AW: Atoms moving too fast with NAMD/Amber

From: Norman Geist (norman.geist_at_uni-greifswald.de)
Date: Thu Jul 26 2012 - 01:42:42 CDT

Hi,

I also use the amber ff and the atoms moving too fast error is most of cases due to an error of the box or simulation setup, not the ff itself. I think the most likely case is that you have overlapping or too close atoms. You should check the VDW energy in the minimization if there are 999999 energies. Also, check the settings of the constraints if you use some. I also had this error one time when I had a broken GPU, so if this is a GPU run, try another GPU to figure out.

In general this error must come with high forces, so the first is to check the energies if their height fits to the systems size.

Norman Geist.

> -----Urspr√ľngliche Nachricht-----
> Von: owner-namd-l_at_ks.uiuc.edu [mailto:owner-namd-l_at_ks.uiuc.edu] Im
> Auftrag von Francesco Pietra
> Gesendet: Donnerstag, 26. Juli 2012 07:49
> An: NAMD
> Betreff: namd-l: Atoms moving too fast with NAMD/Amber
>
> In connection with the suggestions below, is any general protocol to
> follow when atoms move too fast with NAMD-CUDA when using Amber parm7
> ff (prmtop/inpcrd)? It is implied that I am facing this problem after
> succesful minimization and extreme care in settings for heating.
>
> thanks
> francesco pietra
>
> On Thursday 06 May 2010 12:32:13 ipsita basu wrote:
> > Thanks a lot for your reply. I tried timestep 1 before your reply and
> > the job was running fine without giving any error. But the structure
> > become distorted, it squeezes along z axis. Now according to your
> > reply I am fix the langevin temp to zero and heating the system
> > gradually by loop. This part is given below:
> > # gradual temp increase
> > run 1000
> > for { set TEMP 0 } { $TEMP <= 290 } {incr TEMP 1 } {
> > langevinTemp $TEMP
> > run 1000
> > I hope this will not disturb the structure.
> > Any suggestion from your end would be appreciated.
> >
>
> Mmmmh, you probably should double check your structure.
> I never ran into problems setting temperature, langevinTemp and
> LangevinPistonTemp to 303 Kelvin from start.
>
> Make sure to minimize your structure long enough.
> At one point it could be necessary to tune the minimization parameters
> (in
> case of bad contacts which are not removed during a normal
> minimization).
> Once I was forced to use CHARMM with the SD minimizer to get my
> structure
> minimized correctly (after this I could run the simulation in NAMD).
>
> To preserve the structure of you protein you should apply constraints
> to the
> protein backbone (C-Alphas, ~10 kcal/(mol*Angs^2)) and reduce them step
> by
> step. Something like this:
>
> ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
> constraints on
> consref cons.pdb
> conskfile cons.pdb
> conskcol B
> constraintScaling 1.0
>
> # Minimization
> minimize 2000
> reinitvels $temp
>
> run 20000
> constraintScaling 0.5

This archive was generated by hypermail 2.1.6 : Tue Dec 31 2013 - 23:22:18 CST