[lammps-users] pppm and npt

Dear Lammps users,

I simulated SPC/E water with pppm in NPT with the last major release (21 May 2008 C++ version).
Temperature and pressure are set to 300K, and 1 atm, respectively. However, when the simulated system
seems to be equilibrated well for over100 ps (timestep = 1fs), there is a problem occuring
“Error: Out of range atoms -cannot comput pppm” as follows:

---------------- Step 100400 ----- CPU = 73790.7981 (sec) ----------------
TotEng = -44757.2104 KinEng = 8704.7139 Temp = 304.2245
PotEng = -53461.9244 E_bond = 0.0000 E_angle = 0.0000
E_dihed = 0.0000 E_impro = 0.0000 E_vdwl = 10448.7418
E_coul = 184310.0114 E_long = -248220.6776 Press = 148.1073
Volume = 145143.9950
---------------- Step 100450 ----- CPU = 73827.3053 (sec) ----------------
TotEng = -44683.6312 KinEng = 8680.9545 Temp = 303.3941
PotEng = -53364.5858 E_bond = 0.0000 E_angle = 0.0000
E_dihed = 0.0000 E_impro = 0.0000 E_vdwl = 10368.2562
E_coul = 184483.8674 E_long = -248216.7094 Press = 83.9992
Volume = 145435.2436
ERROR: Out of range atoms - cannot compute PPPM

I noted that the average thermo properites obtained from the 100ps (with an equilibration of
10 pico seconds ) are pretty much close to water (denstiy:0.986g/cc, T = 300.01K, P = 5.02 atm).
I also found the former respondences about the issue by Steve. However, these situations
where the problem occured seem different from the case presented here, in which I tend to
belive that the simulated system has been equilibrated.

Additionally, Is the PPPM used with NVT more stable than with NPT ?

Attached is my input and data file for 4800 molecules.
Kindly check my input file. Am I missing some command here?

Thanks

in.h2o (866 Bytes)

Dear Lammps users,

dear <insert name here>,

I simulated SPC/E water with pppm in NPT with the last major release
(21 May 2008 C++ version). Temperature and pressure are set to 300K,
and 1 atm, respectively. However, when the simulated system seems to
be equilibrated well for over100 ps (timestep = 1fs), there is a

well, for bulk water i would not call 100ps "well" equilibrated,
depending on what your initial configuration was.

problem occuring "Error: Out of range atoms -cannot comput pppm" as
follows:

---------------- Step 100400 ----- CPU = 73790.7981 (sec) ----------------
TotEng = -44757.2104 KinEng = 8704.7139 Temp = 304.2245
PotEng = -53461.9244 E_bond = 0.0000 E_angle = 0.0000
E_dihed = 0.0000 E_impro = 0.0000 E_vdwl = 10448.7418
E_coul = 184310.0114 E_long = -248220.6776 Press = 148.1073
Volume = 145143.9950
---------------- Step 100450 ----- CPU = 73827.3053 (sec) ----------------
TotEng = -44683.6312 KinEng = 8680.9545 Temp = 303.3941
PotEng = -53364.5858 E_bond = 0.0000 E_angle = 0.0000
E_dihed = 0.0000 E_impro = 0.0000 E_vdwl = 10368.2562
E_coul = 184483.8674 E_long = -248216.7094 Press = 83.9992
Volume = 145435.2436
ERROR: Out of range atoms - cannot compute PPPM

that could mean that you have some atoms moving too fast.
this can be because of high potential energy (but your output does
not show that), or that you are not rebuilding your neighborlists
often enough. try a smaller delay value.

cheers,
   axel.

I noted that the average thermo properites obtained from the 100ps
(with an equilibration of 10 pico seconds ) are pretty much close to
water (denstiy:0.986g/cc, T = 300.01K, P = 5.02 atm).
I also found the former respondences about the issue by Steve.
However, these situations where the problem occured seem different
from the case presented here, in which I tend to belive that the
simulated system has been equilibrated.

Additionally, Is the PPPM used with NVT more stable than with NPT ?

there should not be much of a difference unless you are using a
very small system.

This error occurs b/c atoms move too far out of a processor's domain
for charge to be mapped to the PPPM grid. Try reneiighboring more
often (Axel's suggestion) or using less processors or a finer PPPM grid.

Steve

2008/7/21 徐志军 <[email protected]>: