Membership in Goal
Contents
Objective
This page describes the design of the reasoner MembershipGoal and its associated tactic MembershipGoalTac.
This reasoner discharges sequent whose goal denotes a membership which can be inferred from hypotheses. Here an basic example of what it discharges :
Analysis
Such sequent are proved by PP and ML. But, these provers have both drawbacks :
- All the visible are added as needed hypotheses, which is most of the time not the case.
- They take quite consequent time to prove it (even with the basic example given here above, the difference in time execution is noticeable).
- If there are too many hypotheses, or if the expression of the
is too complicated, they may not prove it.
This is particularly true when in the list of inclusion expressions of each side of the relation are not equal. For example :
Such a reasoner contributes to prove more Proof Obligations automatically, faster and with fewer needed hypotheses which makes proof rule more legible and proof replay less sensitive to modifications.
Design Decision
Tactic
This part explains how the tactic (MembershipGoalTac) associated to the reasoner MembershipGoal is working.
Goal
The tactic (as the reasoner) should works only on goals such as :
For examples :
In the latter case, the reasoner won't try to prove that x belongs to A and y belongs to B, but that the mapplet belong to the Cartesian product.
Hypotheses
Now we have to find hypotheses leading to discharge the sequent. To do so, the tactic recovers two kinds of hypotheses :
- the ones related to the left member of the goal
(this is the start point):
- the ones denoting inclusion (all but the ones matching the description of the first point) :
Then, it will search a link between those hypotheses so that the sequent can be discharged.
Find a path
Now that we recovered all the hypotheses that could be useful for the reasoner, it remains to find a path among the hypotheses leading to discharge the sequent. Depending on the relations on each side of the inclusion, we will act differently. always represent an expression (may be a domain, a range, etc.).
- The following sequent is provable because
.
- The following sequent is provable because
.
- We can generalized the first two points. This is the Russian dolls system. We can easily prove a sequent with multiple inclusions by going from hypothesis to hypothesis.
- For some relations, positions are needed to be known to continue to find hypotheses, but it is not always necessary.
By using these inclusion and rewrites, it tries to find a path among the recovered hypotheses. Every one of those should only be used once, avoiding possible infinite loop . Among all paths that lead to discharge the sequent, the tactic give the first it finds. Moreover, so that the reasoner does not do the same work as the tactic of writing new hypothesis, it gives all needed hypotheses and added hypotheses in the input.
Reasoner
This part describe how the reasoner MembershipGoal works.
Goal
First, it checks that the goal matches the description made in the part tactic : . Thus, we record the member x as well as the set S
Input
Then, it checks that the input is a hypothesesReasonerInput (an input with an array of predicates). Every given predicates must be contained in the sequent. Only one must be related to the goal's member so that there are no ambiguity. All the other ones must denote inclusion.
Find a path
With the same reasoning as for the tactic, we try to find a path leading to discharge the goal.
Trusted Base
At that point, the reasoner performs the same jobs as the tactic which is quite complicated. That poses one problem : it is hard to proof the reasoner is safe (only doing what it was meant to, not discharging sequent that cannot be proofed). Because reasoners are in the trusted base, we should be absolutely sure of they perform. How to validate the found path ?
As we know, the reasoner condense several inferences rules in only one proof rule. To validate the found path, we have to validate every single inference rule. To achieve it, we use the class created in that purpose : Rules. Each rules contain one predicate and an array of rules (its antecedents). When the path is searched, the rule corresponding is created. When the path is found, we test if the predicate of the rule is equal to the goal. If not, it means the path found was incorrect, so the reasoner fails, else the sequent is discharged.
Example of rules :
Implementation
This section explain how the tactic has bee implemented.
Positions
As it was said, we may sometimes need the position. It is represented by an array of integer. Here are the possible values the array contains (atomic positions) :
- kdom : it corresponds to the domain.
- kran : it corresponds to the domain.
- converse : it corresponds to the child of an inverse
For example, the following expressions at the given positions are equivalent.
Some combinations of atomic positions are equivalent. In order to simplify comparison between positions, they are normalized :
Goal
As explained in the design decision part, goal is checked. If it matches the description here above then x is stored in an attribute. Moreover, from the set S, we compute every pair expression & position equivalent to it. For example, from the set
, the map will be computed :
A pair (expression ; position) is said equal to the goal if and only if there exists a pair equivalent to the goal (GoalExp ; GoalPos) and a pair equivalent to the given pair (Exp ; Pos) such as Pos = GoalPos and Exp is contained in GoalExp.
Hypotheses
As explained in the design decision part, there are two kinds of hypotheses which are recovered. But when hypotheses related to the left member of the goal are stored, the position of x is also record. Then, if there is an hypothesis such as
, then this hypothesis is mapped to the positions
.
Find a path
Let's considered the sequent with the following goal : .
We start with the hypotheses which have a connection with the goal's member. Such a hypothesis gives two informations : the position
and the set
as explained in hypotheses. Then, for each equivalent pair to these one
, we compute set containing
( Find a path 2.). For every new pair, we test if it is contained in the goal.
To be continued.
Untreated cases
Some cases are not treated. Further enhancement may be provided for some.
where
and
are ones of :