发送自 Windows 10 版邮件应用
Dear lammps users,
Dear lammps users,
I wanted to do crack simulation for Al2O3, and nonperiodic boundary
conditions are set for x and y direction, but ewald summation cannot be used
with nonperiodic boundaries, can you give me any suggestions to solve the
problem?
either use periodic boundaries or don't use ewald summation.
axel.
Note that you can still simulate a non-periodic structure in a periodic simulation box. The physical interpretation is an infinite periodic array of finite structures separated by vacuum. The larger the vacuum gap, the more closely it approximates the limit of an isolated structure, but also the larger the vacuum gap, the more kspace points are needed to achieve a desired accuracy level. The kspace_modify slab keyword can help with this, although I am not sure if it works correctly for two non-periodic directions (rod rather than slab geometry). In any case, you should test your simulation method carefully to ensure that it is free of finite-size artifacts.
Note that you can still simulate a non-periodic structure in a periodic
simulation box. The physical interpretation is an infinite periodic array
of finite structures separated by vacuum. The larger the vacuum gap, the
more closely it approximates the limit of an isolated structure, but also
the larger the vacuum gap, the more kspace points are needed to achieve a
desired accuracy level. The kspace_modify slab keyword can help with this,
the slab correction is only available for z direction, so it cannot be
used for two non-periodic directions.
although I am not sure if it works correctly for two non-periodic directions
(rod rather than slab geometry). In any case, you should test your
simulation method carefully to ensure that it is free of finite-size
artifacts.
if the rod is not too wide, using a long cutoff can be very
competitive if not superior in terms of performance and accuracy
compared to the coul/long with kspace solver setup, as interactions in
the non-periodic directions would be computed completely, the
neighborlists are still relatively short and you would not need to
compute kspace contributions for vacuum areas.
axel.
Sorry for not replying to your emails timely, I am still testing for an appropriate simulation method for my problem.
Thank you for your helpful advise.
Best regards,
FU
发送自 Windows 10 版邮件应用