Cooling glass melt with Reax - Temp randomly increases to 858000K

Dear Axel,

Thank you for taking time to reply. The timestep is 0.5fms. Is it too quick for melting? J.C.Fogarty’s work on “A reactive MD of the silica-water interface,” used same timestep and npt ensemble. Thank you.

Ruhil Dongol

The time step that is appropriate may vary depending on the temperature, force field parameters, other fixes that modify the system, etc… The only way to verify you have an accurate time step is to run your simulation with using several time steps and compare the results. Eventually further decreases in the time step size will not effect the results of the simulation. There’s no way to tell just by looking if a time step size is appropriate.

It is especially important to carefully chose a time step when using ReaxFF since the errors may be subtle.


Thank you Ben.


I also suggest you visualize your system with lots of snapshots
around the time that the “blow-up” takes place. Likewise, print
out lots of thermo values in that time window.


Thank you Steve for your suggestions.

Ruhil Dongol