the NEB implementation in Amber10 is not really designed for large systems.
It will become very inefficient, and also puts limits on some variables that
apparently weren't considered when the codes were written.
Amber11 will have an entirely new NEB implementation that has been tested on
systems of tens of thousands of atoms.
if you send the top few lines of your NEB prmtop I can see if there is
anything that might get around this, but I would guess it's that your system
is just too large for this kind of NEB. Amber11 will not have problems with
it.
On Wed, Mar 3, 2010 at 2:12 PM, Guillermina Estiu <gestiu.nd.edu> wrote:
> Hi,
> I am trying to find the path between two conformations of a same
> molecule (~5000 atoms) using NEB in AMBER10.
> I've prepared the .top and .crd files using addles, following the
> instructions given in the manual. It worked after increasing maxdih, for a
> maximum number of 14 copies.
>
> Again, according to the manual, I used the input file in the tutorial (Ross
> Walker’s) to heat up the system.
>
> &cntrl
> imin = 0, irest = 0,
> ntc=1, ntf=1,
> ntpr=50, ntwx=500,
> ntb = 0, cut = 999.0, rgbmax=999.0,
> igb = 1, saltcon=0.2,
> nstlim = 40000, nscm= 0,
> dt = 0.0005,
> ntt = 3, gamma_ln=1000.0,
> tempi=0, temp0=300,
> ineb = 1,skmin = 10,skmax = 10,
> nmropt=1
> /
> &wt type='TEMP0', istep1=0,istep2=35000,
> value1=0.0, value2=300.0
> /
> &wt type='END'
> /
>
> Running the sander.LES.MPI executable, with my system size, results in the
> error:
>
> LES parameters were found
> EXCEEDED MAXLESADJ!
>
> I found the MAXLESADJ parameter to be set to a value of 3000000. I am not
> sure if I just have to increase the value. I’ve found some discussion on the
> list but not clear answer for this.
>
> I have tried the test case of the tutorial, addles works perfectly but
> sander.LES
> crash with an error: Floating point exception signaled at 7170b9: integer
> divide by zero.
>
> There seems to be two different issues here, one related to the size of the
> system. I have seen several posts like this one in the list but not the
> solution. Any help on this?
>
> Thanks,
> Guille
>
>
> _______________________________________________
> 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 Wed Mar 03 2010 - 13:00:05 PST
