Greetings. I am having the following trouble and hopefully someone can share his/her experience and comments to this issue.
I have 512 fixed water molecules partitioned on two processors. After doing a fix_deform, a piece of message saying “Shake atoms 619 620 621 missing on proc 0 at step 14251001” appears. However this does not happen when I running my program on a single processor…I suspect that the ghost list is not large enough to encompass all the associating atoms. Here are some simulation details:
model: SPC/E
N = 512
time step: 2 fs
cut-off: 9 A for both LJ and kspace
domain size: around 25 A by 25 A by 25 A
I tried to increase my skin depth (now it is 7 A), but this scenario still occurs…even when the ghost cutoff is greater than half of the simulation box.
Any kind of help/comment is greatly appreciated. Thank you for the attention.
Greetings. I am having the following trouble and hopefully someone can share
his/her experience and comments to this issue.
I have 512 fixed water molecules partitioned on two processors. After doing
a fix_deform, a piece of message saying "Shake atoms 619 620 621 missing on
proc 0 at step 14251001" appears. However this does not happen when I
running my program on a single processor....I suspect that the ghost list is
not large enough to encompass all the associating atoms. Here are some
simulation details:
model: SPC/E
N = 512
time step: 2 fs
cut-off: 9 A for both LJ and kspace
domain size: around 25 A by 25 A by 25 A
I tried to increase my skin depth (now it is 7 A), but this scenario still
occurs....even when the ghost cutoff is greater than half of the simulation
box.
Any kind of help/comment is greatly appreciated. Thank you for the
attention.
this is a bit of a guess, but could it be that you need to use the
remap option to fix deform?