[AMBER] Coordinate resetting cannot be accomplished, deviation is too large

From: Danny Xu <quantum_mania.yahoo.com>
Date: Wed, 17 Aug 2011 19:13:05 -0700 (PDT)

Hi,

When using pmemd.MPI, I get


     Coordinate resetting cannot be accomplished,
     deviation is too large
     iter_cnt, my_bond_idx, i and j are :       1       1      86     119


whereas serial pmemd works just fine...

Here's my heating.mdin

&cntrl
  imin   = 0,           !no mininisation
  irest  = 0,           !no trajectory restart
  ntx    = 1,           !only coordinates are read in
  dt = 0.002,           !2fs time step
  ntc    = 2,           !SHAKE; bonds involving H are constrainted
  ntf    = 2,           !all H-atom bonds forces are not evaluated
  nsnb   = 25,
  tol = 0.000001,       !Shake tolerance 10 times tighter
  ntb    = 0,           !no periodicity and PME off!
  ntp    = 0,           !No pressure control
  cut    = 9999.0,      !9999A cut off
  igb    = 2,
  rgbmax = 9999.0,      !default max dist between atom pairs Born radii calc.
  saltcon= 0.2,         !salt concentration (M)
  ntt    = 3,           !constant temp using langevin
  tempi  = 0.0,        !initial temp = 0K
  temp0  = 300.0,       !final temp = 300K
  gamma_ln = 1.0,       !1.0 per ps for collision freq
  ntpr = 500,           !print details to log every 500 steps
  ntwx = 500,           !write coordinates to mdcrd every 500 steps
  ntwr = 50000,         !write restart file to heat.rst every 50000 steps
  iwrap  = 0,           !wrap periodic molecules.
  nstlim = 850000,     !run for 850,000 steps
  ntr    = 1,           !restraints
  restraint_wt=2.0,     ! Weight
  restraintmask=".CA,C,O,N",  !Backbone atoms selection
  nscm   = 0,           !No removal of COM motion
  nmropt = 1,           !Scale temp target.
  ig     = -1,          !Random seed based on wallclock.
  ioutfm = 1,           !NetCDF MDCRD.
/
 &wt type='TEMP0', istep1=0,istep2=600000,
   value1=0.0, value2=300.0
/
 &wt type='TEMP0', istep1=600001,istep2=850000,
   value1=300.0, value2=300.0
/
 &wt type='END'
/
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Wed Aug 17 2011 - 19:30:03 PDT
Custom Search