Help regarding Random seeed

Respected Sir,
I am using few command involving random seed i.e. to create velocity, to change type of atoms and to deposit atoms as shown below.

velocity all create 300.0 48274 temp new
set group film type/fraction 2 0.50 12453486
fix 3 adatom deposit 5 2 300 29494 region adatom units box

In principal there should be no effect on the final results of no matter what is the value of random seed is. but in my case i have observe notable difference in the results of diffusion process (i.e. diffusion rate and behaviour). dont know i have done somthing wrong or any thing else, can you please help me to figure out,

1-how to select the random number?
2-what parameters may possibly be affected by changing the random number?
3-Any other possible thing which you think may affect my results if not random number?

Any suggestion, correction or help is highly appreciable

thanks and Regards
imran

Respected Sir,
                         I am using few command involving random seed i.e.
to create velocity, to change type of atoms and to deposit atoms as shown
below.

velocity all create 300.0 48274 temp new
set group film type/fraction 2 0.50 12453486
fix 3 adatom deposit 5 2 300 29494 region adatom units box

    In principal there should be no effect on the final results of no matter
what is the value of random seed is. but in my case i have observe notable
difference in the results of diffusion process (i.e. diffusion rate and
behaviour). dont know i have done somthing wrong or any thing else, can you
please help me to figure out,

1-how to select the random number?

by random.

2-what parameters may possibly be affected by changing the random number?

practically everything outside of ensemble averages.

3-Any other possible thing which you think may affect my results if not
random number?

MD is based on what is essentially a potentially chaotic system.
as a consequence, you will see the same kind of behavior (trajectories
will diverge exponentially, near chaotic regions in phase space, you
will see oscillations, the frequency of those oscillations will double
when you get closer to chaos and so on), thus depending on the part of
phase space you are in and the kind of process you are observing, you
need a different amount of samples to get a converged ensemble
average. on top of that, you have to consider finite size effects,
which can suppress certain processes.

in short. there are likely problems with your simulations, that are
less obvious, because they don't lead to an immediate crash, and you
likely need to stick your nose into a book about statistical mechanics
for a while and make some tests to understand why.

in any case, your description is very vague and thus it is not
possible to give any specific advice. you only describe the symptoms
and not getting anywhere near the real problem.

axel.