Dear Aidan:
I am very grateful for your writing back to me. I agree with you about the point that different seeds have relations with equilibrium properties. I have no idea how to obtain the ideal result. Could I use 3 or 5 seeds and take the average?

Dear Aidan:
I am very grateful for your writing back to me. I agree with you about the
point that different seeds have relations with equilibrium properties. I
have no idea how to obtain the ideal result. Could I use 3 or 5 seeds and
take the average?

it is difficult to evaluate the situation because there is too little
information.
you provide some graphics, but we don't really know what they depict
and how they came to be.

without this information, some more detail of what exactly you are
looking for and a simple and fast to run example input, it is
difficult to even assess what your problem is.

"I agree with you about the point that different seeds have relations
with equilibrium properties. I have no idea how to obtain the ideal
result. Could I use 3 or 5 seeds and take the average?"

It depends. Many properties are "self-averaging", yielding a narrow
distribution with a sharp peak. In this case, sampling the final
instantaneous value, or the final time-average value, is sufficient.
Other properties show large variation between different independent
simulations, in which case it may be necessary to run many simulations
and report the full distribution.