Unexplainable Glitch in NPT Molecular Dynamics (Possible ATK Numerical Error?)

[AsifShah]

Dear Admin,

I have been running NPT simulations (at 300K, details in attached file below) using Martyna Tobias Klien's calculator. The files are attached below.
However, I see my system suffers from a sudden glitch (shown in the pic below). I re-ran my script again but it still shows the glitch.
This has happened for other temperatures also.

Kindly state the possible reasons for this sudden change in temperature. Although grep -rni warning file.log  shows no error or warning.

(The .log file was very large, so I have grep MD in the attached file. It shows how suddenly instant temperature changes to very high values of 1000K+)

Thanks

[Anders Blom]

Possibly some atoms come very close to each other, if the potential is not good enough at keeping them apart.

[AsifShah]

But I am not using any forcefield potential. It is an ab-initio Molecular Dynamic Run.

What could possibly be wrong?

[Anders Blom]

Ah ok. Maybe some oddity in the Grimme term? Can the spike be correlated to some specific feature in the atom positions (if you watch the movie of the atom trajectory)?

[AsifShah]

I will rerun with other Grimme corrections and see if the spike repeats.