[gmx-users] 2019.3 - Fixed Issue 2953: With DD, all intermolecular interactions were ignored?

[Mateusz Bieniek]

Dear Developers,

I have read your latest release and I cannot work out how it affected my
simulations. The overall description is very general and implies that all
simulations done with 2019.0-2 should be discarded. If I
understand correctly, most of the simulations run nowadays use domain
decomposition (DD), and this is your official statement: "when running with
domain decomposition, all intermolecular interactions (when present) were
ignored".

However, in the issue 2953, Berk says "You can run with -ntmpi 1 for the
moment (or apply the fix)". Does this affect only Gromacs compiled without
separate MPI library? Can you comment on how the community was affect by
this?

My next question is about the software development practises. If this is
true that it is a very basic error, how is it possible that the test cases
did not detect this? Are not intermoecular interactions the most basic
interactions that you would expect gromacs to calculate?


Regards,
Mateusz Bieniek