Re: [AMBER] error - during minimization in AMBER10

From: Jason Swails <>
Date: Fri, 5 Mar 2010 23:45:12 -0500

On Fri, Mar 5, 2010 at 10:54 PM, kureeckal ramesh
<> wrote:
> Hi AMBER users,
> I was successful in completing the TrpCage tutorial using AMBER 10 (running in cluster) and was also able to complete the MD analysis for the same.
> Now, I have started working with my sample protein ( ab-initio prediction) and was able to complete only upto heating stage. In the production stage (comprising a total of 10 stages with 5 ns / stage; i. e total of 50 ns simulation), the program got "TERMINATED" after the completion of 1st stage. AT the begining of the 2nd stage, following error is reported:
> -----------------------------------------------------------------------------------------------------------
>  mpirun -np 18 /usr/local/amber10/exe/sander.MPI -O -i -p FT.prmtop -c
> equil1.rst -r equil2.rst -o equil2.out -x equil2.mdcrd
> At line 237 of file _getcor.f
> Fortran runtime error: Bad value during floating point read
> ---------------------------------------------------------------------------------------------------------------

look at equil1.rst. Do you see a " ******** " anywhere? This would
mean that one of the values for one of the coordinates got too large
for the fixed format float used in restart files. If you do see this,
here is the best way forward:

I'm assuming that you're using PME with explicit solvent, since
otherwise particles would not migrate so far away from the origin
(center of mass movement is eliminated every nscm steps). If this is
the case, re-run your first equilibration run with the variable
"iwrap=1" in the &cntrl section of your input file. This will wrap
diffusing molecules that leave your current box back into the other
side (thereby preventing any coordinate from becoming too large for
the restart to handle).

As a side note, if you are just doing regular molecular dynamics
simulations, then I would consider using pmemd instead of sander.MPI.
pmemd scales much better, and in my experience iwrap tends to work
better as well.

Hope this helps,

> I request you to kindly guide me in overcoming this problem
> With the best of regards
> Ramesh K V
>      The INTERNET now has a personality. YOURS! See your Yahoo! Homepage.
> _______________________________________________
> AMBER mailing list

Jason M. Swails
Quantum Theory Project,
University of Florida
Ph.D. Graduate Student
AMBER mailing list
Received on Fri Mar 05 2010 - 21:00:04 PST
Custom Search