[gmx-users] weird nst*out problem

[Bert de Groot]

David wrote:
> 
...

> 
> I think Anton is correct in assuming the right frames are stored, but
> the precision of the float used for time is not more than 7 digits. You
> therefore get these rounding problems.
> 

yes this seems indeed to be the case. The "problem" already starts
earlier:

    25000.00000
    25000.19922
    25000.40039

but this is still ok (for e.g. gmxcheck), but later on it gets more
serious:

   300000.00000
   300000.09375
   300000.18750
   300000.31250
   300000.40625
   300000.50000

or worse:

  3000000.00000
  3000000.25000
  3000000.50000
  3000000.50000
  3000000.75000
  3000001.00000
  3000001.25000
  3000001.50000
  3000001.50000

(note that if you write out less frequently, like most users would do,
the problem starts much later)


If we want to prevent this, would it suffice to use a double for the
time variable in mdrun in our local gromacs version, or would it be better 
to completely switch to double precision?

cheers,

Bert

____________________________________________________________________________
Dr. Bert de Groot

Max Planck Institute for Biophysical Chemistry
Theoretical molecular biophysics group
Am Fassberg 11 
37077 Goettingen, Germany

tel: +49-551-2011306, fax: +49-551-2011089

email: bgroot at gwdg.de
http://www.mpibpc.gwdg.de/abteilungen/071/bgroot
____________________________________________________________________________