[gmx-users] Range checking error: ci barely out of bounds.

[chris.neale at utoronto.ca]

Hello, I previously posted a problem about crashes in parallel but not serial:
http://www.gromacs.org/pipermail/gmx-users/2006-September/024003.html
I resolved that problem by running 5ns in serial and then, once I no  
longer needed to restrain the crystal waters, adding them to the main  
TIP4P SOL group.

I then moved my unrestrained simulation to 4cpu parallel and  
everything went fine for 1.5ns when it crashed (see bottom for text  
output upon crash). The energy is stable when viewing output from each  
2ps. So is the pressure based on the .edr and viewing the .xtc, both  
saved every 10ps.

Total Energy: -8e5 +/- 450
Temperature: 300 +/- 1
Pressure(bar): -2 +/- 116
X(nm): 9.2 +/-0.01
Y(nm): 9.1 +/-0.01
Z(nm): 7.7 +/-0.02

This is a protein in a fully solvated membrane with 57K atoms. I am  
running gromacs-3.3.1 in 800ps segments usgin tpbconv -extend to set  
up new runs. I am using shuffle/sort and when I deshuffle the  
trajectory it looks good for the 1.5ns that I have. tau_p=4.0 for  
semiisotropic XY-Z and tau_t=0.1 for each of 3 groups. Constraints are  
all-bonds with lincs and dt=0.002. It may or may not be relevant that  
I have a double [ pairs ] list for my lipids. I have seen people post  
similar neighboursearching issues and received suggestions regarding  
pressure coupling. However, I do not see any obvious pressure issues  
and I am going on strong-but-not-entirely-proven evidence that this is  
a parallel-specific problem.

    Energies (kJ/mol)
           Angle    Proper Dih. Ryckaert-Bell.  Improper Dih.          LJ-14
     4.66383e+04    1.42350e+04    1.46400e+04    1.47575e+03    2.35022e+04
      Coulomb-14        LJ (SR)        LJ (LR)   Coulomb (SR)   Coul. recip.
     7.84954e+03    1.04413e+04   -9.08592e+03   -6.09933e+05   -4.18189e+05
       Potential    Kinetic En.   Total Energy    Temperature Pressure (bar)
    -9.18426e+05    1.19106e+05   -7.99320e+05    2.99675e+02    2.07330e+02

Grid: 13 x 13 x 10 cells
Grid: 13 x 13 x 11 cells
Grid: 13 x 13 x 10 cells
            Step           Time         Lambda
          696000     1392.00000        0.00000

-------------------------------------------------------
Program mdrun_mpi, VERSION 3.3.1
Source code file: nsgrid.c, line: 226

Range checking error:
Explanation: During neighborsearching, we assign each particle to a grid
based on its coordinates. If your system contains collisions or parameter
errors that give particles very high velocities you might end up with some
coordinates being +-Infinity or NaN (not-a-number). Obviously, we cannot
put these on a grid, so this is usually where we detect those errors.
Make sure your system is properly energy-minimized and that the potential
energy seems reasonable before trying again.

Variable ci has value 1721. It should have been within [ 0 .. 1690 ]
Please report this to the mailing list (gmx-users at gromacs.org)
-------------------------------------------------------

"'Nay. We are but men.' Rock!" (Tenacious D)


#########

Is it significant that ci was 1721, which is much closer to [ 0 ..  
1690 ] than the -82745956412 type numbers that I get when my system is  
exploding?

Many thanks,
Chris Neale.