[gmx-users] Problems with the SHAKE-algorithm

[Björn Fischer]

Dear gromacs users,

I am trying to run a nonane gas phase system (custom force field). It the
beginning, everything looks ok, but after some time an error like this
occurs:

 > >Constraint error in algorithm Shake at step 74147
 > >Wrote pdb files with previous and current coordinates
 > >Inner product between old and new vector <= 0.0!
 > >constraint #1 atoms 191 and 192
 > >   i     mi      j     mj      before       after   should be
 > > 190  14.06    191  15.03     4.25365     1.60942     0.15400
 > > 191  15.03    192  15.03     4.19749     1.90346     0.15400
 > > 192  15.03    193  15.03     0.08362     0.88827     0.15400
 > > 193  15.03    194  15.03     0.40714     0.46199     0.15400
 > > 194  15.03    195  15.03     4.28246     5.11853     0.15400
 > > 195  15.03    196  15.03     3.89289     4.69530     0.15400
 > > 196  15.03    197  15.03     0.30450     0.33541     0.15400
 > > 197  15.03    198  14.06     0.28803     0.34247     0.15400

The timestep, at which this happens, varies with the (randomly
generated) start configuration. The temperature of the system rises
about one order of magnitude, and shortly afterwards all the output
in the md.log file is NaN.

Do you have any suggestions where to look for the source of the problem ?

Thanks in advance for your help!

Björn