I’ve observed some peculiar behaviour when using reax/c and adding molecules during a simulation. (The first observation of it was during the sue of fix deposit, but see below).
I’ve tried to track the issue down, as I suspect it really does originate from reax/c (that’s where the errors are coming from). I ended up with the attached simple example: if I run it unmodified, lammps halts after adding the second molecule with “p0: not enough space for bonds! total=217915748 allocated=25000”. The problem disappears when making the box smaller, also leading to the creation of ghost atoms during the first run (with only one molecule).
Looking at the reax/c code, I figured that it can handle molecule (= bond) additions during a simulation, but only when there are already ghost atoms present which can be replaced by the new atoms. But honestly, I could make sense of what is happening. I do know that the nonperiodic system i was previously studying caused bondchk errors when I wanted to use fix deposit, but I managed to get rid of those by tweaking the system to be periodic (and have ghost atoms prior to new molecule addition).
I understand my comments might sound somewhat vague, but I hope that the attached files can provide some insight.
ch3.txt (219 Bytes)
in.ch3 (1.4 KB)