I have implemented the following function in prolog with开发者_开发知识库 the following code:
abs2(X, Y) :- X < 0, Y is -X.
abs2(X, X) :- X >= 0, !.
How can I implement this function without the use of cut ("!")?
There's the "hidden" cut in the if-then-else construct of Prolog:
abs2(X,Y) :- X < 0 -> Y is -X ; Y = X.
It is something of a quirk, but Prolog does not backtrack on the subgoal that forms the "premise" of an if-then or if-then-else construct. Here, if X < 0 succeeds the first try, then the choice of "then" clause over "else" clause is committed (hence the description of this behavior as a "hidden" cut).
There is more of a role for a cut in the first clause of the predicate abs2/2 as written in the question. As Nicholas points out, the cut at the end of the second clause doesn't have any effect (there are no choice points left when you get there). But as Kaarel points out, there is a choice point left open if the first clause succeeds.
So what I would have written, allowing the use of a cut, is this:
abs2(X,X) :- X >= 0, !.
abs2(X,Y) :- Y is -X.
Nicholas's comments also suggest ways to "arithmetize" the absolute value (rather than use a logic definition) and avoid "cut" that way.
My prolog is a bit rusty, but why do you even need the cut? If you write the predicate properly, backtracking can't succeed, so the cut is unnecessary:
abs(X, Y) :- number(X) , X < 0 , Y is -X .
abs(X, X) :- number(X) , X >= 0 .
No need to use prolog-cut!
Simply write:
abs2(X,Y) :- Y is abs(X).
精彩评论