Hello
I am relatively new to LIGGGHTS-flexible fibers, and I don't understand the effect of neighbor settings quite well. My model has particles of diameter 30e-6 m. Simulation box is cubic with edge of 1000 * particle dia.
When I run with following neighbor settings on my PC:, it runs fine
neighbor 0.001 bin
neigh_modify delay 0 one 5000
However on a different PC with almost same configuration, the above settings don't work (shows segmentation fault !!). When I reduce the skin and modify one setting to following, the code runs, however, it shows very different results compared to the above. Particles now are subjected to higher forces and have larger velocities.
neighbor 30e-6 bin
neigh_modify delay 0 one 5000
Requesting to please help me understand the effect of neighbor setting and what should be an appropriate setting, given particle diameter and simulation box size.
Thanks
Shubham
mschramm | Thu, 01/28/2021 - 18:47
skin distance and max neighbors
Hello,
neighbor skin bin
skin is a parameter (greater than zero please...) that is added to the force cutoff distance (for us, this is the order of the particles diameters).
bin is the type of neighbor finding algorithm that is used.
I recommend keeping the skin parameter around the same size as your particle diameters.
neigh_modify delay 0 one 5000
delay 0 -> rebuild as soon as a flag tells us to.
one 5000 -> The maximum number of neighbors an atom can have...
This is where I assume the discrepancy is coming from. In the first case, your bins are HUGE compared to your simulation domain, and will very quickly get to that max number of neighbors.
If two atoms are in "contact" and not part of the neighbor list, then their forces are not calculated.