[gmx-users] Mdrun box nan error

Nathalie R. Basdevant nrb2001 at med.cornell.edu
Fri May 7 18:22:01 CEST 2004


Dear users,

I am running Gromacs 3.2.1 on an itanium 2 with Linux compiled with Intel
compilers. The system that I am simulating consists of a protein in water
box (OPLS+tip4p). I am simulating under NPT (both Berendsen baths) 300K and
1atm. I am running in parallel (4 processors).
Every once in a while the simulation crashes with the following error
messages (see below). However when I restart the simulation (tpbconv -f .trr
-e .edr ) the simulation passes through the step that had given the error
before without any problem, but eventually stops at a later time (varying
between tens to hundreds of ps) with a similar error, but on a different
atom.
I hope you can help me with this.
Thanks
-- 
Nathalie Basdevant  

IN MD0.LOG:

Step 291924, time 583.848 (ps)  LINCS WARNING
relative constraint deviation after LINCS:
max 0.004130 (between atoms 1632 and 1631) rms nan
bonds that rotated more than 30 degrees:
 atom 1 atom 2  angle  previous, current, constraint length
Grid: 0 x 0 x 0 cells
Fatal error: ci = -1 should be in 0 .. -1 [FILE nsgrid.c, LINE 218]


IN STDERR:

Step 291924, time 583.848 (ps)  LINCS WARNING
relative constraint deviation after LINCS:
max 0.004130 (between atoms 1632 and 1631) rms nan
bonds that rotated more than 30 degrees:
 atom 1 atom 2  angle  previous, current, constraint length
Warning: Only triclinic boxes with the first vector parallel to the x-axis
and the second vector in the x
y-plane are supported.
Warning: Only triclinic boxes with the first vector parallel to the x-axis
and the second vector in the x
y-plane are supported.
         Box (3x3):
            Box[    0]={         nan,          nan,          nan}
            Box[    1]={         nan,          nan,          nan}
            Box[    2]={         nan,          nan,          nan}
         Can not fix pbc.

This 'box' message is repeated several times and finally I get:

Fatal error: ci = -1 should be in 0 .. -1 [FILE nsgrid.c, LINE 218]
Fatal error: ci = -1 should be in 0 .. -1 [FILE nsgrid.c, LINE 218]
Fatal error: ci = -1 should be in 0 .. -1 [FILE nsgrid.c, LINE 218]
Fatal error: ci = -1 should be in 0 .. -1 [FILE nsgrid.c, LINE 218]
MPI: MPI_COMM_WORLD rank 3 has terminated without calling MPI_Finalize()





More information about the gromacs.org_gmx-users mailing list