On Thu, Jan 23, 2014 at 1:56 AM, Saeed Nasiri <s.t.nasiri.gmail.com> wrote:
> Dear all
>
> I see an unusual thing in the simulation output file. the simulation time
> doesn't increase monotonically and in some points time was decrease . I
> restart the simulation for the second times. Is there a problem with the
> simulation?
>
I believe you are misusing MPI. Or perhaps you specified the same output
file for multiple simulations. See my comments below
>
> Best regard
> Nasiri
>
> ===============================================================
> cat mdin.in
>
> equilibration for box
> &cntrl
> imin=0, ntpr=5000, ntwx=5000,
> ntx=5, irest=1,
> tempi=298., temp0=298., ntt=3, cut=12,
> gamma_ln=5., ntb=2, ntp=1, taup=1.0, ioutfm=1, nstlim=5000000,
> dt=.001, ig=-1,
> &end
>
> ===============================================================
> tail -f md2.out
>
> NSTEP = 3685000 TIME(PS) = 7154.000 TEMP(K) = 296.60 PRESS =
> -328.0
> Etot = 75.3663 EKtot = 1996.2872 EPtot =
> -1920.9210
> BOND = 685.6971 ANGLE = 874.3359 DIHED =
> 173.2846
> 1-4 NB = 29.6882 1-4 EEL = 942.3203 VDWAALS =
> -889.6112
> EELEC = -3736.6358 EHBOND = 0.0000 RESTRAINT =
> 0.0000
> EKCMT = 276.3988 VIRIAL = 445.3237 VOLUME =
> 23850.7575
> Density =
> 0.9625
> Ewald error estimate: 0.2486E-05
>
> ------------------------------------------------------------------------------
>
> tail: md2.out: file truncated
>
This means that a new file was opened with the name "md2.out", and the
file was erased up to this point. After this point, another simulation is
writing this output.
NSTEP = 2620000 TIME(PS) = 6089.000 TEMP(K) = 298.70 PRESS =
> 87.8
> Etot = 26.8390 EKtot = 2010.4689 EPtot =
> -1983.6300
> BOND = 657.7941 ANGLE = 864.5939 DIHED =
> 164.2531
> 1-4 NB = 34.7123 1-4 EEL = 943.6344 VDWAALS =
> -831.5673
> EELEC = -3817.0505 EHBOND = 0.0000 RESTRAINT =
> 0.0000
> EKCMT = 274.2455 VIRIAL = 229.0013 VOLUME =
> 23863.1407
> Density =
> 0.9620
> Ewald error estimate: 0.1458E-03
>
> ------------------------------------------------------------------------------
>
This output does not come from the same thread or sander/pmemd job that
created the first chunk of output. This can happen if you misuse MPI. For
example:
mpirun -np X sander -O -i ...
This is running X threads of the serial sander, each of which will run
independently on a single processor and write to exactly the same files.
You need to use "sander.MPI" to run in parallel. (same with 'pmemd' and
'pmemd.MPI')
Another possibility is that you are not using the mpirun from the same MPI
that you used to compile Amber in the first place. More information about
exactly how you ran the simulation (the scripts you used, for instance)
would be necessary to help fully diagnose. Hopefully the guesses I've
provided are of help.
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 Thu Jan 23 2014 - 04:00:04 PST
