Re: [AMBER] power failure + resume

From: Jason Swails <jason.swails.gmail.com>
Date: Fri, 6 Nov 2015 12:00:00 -0500

On Fri, Nov 6, 2015 at 8:59 AM, shahab shariati <shahab.shariati.gmail.com>
wrote:

> Dear amber users
>
> I was running the second prod step. Unfortunately, due to power failure,
> sander was stopped.
>
> My prod.in file is as follows:
>
> &cntrl
> imin=0,irest=1,ntx=5,
> nstlim=2500000,dt=0.002,
> cut=10.0,ntb=2,ntp=1,taup=2.0,
> ntc=2,ntf=2,
> ntpr=2500,ntwx=2500,
> ntt=3,gamma_ln=2.0,
> temp0=300.0,
> /
>

​You never set ntwr here, which means it was left at the default value.
That default value varies based on whether you are using sander, pmemd,
pmemd.MPI, and maybe pmemd.cuda. It is almost certainly more frequently
than 2500 steps, so your restart file is unlikely to correspond to the
output at time step 2445000​. It could belong to any step between 2445000
and 2447500.

Furthermore, you have not set ioutfm=1, which means that your simulation is
printing ASCII mdcrd files. These files are notoriously susceptible to
corruption in cases like these, where the writing buffer may not have fully
flushed.

You should *always* set ioutfm=1 in the input file, since the resulting
trajectories are smaller, faster to read, faster to write, and store
coordinates in higher precision. They are also more fault-tolerant.

I have to have a trajectory containing 5000 ps and 1000 frames.
>
> After power failure, the end of the prod2.out file is as follows:
>
> .
> .
> .
> .
>
> ------------------------------------------------------------------------------
>
> NSTEP = 2445000 TIME(PS) = 11390.000 TEMP(K) = 299.92 PRESS =
> -23.9
> Etot = -92384.3446 EKtot = 23793.4048 EPtot =
> -116177.7494
> BOND = 987.7799 ANGLE = 2608.5156 DIHED =
> 3411.8147
> 1-4 NB = 1154.3951 1-4 EEL = 14108.7303 VDWAALS =
> 13059.2879
> EELEC = -151508.2729 EHBOND = 0.0000 RESTRAINT =
> 0.0000
> EKCMT = 10071.6907 VIRIAL = 10270.6229 VOLUME =
> 385008.3181
> Density =
> 1.0277
> Ewald error estimate: 0.1991E-03
>
> ------------------------------------------------------------------------------
>
> To continue, I made new.in file, as follows: (nstlim =
> 2500000-2445000=55000)
>
> prod complex_solvated
> &cntrl
> imin=0,irest=1,ntx=5,
> nstlim=55000,dt=0.002,
> cut=10.0,ntb=2,ntp=1,taup=2.0,
> ntc=2,ntf=2,
> ntpr=2500,ntwx=2500,
> ntt=3,gamma_ln=2.0,
> temp0=300.0,
> /
>
> After completing run, I concatenated 2 mdcrd files using cpptraj.
>
> I saw mixed.mdcrd trajectory using vmd. In vmd, there are 999 frames
> instead of 1000.
>
> Why? Is my manner wrong?
>

​It might be that the mdcrd file buffer was not fully flushed when the
power died, resulting in a partially-truncated mdcrd file. cpptraj
*should* indicate whether or not it was able to predict the number of
frames in the trajectory file. If it wasn't this is a good indication that
your last frame was truncated.

In which case, cpptraj did the best it could, and simply omitted the final,
corrupt/incomplete frame from the first trajectory, which results in one
fewer than you expect. On the whole, this is not a big problem as the
remaining 999 frames should be perfectly valid.

HTH,
Jason​

-- 
Jason M. Swails
BioMaPS,
Rutgers University
Postdoctoral Researcher
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Fri Nov 06 2015 - 09:30:03 PST
Custom Search