[lammps-users] problem with fix rigid

Dear All

I having some difficulties with running a fix rigid calculation.

I am using a serial 6 Oct 2008 copy of lammps.

I am getting the “ERROR: Out of range atoms - cannot compute PPPM” error

According to the Erroors page on lammps this could either be caused by particles going to another
processor or bad physics. I have changed neighbor_modify

Initial coordinates for this calculation comes from a previous MD calculation that ran with the same
parameters (athough this was not a rigid body calculation, and not using lammps ).

My input and data files are attached

Any help is appreciated



last few steps

---------------- Step 17 ----- CPU = 18.3319 (sec) ----------------
TotEng = 138861.5861 KinEng = 1884.3836 Temp = 300.0340
PotEng = 136977.2026 E_bond = 158642.9664 E_angle = 2447.8646
E_dihed = 43.8213 E_impro = 0.0000 E_vdwl = -21065.2185
E_coul = 21383.3297 E_long = -24475.5607 Press = -190305.4470
---------------- Step 18 ----- CPU = 19.3969 (sec) ----------------
TotEng = 127994.4149 KinEng = 1884.3961 Temp = 300.0360
PotEng = 126110.0187 E_bond = 158642.9664 E_angle = 2447.8646
E_dihed = 43.8213 E_impro = 0.0000 E_vdwl = -31847.3942
E_coul = 21297.8967 E_long = -24475.1360 Press = -244443.1082
---------------- Step 19 ----- CPU = 20.8057 (sec) ----------------
TotEng = -26994828421.2458 KinEng = 1808.8032 Temp = 288.0000
PotEng = -26994830230.0490 E_bond = 158642.9664 E_angle = 2447.8646
E_dihed = 43.8213 E_impro = 0.0000 E_vdwl = -26994988393.4376
E_coul = 21503.4450 E_long = -24474.7086 Press = -37808702170.6445
ERROR: Out of range atoms - cannot compute PPPM

data.rigid (322 KB)

input (495 Bytes)

Your system blew up on step 19 - I would try to figure out why.
Visualizing it might help.



One of the things you need to be careful with in using fix rigid is to make sure all parts of the rigid bodies are in the same central image of the box in the initial configuration. Since you are treating all your waters as rigid bodies, it's likely that some of these have an atom over in another periodic image of the box, so that lammps doesn't know how to put the water molecule together and you get this error (at least, that's my best guess as to what's going on).

It's OK if the rigid body moves across a boundary as the simulation runs, but they all have to start out in the same box, or you have to indicate which image each atom is in in the data file (there's a flag to set to do this). Or, you can use shake to keep the waters contrained, and fix rigid on the rest... shake doesn't have this issue of box boundaries. That's what I've been doing with a solvated rigid body and it seems to be working well.

Hope this helps,

Thank you Steve, Christopher

I have visualized the out put like you suggested but I cannot immediately see what is wrong
Following Christophers suggestion, I held the water molecules using shake and made just
the sulute as rigid bodies. I get a longer run by doing this before it blows up again. My
input and data files are attached.

Thanks again for your help.


input (949 Bytes)

system.inp (443 KB)

I don't undertstand these commands:

fix 1 all shake 0.0001 50 10 b 7 a 8
fix 2 all temp/rescale 1 300.0 310.0 10.0 1.0
fix 3 all rigid group kapkes template

Are you appliying SHAKE to rigid bodies? That would be bad.
You do one or the other. Typically I would apply shake to
the waters and time integrate them with fix nve. Then apply
fix rigid to the solute which does the time integration as part of it.

And I would run intially without a thermostat to see if the
system is reasonably stable.


Thanks Steve, Christopher

Like Christopher pointed out their was indeed the mistake with the LJ values. I wrote them from data on another file
and apperently I forgot to remove the negative signs. The mix up with fix and shake is from jutsling things, things are going
pretty well now. Thanks again for your assistance.