[lammps-users] Error in PPPM coulomb style

Hi everyone:
I’m working with lammps 2001 (yes, fortran version) and I’m having a lot of troubles, hope someone can help me. the problem might be in the coulomb style. I’m using PPPM and I got following error. I don’t know what I should do. Any advice would be appreciated. Thanks in advance.

Step Temp E_nbond E_bond E_long E_total Pressure Volume
0 0.753110 13.490228 0.004121 591.941838 606.565576 0.566927 1500000.0000
Out-of-range Atom: 0 -9.16147532503026 -35.2708412504378
12.4769819020427 15 -2 32 -1
-2 6 3 -2 6
ERROR: Cannot compute PPPM

-300.00000 300.00000 xlo xhi
-25.000000 25.000000 ylo yhi
-25.000000 25.000000 zlo zhi

If I increase at the same time X and Y axis to -86.000 86.000 without changing Z , it works well. But I really need the former simulation box.

my input script is:

units		lj
dimension       3
# periodic in al dimmensions
periodicity     0 0 0
              
maximum cutoff	57.00          
neighbor	0.5 1 1 10 1
true flag	3
extra memory    60.0 60.0 60.0 60.0

angle style 	class2

read data       [dna.in](http://dna.in)

bond style	harmonic
bond coeff	1  200.0 1.1 
bond coeff	2  200.0 1.1 
bond coeff	3  200.0 1.1 

nonbond style   lj/cutoff   2.3        1
nonbond coeff   1 1  1.0000000 2.0000000  2.2449240966 
nonbond coeff   1 2  1.0000000 1.5000000  1.6836930724 
nonbond coeff   1 3  1.0000000 1.5000000  1.6836930724 
nonbond coeff   1 4  1.0000000 2.0000000  2.2449240966
nonbond coeff   2 2  1.0000000 1.0000000  1.1224620483
nonbond coeff   2 3  1.0000000 1.0000000  1.1224620483
nonbond coeff   2 4  1.0000000 1.5000000  1.6836930724 
nonbond coeff   3 3  1.0000000 1.0000000  1.1224620483
nonbond coeff   3 4  1.0000000 1.5000000  1.6836930724 
nonbond coeff   4 4  1.0000000 1.5000000  1.6836930724 

coulomb style   pppm 57.00  0.001
dielectric      0.17806267

Hi Paul. It would certainly be easier for us to track down and diagnose what the problem is if you were using the current version of LAMMPS. Also, if you could post the data file that your input script calls for. Your timestep size seems large. Usually when these PPPM errors occur, the real problem has nothing to do with PPPM. Atoms get blown out of the box for other reasons (usually bad initial configuration), and then PPPM is the first part of LAMMPS that complains about it. What you are describing sounds like there are atoms overlapping through the periodic boundary conditions when you use the smaller box size and not when you use the larger box size. One good sanity check is to inspect the initial configuration visually, with the periodic bounday conditions applied for the smaller box size. Or run a script that checks nearest particle-to-particle distance (like distance.py in the Pizza.py toolkit) to make sure you don’t have overlaps. http://www.cs.sandia.gov/~sjplimp/pizza.html

Paul