Keren Lasker wrote:
Currently, once a particle is deactivated, various things will skip over it/remove it so trying to reactivate it afterwards would be a bit of a mess. The problem is that either we have to put checks everywhere in the code or we have to remove inactive particles from various lists (like the set of particles to be checked for nonbonded interactions).Sure - if we're going to allow particle deletion, there's no reason to have active/inactive.We might need it for discrete optimization via MC or inference.I would prefer if you'll leave it till we sort out the discrete optimization framework.
As a result, we should probably remove the methods now. If we want them later, we can put them back and change the other code.
What would you want to do with inactivating particles as opposed to deleting them?