Open ehhartmann opened 2 months ago
gmx trjcat
is the cause of the instability. Often it struggles when only a few frames are generated; in my experience, it becomes more stable with a double-digit number of frames or more.
I'd keep the current behavior for now, as it works with rejection and rejection-free KMC. Otherwise, the user has one more option to make a mistake without gaining a real choice.
For rejection-free KMC it is not necessary to truncate trajectories. All we need to continue the run is a trr (and gro) file with the frame indicated by the recipe. For this we could use gmx trjconv -dump.
For rejection KMC/ extrande, we can keep the truncation as an option but the run should not fail if it doesn't work. This is just for analysis anyway.