[gmx-users] Odd results from rerun for solvent water

[Mecklenfeld Andreas]

Dear Gromacs users,


I've encountered an odd problem regarding md -rerun. I'm looking at a small 
molecule dissolved in water (TIP3P or TIP4P/2005).  I've generated the 
trajectory of the system and used this trajectory to perform a rerun. The 
.trr seems fine (visualized in VMD, analyzed with gmx rdf) and the 
corresponding .edr yields normal results as well.
For the rerun, the force field and simulation parameters don't change, but I 
obtain a totally different outcome, like extremly high Coulomb recip. 
(factor 10^2) and kinetic energy / temperature (factor 10^5) terms. I've 
performed the rerun using different Gromacs versions (5.1.2, 2016.1 and 
2018.4) all with the same behaviour.
I've noticed, that using "coulombtype = Ewald" instead of "PME" gives 
realistic temperatures and that other solvents don't result in such errors. 
Furthermore, altering the number of PME ranks (down to 1) or reducing
ewald-rtol don't affect this behaviour.

I'm kind of confused right now and would appreciate any help.
Thanks in advance and kind regards

Andreas

   
-- M. Sc. Andreas Mecklenfeld
Technische Universität Braunschweig
Institut für Thermodynamik
Hans-Sommer-Straße 5
38106 Braunschweig
Deutschland / Germany

Tel: +49 (0)531 391-2634
      +49 (0)531 391-65685
Fax: +49 (0)531 391-7814

http://www.ift-bs.de