GRADIENT -- Illegal Value for the Total Potential Energy

[heman3333]

So I am running the simulation in NVE ensemble. For few picoseconds the simulation is running fine but after that
it gives the warning " Warning, system moved too much since last neighbor list update, try lowering nlupdate VDW ", which
I am not able to understand. Further the simulation got terminated giving the error "GRADIENT -- Illegal Value for the Total Potential Energy". Currently I am stuck here.
I am using Tinker-HP v1.2.

[louislagardere]

That means that there is some instability in the simulation that lead to some movements of atoms that are too large given the frequency of update of the neighbour lists. To circumvent that, you can try reducing the "nlupdate" parameter in the key file (nludapte x), default value is 20 (time steps) for a 2fs simulation. Of course this can also mean that there is a more fondamental issue in your simulation that leads to some instability.

[heman3333]

In fact, I have ran the simulation using the same key file without any error for some aqueous salt solutions.
This time, I have taken a little bit higher concentration and I ended up with this error.
Thanks for the suggestions and I will check nlupdate parameter in my key file.

[heman3333]

Further, The same simulation with the same key file, running fine on other cluster. But giving error on another cluster.
Does it also depend on the computer architecture too ?