Hi? I have performed NVT simulation at 1500 K using for 500 ps ReaxFF.
To check the temperature during the simulation time, I plotted the temperature profile, and I found something strange about 360 ps.
Temperature suddenly dropped to 0 K at about 360 ps.
I wanted to know why temperature dropped to 0 K, so I plotted energy, and I found potential energy more affect total energy than kinetic energy.
So, I plotted sub-categories energy such as bond energy, atom energy, lone-pair energy, etc.
However, I didn’t found what energy has the greatest effect on the potential energy at 360 ps.
(According to documentation of lammps, emol (molecule energy) and efi (electric field energy) are always 0, so I didn’t write those in the output file.)
Furthermore, summation of sub-categories energy is different to potential energy.
Following figure is the difference between potential energy and summation of sub-categories energy.
Why does the difference between potential energy and summation of sub-categories energy occur?
Hi? I have performed NVT simulation at 1500 K using for 500 ps ReaxFF.
To check the temperature during the simulation time, I plotted the temperature profile, and I found something strange about 360 ps.
Temperature suddenly dropped to 0 K at about 360 ps.
which means, that something very bad must have happened at the point. possibly some memory corruption, or a very close contact of two atoms due to bad simulation settings. difficult to say from the outside what exactly.
you can do three things:
visualize the trajectory and see, if there is anything specific happening around the 360ps mark.
restart your calculation for before the mark where things went bad, and see if there is a recurrence of the unexpected event around the same mark. or perhaps after a certain amount of simulation time
check with your system manager or sysadmin, if there are any indications of hardware issues on the machine this simulation was running on. try to see, if this event is reproducible on the same or a different hardware.
