How to solve the problem "ERROR: failed to allocate 8830876800 bytes for array list:three_bodiesapplication"?

Dear lammps developers and users,

I have met some problems when using a reactive force field developed by Duin to simulate a C-Ni sample.
First, the error reminder said “Neighbor list overflow, boost neigh_modify one”, and then I increased the value of parameter “one”,
however another error “failed to allocate 8830876800 bytes for array list:three_bodiesapplication” occoured. I tried to solve this
problem by icreasing the value of the parameter “safezone” and “mincap”, but the problem is still over there.
It is weird that using this potential I simulated graphene and Ni with same or even larger size, and the obove mentioned problems did not happen.
Even more, I simulated other C-Ni samples with similar size but with different configurations, and the the obove mentioned problems also did not happen.
I have carefully checked the C-Ni sample used this time, but no problem was founded. Please give me some instructions, the input script (lammps.in) and the
C-Ni sample file (model.dat) were sended to you as attachments. The version of the lammps used is “lammps-12Dec18”.
Thanks in advance.

从QQ邮箱发来的超大附件


lammps.in (697B, 无限期)
进入下载页面


model.dat (1.12M, 无限期)
进入下载页面

the problem is (like so many problems reported here), a problem of bad input. you haven’t checked it carefully enough.
please see the attached visualization of your input geometry with the simulation box.
as you can see those don’t match and thus through PBC coordinates will be folded back into the box creating close contacts and thus creating all the neighborlist and similar weirdness.

axel.

vmdscene.jpg