[gmx-users] Error using double precision

[Mark Abraham]

edvin.erdtman at nat.oru.se wrote:
>> edvin.erdtman at nat.oru.se wrote:
>>> I installed both double and single precision gromacs on a new system. When running a membrane MD with double precision 8 processor mpi calc, this error appars in the error output file:
>>
>> this usually means you have an unequilibrated system.
> 
> But it worked with single precision! Or do I need to make a energy minimization first in DP?

Check out http://wiki.gromacs.org/index.php/Steps_to_Perform_a_Simulation

MD simulations are chaotic, and from the same starting point (as 
measured to single precision) you will observe divergent behaviour, 
increasingly so as time progresses. The fact that only one of the 
simulations crash is likely just a numerical artefact.

Mark