Re: [AMBER] System blows up after several nanoseconds using Amber 12

From: Jason Swails <jason.swails.gmail.com>
Date: Fri, 9 Nov 2012 23:34:26 -0500

Try using a smaller cutoff like 8 or 9. 15 is a very large cutoff for PME calculations and it makes very little difference in the results.

Also, what is the latest bugfix that had been applied (what is the output of ./patch_amber.py --patch-level when you are in AMBERHOME?). And what GPU are you using?

And are you using 1 GPU, or pmemd.cuda.MPI with multiple GPUs?

--
Jason M. Swails
Quantum Theory Project,
University of Florida
Ph.D. Candidate
352-392-4032
On Nov 9, 2012, at 10:19 PM, Sasha Perkins <slp300.psu.edu> wrote:
> Hello,
> 
> I have been running MD simulations using Amber 10 and 11 before, however, I
> recently switched to Amber 12 due to its efficient computational capacity.
> The same systems I have previously ran in Amber are now unexpectedly
> blowing up after several nanoseconds of production while using GPU (CUDA)
> Version of PMEMD version 12.1. Additionally, the system continues to run
> even with the "******" up until the maximum steps I put in the input file
> although the energy values and densities do not make sense. Using VMD,
> nothing visually appears to be out of the ordinary and this did not happen
> using prior versions of Amber. The whole last "bad" restart file contains
> all ****** and thus cannot be used for future simulations. It was suggested
> to me to use iwrap=1, but this still occurs even with iwrap=1.
> Additionally, using the check overlap tool on the trajectory files, no
> atoms were listed to be within the default values so I'm not sure why this
> is still happening. If I restart the simulation from my last good restart
> file, everything runs fine for several more nanoseconds. Energies and
> densities look stable before the "blow up".
> 
> Although the simulation goes back to normal if I use the last good restart
> file of my simulation, I just want to make sure there is not something very
> wrong with my system. Below is my input file as well as a section of the
> output file that contains the "*******" and as mentioned, the simulation
> keeps going until the max number of steps but clearly the energy values and
> density values do not make sense.
> 
> Input :
> 
> &cntrl
> imin = 0, irest=1, ntx=5,
> ntb=2, ntp=1, taup=1.0,
> pres0=0.98,
> ig=-1,
> iwrap=1,
> cut = 15.0,
> ntc=2, ntf=2,
> temp0=298.0,
> ntt=3, gamma_ln=5,
> nstlim= 1000000, dt =0.002,
> ntpr=1000, ntwx= 1000, ntwr = 1000
> &end
> 
> 
> Section of Output:
> 
> NSTEP =    87000   TIME(PS) =   15802.000  TEMP(K) =   331.33  PRESS =
> -277.3
> Etot   =   -137358.4935  EKtot   =      9086.1172  EPtot      =
> -146444.6107
> BOND   =      3244.8075  ANGLE   =      5467.3853  DIHED      =
> 7265.7719
> 1-4 NB =       188.2156  1-4 EEL =   -187788.4759  VDWAALS    =
> -5104.3570
> EELEC  =     30282.0419  EHBOND  =         0.0000  RESTRAINT  =
> 0.0000
> EKCMT  =      1222.6766  VIRIAL  =      1870.1499  VOLUME     =
> 108126.1735
>                                                    Density    =
> 1.1967
> ------------------------------------------------------------------------------
> 
> wrapping first mol.:   758055.52943   122927.92369-12401192.44732
> wrapping first mol.:   758055.52943   122927.92369-12401192.44732
> 
> NSTEP =    88000   TIME(PS) =   15804.000  TEMP(K) =*********  PRESS =
> 6204.7
> Etot   = **************  EKtot   = **************  EPtot      =
> **************
> BOND   =         0.0000  ANGLE   =   1198214.2838  DIHED      =
> 23509.4222
> 1-4 NB =         0.0000  1-4 EEL =        -4.2558  VDWAALS    =
> **************
> EELEC  =  42304898.1122  EHBOND  =         0.0000  RESTRAINT  =
> 0.0000
> EKCMT  =  12582912.0000  VIRIAL  =   5661878.4412  VOLUME     =
> 51661802.8483
>                                                    Density    =
> 0.0025
> ------------------------------------------------------------------------------
> 
> wrapping first mol.:   987146.16498   289271.90065-31488848.29287
> wrapping first mol.:   987146.16498   289271.90065-31488848.29287
> 
> NSTEP =    89000   TIME(PS) =   15806.000  TEMP(K) =*********  PRESS =
> 6174.0
> Etot   = **************  EKtot   = **************  EPtot      =
> 89020168.5599
> BOND   =         0.0000  ANGLE   =   1347576.8634  DIHED      =
> 23526.4752
> 1-4 NB =         0.0000  1-4 EEL =        -5.6433  VDWAALS    =
> 11598652.0389
> EELEC  =  76050418.8256  EHBOND  =         0.0000  RESTRAINT  =
> 0.0000
> EKCMT  =  12582912.0000  VIRIAL  =   -754075.8966  VOLUME     =
> 100049641.6333
>                                                    Density    =
> 0.0013
> ------------------------------------------------------------------------------
> 
> 
> Please let me know if you need any additional information or have an idea
> as to what could be happening, I realize this could be system specific if
> it's something with my parameters, however, I figured it would have
> appeared earlier in the simulation if that was the case. I'm using GAFF
> default values with charges I obtained through Gaussian.
> 
> Thanks for your time and help,
> 
> Sasha Perkins
> Penn State MatSE graduate student
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber

Received on Fri Nov 09 2012 - 21:00:03 PST