problem with NVE simulation

Hi all

I am doing an NVE simulation (with good energy conservation, around 0.01%)
with a silica nanoparticle thermalized at 500 K immersed in a water bath at 310 K.
The silica nanoparticle has 10K atoms and the water bath 100K atoms.
I run the simulation for 600ps, and the temperature
reaches asymptotic behavior after 150ps. In principle both system should reach
the same temperature due to the heat exchanged, but actually there is a difference
of almost 15 K between the final temperature of both system, water bath is at 296 K
and silica nanoparticle is at 311 K.

Why is this happen? How can I solve this problem?

Attached you will find my input file



I don’t see any obvious issue with your input script. Have you visualized
the system to verify something funny is not happening? Is there
any drift of the nanoparticle or water that would add a velocity bias
to the temperature? You could try using compute temp/com for both
silica and water to see if it changes. Since you are dumping ke/atom
you can also verify for yourself that the Temp LAMMPS is calculating
matches the summed KE of the particles.