Section: New Results

Computing Reveals

The reveals relation has been introduced in [33] between events in occurrence nets. Essentially, event $a$ is said to reveal event $b$ if in any maximal run that contains $a$, $b$ must also occur, be it before, after, or concurrently with $a$, and even if $a$ and $b$ are not causally related. Information of this kind is useful for diagnosis; a sensor for event $a$ may render a sensor for $b$ unnecessary.

In [33] , the reveals relation was shown to be decidable for occurrence nets that represent unfoldings of safe Petri nets. However, the upper bound was prohibitive for computing the relation in practice. In [51] we address this problem and drastically reduce the upper bound. We also propose efficient algorithms to actually compute the relation on a given occurrence net.