在Wolf Goat Cabbage拼图的求解器中堆栈溢出

Question

我试图在Prolog中编写一个程序来解决众所周知的Wolf Goat Cabbage难题。鉴于一位农民想要用他的狼，山羊和卷心菜过河。这艘船只能同时容纳两只，他不能把狼与山羊或山羊一起放入白菜。

我知道Stackoverflow上有解决此问题的方法。但我想在我的代码中找到学习目的的错误。这是我的代码。它导致所谓的本地堆栈溢出，我猜逻辑中存在错误。由于我对每个块进行了评论，因此应该很容易理解。

% Helper function to check if first list
% is fully contained in second one.
subset([], []).
subset([E|Tail], [E|NTail]):-
    subset(Tail, NTail).
subset([_|Tail], NTail):-
    subset(Tail, NTail).

% There is space for two objects on the
% boat, but one of them must be the farmer.
crew(farmer).
crew(farmer, wolf).
crew(farmer, goat).
crew(farmer, cabbage).

% The riverside is safe if either the
% farmer is there, or not both wolf and
% goat or both goat and cabbage are there.
safe(Side) :-
    member(farmer, Side).
safe(Side) :-
    not(member(wolf, Side)),
    not(member(goat, Side)).
safe(Side) :-
    not(member(goat, Side)),
    not(member(cabbage, Side)).

% To embark, objects from one riverside,
% the crew must be valid an the riverside
% must stay safe. Disembarking is easy.
embark(Crew, Side, New) :-
    subset(Crew, Side),
    crew(Crew),
    safe(Side),
    delete(Side, Crew, New).
disembark(Crew, Side, New) :-
    append(Side, Crew, New).

% Crossing the river from one or the other side.
cross(Left, Right, Nextleft, Nextright) :-
    embark(Left, Crew, L),
    disembark(Right, Crew, R),
    cross(L, R, Nextleft, Nextright).
cross(Left, Right, Nextleft, Nextright) :-
    embark(Right, Crew, R),
    disembark(Left, Crew, L),
    cross(L, R, Nextleft, Nextright).

% Find solution to bring all objects from left
% riverside to right. Run this after consulting
% the file.
% cross([farmer, wolf, goat, cabbage], [], [], [farmer, wolf, goat, cabbage]).

这段代码有什么问题？我只是想深入了解Prolog。

Answer 1

SWI-Prolog基于外部参照的编辑器强调了第一个问题：从未使用过crew / 2：

然后你可能想要这个定义：

crew([farmer]).
crew([farmer, wolf]).
crew([farmer, goat]).
crew([farmer, cabbage]).

编辑我认为下一个问题在你召集工作人员/ 1的方式中很明显：

embark(Crew, Side, New) :-
    subset(Crew, Side),
    crew(Crew),
    safe(Side),
    delete(Side, Crew, New).

您正在传递已经形成的工作人员而不是子集/ 2生成的工作人员。 如果进入调试模式

?- leash(-all),trace.
true.
[trace] 3 ?- cross([farmer, wolf, goat, cabbage], [], [], L).
Call: (6) stackoverflow:cross([farmer, wolf, goat, cabbage], [], [], _G1474)
...

Exit: (8) stackoverflow:subset([farmer, wolf, goat, cabbage], [farmer, wolf, goat, cabbage])
Call: (8) stackoverflow:crew([farmer, wolf, goat, cabbage])
Fail: (8) stackoverflow:crew([farmer, wolf, goat, cabbage])
Redo: (11) stackoverflow:subset([cabbage], _G1560)
...
Exit: (8) stackoverflow:subset([farmer, wolf, goat, cabbage], [farmer, wolf, goat])
Call: (8) stackoverflow:crew([farmer, wolf, goat, cabbage])
Fail: (8) stackoverflow:crew([farmer, wolf, goat, cabbage])
Redo: (10) stackoverflow:subset([goat, cabbage], _G1557)
...

也就是说，船员总是失败......

Answer 2

事实证明该程序存在一些问题，大多数情况下不限制深度优先搜索并允许循环如（[F，G，W，C]，[]） - ＆gt;（[W ，C]，[F，G]） - ＆gt;（[F，G，W，C]，[]），它将无限下降。在诸如安全之类的事情中也存在一些小的逻辑错误（不允许山羊或狼留下很少的选择）。

作为一种学习经历，我经历了这种方法。我喜欢它捕获的思维过程，并希望看到它的发展。在这个过程中，我重新组织了一些，但它仍然应该是可识别的。我用prev添加了一个“no undo”移动检查，并且在一个“没有空的右侧”检查中添加了一个关闭，以切断对骨头路径的搜索。我还为prolog解释器添加了一些基础知识，这些解释器更加简单。但也许看到这将有助于另一个。

最后在SWI-Prolog上进行了测试。

member(X, [X|_]).
member(X, [_|Tail]) :- member(X, Tail).

subset([],_).
subset([X|L],Set):-
    member(X,Set),
    subset(L,Set).

delete([], _, []).
delete(List, [], List).
delete([X|Tail], [X|STail], Res) :- 
    delete(Tail, STail, Res).
delete([X|Tail], Sublist, Res) :- 
    member(X, Sublist),
    delete(Tail, Sublist, Res).
delete([X|Tail], Sublist, [X|Res]) :-
    not(member(X, Sublist)),
    delete(Tail, Sublist, Res).

append([],L,L).
append([H|T],L,[H|TL]) :- append(T,L,TL).

crew([farmer]).
crew([farmer, wolf]).
crew([farmer, goat]).
crew([farmer, cabbage]).

safe([]).
safe(Side) :-
    member(farmer, Side).
safe(Side) :-
    not(subset([goat, wolf], Side)),
    not(subset([goat, cabbage], Side)).

embark(Side, Crew, Remain, Prev) :-
    crew(Crew),
    subset(Crew, Side),
    not(Crew = Prev),
    delete(Side, Crew, Remain),
    safe(Remain).
disembark(Side, Crew, Remain) :-
    append(Side, Crew, Remain),
    safe(Remain).

cross([], Right, [], _) :-
    subset([farmer, wolf, goat, cabbage], Right).
cross(Left, Right, Move, Prev) :-
    embark(Left, Crew, NewLeft, Prev),
    disembark(Right, Crew, NewRight),
    cross(NewLeft, NewRight, Othermoves, Crew),
    Move = [[toright|Crew]|Othermoves].
cross(Left, Right, Move, Prev) :-
    embark(Right, Crew, NewRight, Prev),
    not(NewRight = []),
    disembark(Left, Crew, NewLeft),
    cross(NewLeft, NewRight, Othermoves, Crew),
    Move = [[toleft|Crew]|Othermoves].

solve(Left, Right, Moves) :-
    cross(Left, Right, Moves, []).