On Fri, Feb 5, 2016 at 11:10 AM, Batuhan Kav <bkav13.ku.edu.tr> wrote:
> Dear All,
>
> I have a system of two trisaccharides in explicit water box. I run my
> simulations in NPT with pmemd.cuda, however info files always output to
> zero pressure. I use MC barostat for pressure control.
>
> I run minimization, heating and a short equilibration runs with sander,
> then I proceed to use pmemd.cuda for the production runs. Until I switch to
> cuda implementation, info files output nonzero pressure values, however
> with cuda I always get pressure=0, as below.
>
> NSTEP = 900000 TIME(PS) = 2050.000 TEMP(K) = 300.67 PRESS =
> 0.0
> Etot = -39710.3760 EKtot = 7873.7148 EPtot =
> -47584.0908
> BOND = 51.7412 ANGLE = 78.5973 DIHED =
> 7.8085
> 1-4 NB = 43.2666 1-4 EEL = 1162.9210 VDWAALS =
> 8634.6618
> EELEC = -57563.0872 EHBOND = 0.0000 RESTRAINT =
> 0.0000
> EKCMT = 0.0000 VIRIAL = 0.0000 VOLUME =
> 131390.9400
> Density =
> 0.9999
>
> For instance, while using sander for initial equilibration run, I get
> (however negative)
>
> NSTEP = 100000 TIME(PS) = 250.000 TEMP(K) = 297.05 PRESS =
> -17.2
> Etot = -39803.0263 EKtot = 7779.0082 EPtot =
> -47582.0344
> BOND = 47.0929 ANGLE = 79.1150 DIHED =
> 5.0144
> 1-4 NB = 45.7862 1-4 EEL = 1159.6999 VDWAALS =
> 8796.6486
> EELEC = -57715.3915 EHBOND = 0.0000 RESTRAINT =
> 0.0000
> EKCMT = 3838.4464 VIRIAL = 3887.1530 VOLUME =
> 131307.0964
> Density =
> 1.0005
>
> I checked my input files couple of times and couldn’t catch an
> error/mistake. I just wanted to ask if these nonzero pressures values are
> normal (which I don’t think). Also, what might be the reason that I am
> getting zero pressure with cuda implementation? I attach my input files.
> 03_Prod_ini.in is the input file for short equilibration with sander, and
> 03_Prod.in is the actual production run input.
>
A pressure of -17.2 is not "fake" or incorrect -- it is correctly
calculated from the virial theorem and simply indicates that the box is too
big and that particles are being "sucked in" stronger than being forced
against the external pressure.
Note that the isothermal compressibility for most liquids is incredibly
small, such that very large fluctuations in pressure correspond to very
small fluctuations in volume. So while the volume is fluctuating very
slightly around some average, the pressure is fluctuating a lot more.
Pressures between -1000 and 1000 bar are not unusual.
As for the reporting of 0 for pmemd (and pmemd.cuda), I've added a warning
to the output file for future releases.
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 Feb 05 2016 - 10:30:04 PST
