Suppose you have a database with the following content:
son(a, d).
son(b, d).
son(a, c).
son(b, c).
So a and b are sons of d and c. Now you want to know, given a bigger database, who is brother to who. A solution would be:
brother(X, Y) :-
son(X, P),
son(Y, P),
X \= Y.
The problem with this is that if you ask "brother(X, Y)." and start pressing ";" you'll get redundant results like:
X = a, Y = b;
X = b, Y = a;
X = a, Y = b;
X = b, Y = a;
I can understand why I get these results but I am looking for a way to fix this. What can I do?
Prolog will always try to find every possible solution available for your statements considering your set of truths. The expansion works as depth-first search:
son(a, d).
son(b, d).
son(a, c).
son(b, c).
brother(X, Y) :-
son(X, P),
son(Y, P),
X \= Y.
brother(X, Y)
_______________________|____________________________ [son(X, P)]
| | | |
X = a, P = d X = b, P = d X = a, P = c X = a, P = b
| | | |
| ... ... ...
|
| (X and P are already defined for this branch;
| the algorithm now looks for Y's)
|__________________________________________ [son(Y, d)]
| |
son(a, d) -> Y = a son(b, d) -> Y = b
| |
| | [X \= Y]
X = a, Y = a -> false X = a, Y = b -> true
|
|
solution(X = a, Y = b, P = d)
But, as you can see, the expansion will be performed in all the branches, so you'll end up with more of the same solution as the final answer. As pointed by #Daniel Lyons, you may use the setof built-in.
You may also use the ! -- cut operator -- that stops the "horizontal" expansion, once a branch has been found to be valid, or add some statement that avoids the multiple solutions.
For further information, take a look at the Unification algorithm.
First, I would advise against updating the Prolog database dynamically. For some reasons, consider the article
"How to deal with the Prolog dynamic database?".
You could use a combination of the builtin setof/3 and member/2, as #DanielLyons has suggested in his answer.
As yet another alternative, consider the following query which uses setof/3 in a rather unusual way, like this:
?- setof(t,brother(X,Y),_).
X = a, Y = b ;
X = b, Y = a.
You can eliminate one set with a comparison:
brother(X, Y) :-
son(X, P),
son(Y, P),
X \= Y, X #< Y.
?- brother(X, Y).
X = a,
Y = b ;
X = a,
Y = b ;
false.
Since X and Y will be instantiated both ways, requiring X be less than Y is a good way to cut the solutions in half.
Your second problem is that X and Y are brothers by more than one parent. The easiest solution here would be to make your rules more explicit:
mother(a, d).
mother(b, d).
father(a, c).
father(b, c).
brother(X, Y) :-
mother(X, M), mother(Y, M),
father(X, F), father(Y, F),
X \= Y, X #< Y.
?- brother(X, Y).
X = a,
Y = b ;
false.
This method is very specific to this particular problem, but the underlying reasoning is not: you had two copies because a and b are "brothers" by c and also by d—Prolog was right to produce that solution twice because there was a hidden variable being instantiated to two different values.
A more elegant solution would probably be to use setof/3 to get the solutions. This can work even with your original code:
?- setof(X-Y, (brother(X, Y), X #< Y), Brothers).
Brothers = [a-b].
The downside to this approach is that you wind up with a list rather than Prolog generating different solutions, though you can recover that behavior with member/2.
This should work. But I think it can be improved (I am not a Prolog specialist):
brother(X, Y) :-
son(X, P1),
son(Y, P1),
X #< Y,
(son(X, P2), son(Y, P2), P1 #< P2 -> false; true).
If you're using Strawberry Prolog compiler,you won't get all the answers by typing this:
?- brother(X, Y),
write(X), nl,
write(Y), nl.
In order to get all the answers write this:
?- brother(X, Y),
write(X), nl,
write(Y), nl,
fail.
I hope it helps you.:)
I got to an answer.
% Include the dictionary
:- [p1]. % The dictionary with sons
:- dynamic(found/2).
brother(X, Y) :-
% Get two persons from the database to test
son(X, P),
son(Y, P),
% Test if the two persons are different and were not already used
testBrother(X, Y).
% If it got here it's because there is no one else to test above, so just fail and retract all
brother(_, _) :-
retract(found(_, _)),
fail.
testBrother(X, Y) :-
X \= Y,
\+found(X, Y),
\+found(Y, X),
% If they were not used succed and assert what was found
assert(found(X, Y)).
It always returns fails in the end but it succeeds with the following.
brother(X, Y). % Every brother without repetition
brother('Urraca', X). % Every brother of Urraca without repetition
brother('Urraca', 'Sancho I'). % True, because Urraca and Sancho I have the same father and mother. In fact, even if they only had the same mother or the same father it would return true. A little off context but still valid, if they have three or more common parents it would still work
It fails with the following:
brother(X, X). % False because it's the same person
brother('Nope', X). % False because not is not even in the database
brother('Nope', 'Sancho I'). % False, same reason
So like this I can, for example, ask: brother(X, Y), and start pressing ";" to see every brother and sister without any repetition.
I can also do brother(a, b) and brother(b, a), assuming a and b are persons in the database. This is important because some solutions would use #< to test things and like so brother(b, a) would fail.
So there it is.
Related
I'm trying to program the unification algorithm in Prolog to verify if two expressions can unify by returning boolean True/False:
EDIT.
I found this implementation usefull:
from: http://kti.mff.cuni.cz/~bartak/prolog/data_struct.html
unify(A,B):-
atomic(A),atomic(B),A=B.
unify(A,B):-
var(A),A=B. % without occurs check
unify(A,B):-
nonvar(A),var(B),A=B. % without occurs check
unify(A,B):-
compound(A),compound(B),
A=..[F|ArgsA],B=..[F|ArgsB],
unify_args(ArgsA,ArgsB).
unify_args([A|TA],[B|TB]):-
unify(A,B),
unify_args(TA,TB).
unify_args([],[]).```
Here is a partial implementation of something like the Martelli and Montanari unification algorithm described at https://en.wikipedia.org/wiki/Unification_(computer_science)#A_unification_algorithm. The comments for each part refer to the corresponding rewrite rule from the algorithm. Note that there is no need for an explicit conflict rule, we can just fail if no other rule applies.
% assuming a universe with function symbols g/2, p/2, q/2
% identical terms unify (delete rule)
unify(X, Y) :-
X == Y,
!.
% a variable unifies with anything (eliminate rule)
unify(X, Y) :-
var(X),
!,
X = Y.
% an equation Term = Variable can be solved as Variable = Term (swap rule)
unify(X, Y) :-
var(Y),
!,
unify(Y, X).
% given equal function symbols, unify the arguments (decompose rule)
unify(g(A, B), g(X, Y)) :-
unify(A, X),
unify(B, Y).
unify(p(A, B), p(X, Y)) :-
unify(A, X),
unify(B, Y).
unify(q(A, B), q(X, Y)) :-
unify(A, X),
unify(B, Y).
Examples:
?- unify(q(Y,g(a,b)), p(g(X,X),Y)).
false.
?- unify(q(Y,g(a,b)), q(g(X,X),Y)).
false.
?- unify(q(Y,g(a,a)), q(g(X,X),Y)).
Y = g(a, a),
X = a.
One or two things remain for you to do:
Generalize the decompose rule to deal with arbitrary terms. You might find the =.. operator useful. For example:
?- Term = r(a, b, c), Term =.. FunctorAndArgs, [Functor | Args] = FunctorAndArgs.
Term = r(a, b, c),
FunctorAndArgs = [r, a, b, c],
Functor = r,
Args = [a, b, c].
You will need to check if two terms have the same functor and the same number of arguments, and whether all corresponding pairs of arguments unify.
Find out if your professor would like you to implement the occurs check, and if yes, implement it.
predicate change_pos(E1, E2,Lin,Lout).
The Lin has any number of elements, and I need to change all occurences of E1 to E2, and vice-versa. And return in Lout.
I was thinking to do something like this:
change(X, Y, [], []).
change(X, Y, [X|L], [Y,L1]):- change(X,Y,L,L1).
change(X, Y, [Z|L], [Z,L1]:- X \== Z, change(X,Y,L,L1).
But this way is not swiping two number of the list
I'm supposing, since this is homework, it's an exercise to learn list processing and recursion. But in Prolog, a common tool for processing each term in turn in a list is maplist:
% Rule for changing one element
change_element(X, Y, X, Y).
change_element(X, Y, Y, X).
change_element(X, Y, Z, Z) :- dif(X, Z), dif(Y, Z).
% Rule for changing a list
change(X, Y, L1, L2) :-
maplist(change_element(X, Y), L1, L2).
Which yields:
?- change(a, b, [a,b,c,b,a], L).
L = [b, a, c, a, b] ? ;
no
?-
For a determinate solution, you can use if_/3:
change1(X, Y, A, B) :-
if_(=(Y, A), B = X, A = B).
change2(X, Y, A, B) :-
if_(=(X, A), B = Y, change1(X, Y, A, B)).
change(X, Y, L1, L2) :- maplist(change2(X, Y), L1, L2).
Which yields:
?- change(a, b, [a,b,c,b,a], L).
L = [b, a, c, a, b].
?-
You're almost there. Your base case (the empty lists) and your second rule (swap X for Y) are basically fine (apart from the details pointed out in the comments). However, you are missing a rule for vice-versa (swap Y for X). And in your last rule you likely want to make sure that Z differs not only from X but also from Y, otherwise Z would be subject to rule two or three.
change(X, Y, [], []).
change(X, Y, [X|L], [Y|L1]) :-
change(X,Y,L,L1).
change(X, Y, [Y|L], [X|L1]) :- % <- vice versa case
change(X,Y,L,L1).
change(X, Y, [Z|L], [Z|L1]) :-
dif(X,Z), % <- neither X=Z
dif(Y,Z), % <- nor vice versa
change(X,Y,L,L1).
Here are some example queries. What does [1,2,3,4] look like after swapping 1 with 2 and vice versa?
?- change(1,2,[1,2,3,4],L).
L = [2,1,3,4] ? ;
no
What did [2,1,3,4] look like before swapping 1 with 2 and vice versa?
?- change(1,2,L,[2,1,3,4]).
L = [1,2,3,4] ? ;
no
Which elements have been swapped in [1,2,3,4] if the resulting list is [2,1,3,4] ?
?- change(X,Y,[1,2,3,4],[2,1,3,4]).
X = 1,
Y = 2 ? ;
X = 2,
Y = 1 ? ;
no
This question already has answers here:
Prolog, X before Y in a List
(4 answers)
Closed 6 years ago.
I am going to write predicate which is true iff only and only when element X occurs before Y on list L
before(L, X, Y) :-
nth1(PX, L, X),
nth1(PY, L, Y),
PX < PY.
Above, you can see my solution. What do you think about it ?
When it comes to my specific question:
My predicate returns true when there is exists at least one pair that Y followed X. How to define predicate such that it is true for each pair ?
The solution you show works for the "if one exists" case, but is somewhat imperative in nature. That is, it's a little bit like a C program translated to Prolog. Imperative means you are telling the computer, using the programming language, what steps to execute in order to achieve your results.
To be more declarative or relational, your "exists" solution could be expressed nicely as a DCG:
... --> [].
... --> [_], ... .
before(X, Y) --> ... , [X], ... , [Y], ... .
(NOTE: You can in Prolog have a predicate named ..., which is shown here.) This describes the relationship of X and Y in the list. It does not describe steps to execute, but instead describes the relationship of X and Y in a sequence. This solution has been shown before on SO.
Following this approach (where we describe the relationship of X and Y), one way (not necessarily the only way) to express that all the X precede all the Y would be:
before_all(X, Y) -->
{ dif(X,Y) },
any_sequence_but(Y), [X], any_sequence_but(Y), [Y], any_sequence_but(X).
any_sequence_but(_) --> [].
any_sequence_but(Y) --> [X], { dif(X,Y) }, any_sequence_but(Y).
Which yields a solution like this:
?- phrase(before_all(X,Y), [b,a,b,c,a,b,d]).
X = b,
Y = d ;
X = a,
Y = d ;
X = b,
Y = d ;
X = c,
Y = d ;
X = a,
Y = d ;
X = b,
Y = d ;
false.
?-
If the condition should hold for all pairs, the condition should hold for at least one pair, while its converse shouldn't be true for any pair.
I took the liberty of renaming your before/3 to beforeSome/3.
beforeSome(L, X, Y) :-
nth1(PX, L, X),
nth1(PY, L, Y),
PX < PY.
beforeAll(L, X, Y) :-
beforeSome(X,Y),
not(beforeSome(L, Y, X)).
Which yields the desired results:
?- beforeAll([1,2,3,1,4,5], 1, 4).
true.
?- beforeAll([1,2,3,1,4,5], 1, 2).
false.
Please note that your use of nth1/3 precludes it being used with uninstantiated variables. In other words, beforeAll([1,2,3,1,4,5], X, Y). is false.
A better implementation of beforeSome/3 would be something like
beforeSome([X|T], X, Y) :-
member(Y, T).
beforeSome([_|T], X, Y) :-
beforeSome(T, X, Y).
% no change needed, but repeated here for completeness' sake
beforeAll(L, X, Y) :-
beforeSome(X,Y),
not(beforeSome(L, Y, X)).
I have a simple knowledge base that encodes a family tree. Some important rules in this representation are as follows:
% fathers
father(michael,cathy).
father(michael,sharon).
father(charles_gordon,michael).
father(charles_gordon,julie).
father(charles,charles_gordon).
father(jim,melody).
father(jim,crystal).
father(elmo,jim).
father(greg,stephanie).
father(greg,danielle).
% mothers
mother(melody,cathy).
mother(melody,sharon).
mother(hazel,michael).
mother(hazel,julie).
mother(eleanor,melody).
mother(eleanor,crystal).
mother(crystal,stephanie).
mother(crystal,danielle).
% parents
parent(X,Y) :- father(X,Y).
parent(X,Y) :- mother(X,Y).
% men
male(michael).
male(charles_gordon).
male(charles).
male(jim).
male(elmo).
male(greg).
% women
female(cathy).
female(sharon).
female(julie).
female(hazel).
female(eleanor).
female(melody).
female(crystal).
female(stephanie).
female(danielle).
person(X) :- male(X) ; female(X).
parent(X,Y) :- father(X,Y) ; mother(X,Y). % X is parent of Y
child(X,Y) :- parent(Y,X).
elder(X,Y) :- parent(X,Y). % X is an elder of Y, meaning X is a parent or an ancestor of Y
elder(X,Y) :- parent(X,Z), elder(Z,Y).
junior(X,Y) :- child(X,Y). % X is a junior of Y, meaning X is a child or some descendant of Y
junior(X,Y) :- child(X,Z), junior(Z,Y).
I am attempting to find the nearest elder between two individuals(predicate ne(X,Y,Z)). This individual Z is the elder of both X and Y, and no junior of Z is also an elder of BOTH X and Y.
My attempt looks like this:
ne(X,Y,Z) :- person(X),
person(Y),
X \= Y,
elder(Z,X),
elder(Z,Y),
junior(A,Z),
not(elder(A,X)),
not(elder(A,Y)).
but this is somehow incorrect, because whenever I run ?- ne(stephanie,cathy,Z). i get
Z = jim ;
Z = jim ;
Z = jim ;
Z = jim ;
Z = elmo ;
Z = elmo ;
Z = elmo ;
Z = elmo ;
Z = eleanor ;
Z = eleanor ;
Z = eleanor ;
Z = eleanor ;
but i'm only supposed to get one answer, and I can't figure out what's wrong. Thanks!
from this graph
seems this answer is correct
?- ne(stephanie,cathy,A).
A = eleanor ;
A = jim.
here is my attempt to ne/3
ne(X,Y,Z) :-
setof(A, (
elder(A, X),
elder(A, Y),
X \= Y,
\+ (elder(A, T), elder(T, X) , elder(T, Y) ))
, As), member(Z, As).
not sure it's the better way...
Setof/3 (joined with member/2) is used to eliminate duplicate answers, since we get
?- aggregate(count,A^ne(stephanie,cathy,A),N).
N = 32.
with this core logic
ne(X,Y,A) :-
elder(A, X),
elder(A, Y),
X \= Y,
\+ (elder(A, T), elder(T, X) , elder(T, Y)).
note variable A replaces locally the original Z
edit
I didn't took in account the savvy comment by #Boris, but after removing the duplicate parent/2 definition, the setof/3+member/2 trick become useless.
Just a little question because I do not understand something on a program I have at one of my finals :
p(A, [A | _]).
p(B, [_, _ | C]) :- p(B, C).
q(D, [_, D | _]).
q(E, [_, _ | F]) :- q(E, F).
r(G, H) :- p(G, H).
r(I, J) :- q(I, J).
The question is to make the Tree of research of that with the purpose : r(X, [a,b,c]).
So actually, two possibilities :
r(G, H) :- p(G, H). (With G = X, and H = [a,b,c]).
r(I, J) :- q(I, J). (With I = X, and J = [a, b, c]).
If we take the 1st one, we have : p(X, [a,b,c]).
So we can use the 1st rule : p(A, [A | _]). (With A = X).
But I don't understand why when I launch the SWI-Prolog, a goes into X ...
Thanks !
Satisfying r(X, [a,b,c]) with the first rule goes into satisfying p(X, [a,b,c]).
Your first rule for p/2, p(A,[A|_]), states that a goal p(X,L) is satisfied when L is a list and its first element unifies with X.
In your case that list is [a,b,c] and X is unified with a.