[gmx-users] program crashed

[Mark Abraham]

> Mark,
>
> Thank you for your reply. When I checked step38885.pdb, I found a
> polymer chain in my system fall apart (bond broken). If I visualize the
> trajectory. it is fine till step38000 (last frame in my trr file).

OK... bonds don't just "break" in MD simulations. Bonded potentials don't
allow for it, and often bond-length constraint algorithms are being used
anyway. If you're not using such algorithms, it might be helpful to look
at the time evolution of the bond length of the bond that breaks. I have
heard of weird cases where some resonance got established in the system
eventually leading to some catastrophic failure (think Tacoma Narrows
Suspension Bridge!) - if so then the time evolution should show such
behaviour. I can't explain the KE doubling though...

Mark