per particle quantity update in parallel simulation

is there any
way that two owned particles with same id can exist at the same time?

There isn't. But it is up to you to insure that info
associated with that particle (e.g. the stored trajectory
points, or what you call a local array), is carried along
with the particle when it migrates to a new processor.

There are fix callback functions that do this, but you have
to have them in your fix and write them correctly. Those callbacks
will be invoked at every reneighboring.