Negation Normal Form
This page describe the Negation Normal Form used in Rodin and show how it has been implemented.
Definition
A predicate is in Negation Normal Form when negation occurs only on atomic expression. To compute such a normal form, the following rewrite rules (which are already implemented in Rodin) are reused :
SIMP_NOT_NOT :
DISTRI_NOT_AND :
DISTRI_NOT_OR :
DERIV_NOT_IMP :
DERIV_NOT_FORALL :
DERIV_NOT_EXISTS :
However, no rule has been defined for the case where the inner predicate is an equivalence, such as . The latter is equivalent to both and . In this latter case, the way the tactic will work is yet to be defined. Since the "" is to be kept, the choice on which predicate the not will be applied to is ambiguous. Here is an example showing that difficulty :
(1) which is equivalent to (2)
A solution would be to applied systematically on the left expression. In that case, simple simplification in the right member as in (1) would not be applied, thus complicating the left expression.
A better solution would to apply the on a member containing already a . But it can be hard to find an expression with a which has already been put in its Negation Normal Form as for the right member in (2)
Implementation
Tactics
When a predicate with a negation is matched, rules are applied following a pre-order traversal of that predicate.
INSTEAD OF
This way, tree proof is not unnecessarily complicated, and so more legible. The space used on disk is also smaller.