回复: some questions about serial and parallel environments

I found that my problem is because of this command

“region del block 49.0 51.0 49.0 51.0 -0.6 0.6
delete_atoms region del”

if i delete this command,the problem will not emerge.
if i add this command and the result will be different.

Tian

if the results is different when i run the simulation under the serial and
parallel environments , and the result under the parallel environment(use
16 cores) is wrong, what is the most possible problem?

the biggest possible problem right now is, that your question is far
too vague and lacking any kinds of meaningful detail to give a useful
answer.

axel.

I found that my problem is because of this command

"region del block 49.0 51.0 49.0 51.0 -0.6 0.6
delete_atoms region del"

if i delete this command,the problem will not emerge.
if i add this command and the result will be different.

check your output and see, if the if the exact same number of atoms
are deleted in the different cases.

axel.

I perform a control-experiment.
In another input,is ok whether i add this command or not.i mean i didnt need to delete any atoms in my experiment.
I just add this command,and this command is delete no atoms.I run two simulations ,one of them add this command and the other is not.
The result shows that the one which add the command goes wrong and the other is consist with my prediction under parallel environments(both are right under serial environments).

So in my initial input i use this command to delete atoms and i think that is why output goes wrong

I perform a control-experiment.
In another input,is ok whether i add this command or not.i mean i didnt need
to delete any atoms in my experiment.
I just add this command,and this command is delete no atoms.I run two
simulations ,one of them add this command and the other is not.
The result shows that the one which add the command goes wrong and the other
is consist with my prediction under parallel environments(both are right
under serial environments).

So in my initial input i use this command to delete atoms and i think that
is why output goes wrong

two things:

1) you have to define "wrong" in this context. you are arguing in a
very non-scientific way here.
2) the only side effect from calling delete_atoms is that you trigger
a neighbor list build. if you have a simulation, where the correctness
of the output (whatever that means, see 1) depends on when neighbor
lists are updated, then you have likely other, very serious problems
in your input.

it is impossible to argue details without any kind of tangible
information let alone proof outside of your say-so.

axel.