These versions can be more efficient when they are called many times, since they do not require allocating a new vector on each call.
See Implementation
These versions can be more efficient when they are called many times, since they do not require allocating a new vector on each call.