I am having a difficult time understand recursion in prolog. I can read examples and sometimes understand, but I mostly have a difficult time implementing them. For example, could someone code me how to find the summation all the elements in a list, and go through it? and tips on how to approach a question like this? Thanks!
A general "good" explanation is not possible, because a good explanation needs to link to the previous knoledgment of the person. I'm going, by example, assume you are able to made a "proof by induction".
Step1: Let start by the initial fact, "the sum of a set with a single element is the element itself". In prolog:
sum([A],A).
Step2: if the sum of a set Q is SQ, the sum of this set adding one element H is H+SQ. In prolog:
sum([H|Q],R) :- sum(Q,SQ), R is H+SQ.
thats all, you have the problem solved. But...
In general, we try to start by the most basic set, the empty one, so replace "step 1" that becames now: the sum of the elements of an empty set is 0:
sum([],0).
Finally, prolog is more efficiente if the rules are tail recursive (if the execution environment is not able to optimice by itself). That means a little change:
sum([],R,R).
sum([H|Q],SQ,R) :- T is SQ+H, sum(Q,T,R).
these rules can be understood as: Assume (assert) that sum of Q is SQ. In this case, sum of set Q plus an element H is SQ+H. The first one means, when there are no more elements in the pending set, the result is directly the acumulated sum.
Thinking recursively can be hard. See my answer to the question "Prolog programming - path way to a solution" for links to good resources on how to think recursively.
For instance, most recursive problems can be broken down into a few (1 or 2) special cases, and then, the general case. In your case — computing the sum of a list of numbers — one might look at it has having 1 or two special cases. First, you have to make a decision: What is the sum of an empty list? One might argue either that the sum of an empty list is zero, or that an empty list has no sum. Either is, arguable, a perfectly valid point-of-view.
In either event, the special cases are
[]. The empty list. The sum of the empty list is either 0, or nothing (in which case your predicate should fail.)
[100]. A list of length one. The sum of a list of length 1 is obviously that value of the first and only entry.
And the more general case:
[100,101,102]. The sum of a list of length greater than 1 can be computed by taking the value of the first item in the list and computing the sum of the remainder. Note that
The solution is defined in terms of itself, and
The problem is made smaller (by removing the 1st item from the list).
Eventually, the problem will degenerate into one of the special cases, right?
Given all that, let us suppose that we've decided that the sum of the empty list is to be 0. That means our 2nd special case (a single element list) goes away, leaving us with a solution that can be described as
The sum of an empty list is 0.
The sum of a non-empty list is computed by
removing the 1st item from the list,
computing the sum of the remaining items,
adding the value of the 1st item to the sum of the remainder.
And since prolog is a declarative language, our solution is going to be pretty much identical to the description of the solution:
sum_of_list( [] , 0 ) .
sum_of_list( [N|Ns] , S ) :-
sum_of_list(Ns,T) ,
S is T+N
.
The c
Related
I'm trying to delete an element form a list of lists. This is an example of what I'm trying to do:
List=[[1,2],[1,3],[4,2]]
Max=2
delete(List,[_|Max],List2)
List2=[[1,3]]
Explanation: so given List I'm trying to use delete/3 to remove every element that has Max as its second element in its sub list. I'm not sure if this is possible with the standard delete/3 predicate. Thanks for the help!
This is what I have tried so far for my own predicate. I realize this isn't working and won't do what I want it.
second_element(E, []).
second_element(E,L):-
member(X,L),
X==[_,E],
delete(L,X,L2),
second_element(E,L2).
I'm not going to just give you the answer to your assignment. But I can give you a pattern commonly used in Prolog that it follows. This plus the hint I provided in comments are all the tools you need to solve the problem.
If you have a list, L, and you want another list, RL, that omits those elements of L that meet a certain criteria, then this is handled with a common recursive pattern:
filtered_list([], []). % The empty list is a filtered version of the empty list
filtered_list([H|T], R) :-
( meets_criteria_for_omission(H)
-> R = RT % R will not have H if it meets criteria for omission
; R = [H|RT] % R will have H if it does not meet criteria for omission
),
filtered_list(T, RT).
In the case of the original problem, the criteria for omission is that the element being checked is a list whose second element is given. That means that element being checked against must be provided as an argument. And the criteria to check is whether it is the second element of the given list.
This is as much information as I can give without simply handing over the complete answer, but it's 85% of the way there. Time to use the noodle.
I have a number of lists containing letters and I have written a predicate that checks whether or not there are duplicates present in one of these given lists:
noDuplicates([]).
noDuplicates([H|T]):-
not(member(H, T)),
noDuplicates(T).
I have 10 lists and I want to know if there are no duplicates in any of them, so I made them into sublists of one big list, something like:
[[A,B,C], [C,A,D], [E,F,G]...]]
(So there can be duplicates in the big list, but not the individual sublists).
I get that I have to do the duplicates test 10 times; once for every sublist, but how do I write this in Prolog? I could probably write it down 10 times, but my guess is I can use recursion to make prolog repeat itself until all sublists have been checked.
So basically: I want this predicate to repeat itself N times, until N is 10. I'm really struggling with it though. Does anyone have any idea on what to do?
Let us generalize the question as follows:
You have a predicate p/1 that expresses what you want for a single list.
Thus, to lift this definition to a list of such lists, you can define a predicates ps/1 as follows:
ps([]).
ps([L|Ls]) :-
p(L),
ps(Ls).
Every time you see this pattern, you can use maplist/2. That is, the above is equivalent to:
ps(Ls) :- maplist(p, Ls).
The goal maplist(p, Ls) is true iff p holds for each element L of Ls.
Note that it will limit your understanding of Prolog if you think in terms of "looping" and "repeating". These are imperative notions and only make sense when the list is already fully instantiated. However, from Prolog, we expect more than that: We expect a full-fledged relation to also generate lists for which the relation holds. And in such cases, there is nothing to "repeat" yet: We start from nothing, and ask Prolog what solutions there are in general.
Thus, think in terms of describing when the relation ps/1 holds for lists of lists:
It holds for the empty list [].
It holds for the list [L|Ls] if our initial predicate (p/1) holds for L, and ps/1 holds for the remaining list Ls.
This declarative reading is applicable in all directions, no matter how many list elements are already instantiated, if any. It works for 10 lists just as well as for zero and 50.
I want to define predicate which takes a list, adds an element to the list, let's say the number "1", and then returns the list.
I've found out I can add elements to a list using append/3, but I want to use in inside another predicate, thus why I want it to return "my modified list".
My object-oriented mindset tells me to ask the interpreter something like: ?-append(X,5,X). , so that it takes the list X, adds 5 to it, and returns "the new X", but I know that's not how unification works, so my mind is kinda in a glitch.
Can anyone please try to explain how something like this could be achievable?
You are already very close to the solution, so I only rephrase what you are beginning to sense already:
First, you cannot modify a list in pure Prolog.
Instead, you should think in terms of relations between entities. In your case, think in terms of relations between lists.
So, "adding the number 1" to a list is a relation between two lists, which could look like this:
list_with_one(Ls, [1|Ls]).
Note that this works in all directions! You can use it to:
generate answers
test particular cases
"reverse" the direction etc.
So, all you need to do in your case is to think in terms of relations between lists: One without an element, and how this relates to a different list with the element.
Obviously, these two lists will be indicated by different variables and different arguments.
Note in particular that append(X, 5, X) cannot hold: First of all, append/3 is meant to be a relation between lists, and 5 is not a list. Second, assuming you wrote for example append(Xs, [5], Xs), then this would be true if there where a list Xs such that if the element 5 were appended to Xs, the resulting list would again be Xs. Good luck finding such a list... Note also the naming convention to denote lists by letting the variable name end with an s.
It is also falls a bit short to blame this on your "object-oriented mindset", since you can have object oriented programming in Prolog too.
Although lists in Prolog cannot be modified, it is possible to add elements to the end of a list with an unspecified length. In this way, items can be "appended" to a list without creating another list:
:- initialization(main).
append_to_list(List,Item) :-
append_to_list(List,Item,0).
append_to_list(List,Item,Index) :-
% using SWI-Prolog's nth0 predicate
(nth0(Index,List,Check_Item),
var(Check_Item),
nth0(Index,List,Item));
(Next_Index is Index+1,
append_to_list(List,Item,Next_Index)).
main :-
A = [1,2,3|_],
append_to_list(A,4),
append_to_list(A,7),
writeln(A).
In this example, A becomes [1,2,3,4,7|_].
let's say we have a list of elements:
[(a,b); (c,d); (e,f)]
What function would check if element (lets say A, where A=(x,y)) is in the list or not?
Use List.mem to do the search for a match.
let a = (3,4)
List.mem a [(1,2); (3,4); (5,6)]
You can also use List.memq if you want to check if the two items both reference the same entity in memory.
Here's a hint on how to write this yourself. The natural way to to process a list is to initially look at the first element, then check the remainder of the list recursively until you have an empty list. For this problem, you could state the algorithm in English as follows:
If the list is empty then the item is not in the list,
else if the first list element equals the item then it is in,
else it is the answer to (Is the item in the remainder of the list?)
Now you just need to translate this into OCaml code (using a recursive function).
In general, if you can describe what you want to do in terms of smaller or simpler parts of the problem, then writing the recursive code is straightforward (although you have to be careful the base cases are correct). When using a list or tree-structured data the way to decompose the problem is usually obvious.
Yes this is a homework/lab assignment.
I am interesting in coming up with/finding an algorithm (I can comprehend :P) for using "backtracking" to solve the subset sum problem.
Anyone have some helpful resources? I've spent the last hour or so Googling with not much like finding something I think I could actually use. xD
Thanks SO!
Put the data in a vector.
Then write a routine that has 3 arguments: the vector, an index, and a sum.
Call this routine with the following arguments: the vector, 0, 0.
The routine should do the following tasks:
check if we reached the end of the vector (index==size). If this is the case, we can return immediately.
call itself with arguments: the vector, index+1, sum+vector[index]
(in this case we add the element at the index to the sum and continue with the vector)
call itself with arguments: the vector, index+1, sum
(in this case we don't add the element at the index to the sum, but still continue)
I deliberately left out 2 parts in this algorithm:
first, you should check the sum at some point. If it is zero, then you found a correct subset
second, you should also pass knowledge about which elements you used, so if the sum is zero, you can print out the subset. Consider using an STL::set for this.
Alternatively, you can use the return value of the function to determine whether a correct subset has already been found or not.
The complexity of the algorithm is O(2^N) so it will be very slow for big sets.
Have fun.