Non-numeric atom coords - simulation unstable (../domain.cpp:548)

LAMMPS
1

Greetings, LAMMPS community,

I’m currently conducting simulations involving dense active Brownian particles within an infinite polymer network. To achieve this, I’ve set the size of the polymer chains equal to the dimensions of the simulation box.

However, I’m encountering the error “Non-numeric atom coords - simulation unstable (…/domain.cpp:548).” This issue arises when the size of the polymer network matches the dimensions of the simulation box. Conversely, if I reduce the size of the network relative to the box, the simulation runs without errors. Unfortunately, this compromises our ability to achieve an infinite chain.
In support of this, I am attaching the input files for reference.
System_SL_3MS_0.5_F120_Prod_hext.in (1.5 KB)

Your insights on resolving this error while maintaining an infinite polymer network would be greatly appreciated.

2

Hi @ramanand_yadav,

This error has been discussed extensively in the forum and previous mailing-list. You might find discussions in the archives that might help you better understand what is going on in your simulation.

Without your system input it is hard to get what is going on but I suspect setting an infinite polymer matching exactly the size of the box may lead to overlapping atoms through boundary condition which is an non-physical starting configuration. But take it as it is, a simple guess.

1 Like
3

Hi Germain,
I’ve reviewed the discussion on the forum and identified a solution for deleting overlapping atoms. However, upon implementing the deletion, we observed that particles at the boundary are being deleted. Consequently, this results in the absence of bonds among particles in periodic images. I’ve attached the system file for your reference. Let me know if you need further assistance.
system.data (251.6 KB)

thanks and regads
Ramanand

4

You need to define the bonds to the deleted atoms as bonds to their periodic images. That will look like you have bonds across the box and LAMMPS will complain about inconsistent image flags, but the bonds will be computed between the closest periodic images and thus the simulation will proceed as it should.
This is a known complication of having “infinite” polymers through periodic boundaries.