.00436578
293 16.5632 0.00161248 0.00536912
294 16.6162 0.00130411 0.00616602
295 16.6691 0.000959568 0.00673531
296 16.722 0.000591514 0.00705815
297 16.7749 0.000212816 0.00713869
298 16.8278 -0.000163759 0.00698113
299 16.8807 -0.000525788 0.00658841

Here is the output:

Reading data file …
triclinic box = (0 0 0) to (100 7.1856 100) with tilt (-2.33474 0 0)
1 by 1 by 1 processor grid
2 atoms
2 atoms in group qcfix
Setting up run …
Memory usage per processor = 1.43775 Mbytes
Step Temp E_pair E_mol TotEng Press
0 0 1592.2644 0 1592.2644 283701.86
100 456849 -452.83472 0 -393.78235 -88638.495
200 907857.71 -520.78728 0 -403.43746 -100195.76
300 68298.13 1549.3042 0 1558.1325 276547.94
400 749524.18 -509.96087 0 -413.07727 -98579.161
500 455732.79 -461.95887 0 -403.05079 -90344.976
600 176050.02 1535.5116 0 1558.2678 274409.39
700 686560.73 -486.23977 0 -397.49483 -94394.455
800 772098.66 -499.87471 0 -400.07312 - 96718.07
900 188746.51 1529.1259 0 1553.5233 273346.8
1000 483465.29 -468.82317 0 -406.33038 -91568.618
1100 669902.17 -500.76466 0 - 414.17301 -97075.692
1200 114527.28 1539.4554 0 1554.2592 274962.19
1300 907850.02 -519.66491 0 -402.31608 -99998.492
1400 494172.58 -456.41847 0 - 392.54165 -89237.109

Any ideas?

I should also add that the size of the x and z dims of the box don’t have any effect on this energy conservation problem, while increasing the y dimension does, in fact, result in proper energy conservation.

I changed the pair style to linear and now everything is working properly. I am going to look over the spline routine to see what the problem was. I’ll be more than happy to contribute, if there is, in fact, a problem.

You can use the pair_write command to write the eng,force values
for a potential to a file after it has been splined. Then if you
plot it you may see the problem.

Steve