Hi all,
I am using fix gcmc+fix rigid/nvt/small to test the combination of gcmc and rigid, and get error “ERROR on proc 0: Non-numeric atom coords - simulation unstable”. The input has been uploaded. Anyone could give me some suggestions? Thanks.
Dongbo Wang
CH4.txt (580 Bytes)
in.gcmc.ch4 (3.03 KB)
Hi all,
I am using fix gcmc+fix rigid/nvt/small to test the combination of gcmc and rigid, and get error “ERROR on proc 0: Non-numeric atom coords - simulation unstable”. The input has been uploaded. Anyone could give me some suggestions? Thanks.
this would happen if you would insert a molecule in such a way that it would cause close contacts and then overflows energies, which will lead on unexpected behavior when applying the metropolis criterion. This configurations thus need to be avoided. This can be achieved with the “overlap_cutoff” keyword.
For modeling of CH4 I would also seriously consider switching to a united atom model and then avoid all the complications from using molecules and charges and fix rigid.
Please also note, that when inserting a CH4 molecule close to another, you may induce significant rotational energy, which may cause the rigid body integrator to diverge. So you probably also need to reduce the time step. For the near spherical CH4 molecule those rotations have very little friction from adjacent molecules and thus can reach high speeds.