Re: Rattle algorithm failure while restarting metadynamics simulation

From: Giacomo Fiorin (giacomo.fiorin_at_gmail.com)
Date: Fri Oct 29 2021 - 10:01:01 CDT

Hello Monika,

On Fri, Oct 29, 2021 at 10:43 AM Monika Kumari <
Monika.Kumari_at_chemistry.iitd.ac.in> wrote:

> Dear all,
> I am running a well-tempered metadynamics simulation with 2 fs timestep
> using NAMD which is running fine but when I restart the run, it gives the
> rattle algorithm error. What can the possible things I might be
> overlooking. Please suggest.
>

If the instability is only at the restart, can you confirm that you are
also loading the correct .colvars.state file alongside the correct NAMD
coor/vel/xsc files?
https://urldefense.com/v3/__https://colvars.github.io/colvars-refman-namd/colvars-refman-namd.html*sec:colvars_input__;Iw!!DZ3fjg!qfab6XpgNYgTnJ0ARPo_iMUrteSFBGkrjXHO1fvhMZYiWjaaW40sFoUySf5pztCvKA$
(If you used a NAMD patched with Plumed, you can follow the corresponding
instructions for restarting).

> Though after reducing the time step, the error disappears.
>
> Also, when I ran two simulations parallelly with exactly same files, the
> colvar trajectory was different. Being molecular dynamics a deterministic
> simulation, how could it happen?
>

This is false. Even with a completely deterministic set of options, the
limited precision of floating-point math adds tiny errors that are
completely harmless for the sake of physical accuracy but lead over time to
diverging trajectories. And if you use a non-deterministic thermostat or
barostat (very common in NAMD) determinism may go out the window from the
very first step.

Giacomo

>
> Thanks
> Monika
>

This archive was generated by hypermail 2.1.6 : Fri Dec 31 2021 - 23:17:12 CST