Prolog and limitations of backtracking - prolog

This is probably the most trivial implementation of a function that returns the length of a list in Prolog
count([], 0).
count([_|B], T) :- count(B, U), T is U + 1.
one thing about Prolog that I still cannot wrap my head around is the flexibility of using variables as parameters.
So for example I can run count([a, b, c], 3). and get true. I can also run count([a, b], X). and get an answer X = 2.. Oddly (at least for me) is that I can also run count(X, 3). and get at least one result, which looks something like X = [_G4337877, _G4337880, _G4337883] ; before the interpreter disappears into an infinite loop. I can even run something truly "flexible" like count(X, A). and get X = [], A = 0 ; X = [_G4369400], A = 1., which is obviously incomplete but somehow really nice.
Therefore my multifaceted question. Can I somehow explain to Prolog not to look beyond first result when executing count(X, 3).? Can I somehow make Prolog generate any number of solutions for count(X, A).? Is there a limitation of what kind of solutions I can generate? What is it about this specific predicate, that prevents me from generating all solutions for all possible kinds of queries?

This is probably the most trivial implementation
Depends from viewpoint: consider
count(L,C) :- length(L,C).
Shorter and functional. And this one also works for your use case.
edit
library CLP(FD) allows for
:- use_module(library(clpfd)).
count([], 0).
count([_|B], T) :- U #>= 0, T #= U + 1, count(B, U).
?- count(X,3).
X = [_G2327, _G2498, _G2669] ;
false.
(further) answering to comments
It was clearly sarcasm
No, sorry for giving this impression. It was an attempt to give you a synthetic answer to your question. Every details of the implementation of length/2 - indeed much longer than your code - have been carefully weighted to give us a general and efficient building block.
There must be some general concept
I would call (full) Prolog such general concept. From the very start, Prolog requires us to solve computational tasks describing relations among predicate arguments. Once we have described our relations, we can query our 'knowledge database', and Prolog attempts to enumerate all answers, in a specific order.
High level concepts like unification and depth first search (backtracking) are keys in this model.
Now, I think you're looking for second order constructs like var/1, that allow us to reason about our predicates. Such constructs cannot be written in (pure) Prolog, and a growing school of thinking requires to avoid them, because are rather difficult to use. So I posted an alternative using CLP(FD), that effectively shields us in some situation. In this question specific context, it actually give us a simple and elegant solution.
I am not trying to re-implement length
Well, I'm aware of this, but since count/2 aliases length/2, why not study the reference model ? ( see source on SWI-Prolog site )

The answer you get for the query count(X,3) is actually not odd at all. You are asking which lists have a length of 3. And you get a list with 3 elements. The infinite loop appears because the variables B and U in the first goal of your recursive rule are unbound. You don't have anything before that goal that could fail. So it is always possible to follow the recursion. In the version of CapelliC you have 2 goals in the second rule before the recursion that fail if the second argument is smaller than 1. Maybe it becomes clearer if you consider this slightly altered version:
:- use_module(library(clpfd)).
count([], 0).
count([_|B], T) :-
T #> 0,
U #= T - 1,
count(B, U).
Your query
?- count(X,3).
will not match the first rule but the second one and continue recursively until the second argument is 0. At that point the first rule will match and yield the result:
X = [_A,_B,_C] ?
The head of the second rule will also match but its first goal will fail because T=0:
X = [_A,_B,_C] ? ;
no
In your above version however Prolog will try the recursive goal of the second rule because of the unbound variables B and U and hence loop infinitely.

Related

Why does my list reversal only work correctly in one direction?

I have produced the following code.
list_reverse([],[]).
list_reverse([X],[X]).
list_reverse(Ls,[R|Rs]) :-
last_elem(Ls,R),
without_last_elem(Ls,Next),
list_reverse(Next,Rs).
last_elem([E],E).
last_elem([_|Xs],E) :-
last_elem(Xs,E).
without_last_elem([X,_|[]],[X|[]]).
without_last_elem([X|T0],[X|T1]) :-
without_last_elem(T0,T1).
Swipl:
?- list_reverse([1,2,3],X).
X = [3, 2, 1] ;
false.
This is exactly what I want.
However if I go in the opposite direction I get success, followed by non-termination.
?- list_reverse(X,[1,2,3]).
X = [3, 2, 1] ;
C-c C-cAction (h for help) ? a
abort
% Execution Aborted
What I am struggling to understand is why I first get a correct solution for X. Is my program correct or not?
I am not worried about reversing a list as much as I am about this pattern of getting a correct solution followed by non-termination. It is a pattern I have already come across a few times.
I am [worried] about this pattern of getting a correct solution followed by non-termination.
This is due to the very specific notion of (universal) termination in Prolog. In other programming languages termination is a much simpler beast (still an undecidable beast nevertheless). If, say, a function returns then it terminates (for that case). But in Prolog, producing an answer is not the end as there might be further solutions or just an unproductive loop. In fact, it's best not to consider your query ?- list_reverse(X,[1,2,3]). but rather the following instead.
?- list_reverse(X,[1,2,3]), false.
In this manner all distracting answers are turned off. The only purpose of this query is now either to show termination or non-termination.
After that,
you can either try to follow Prolog's precise execution path but that is as insightful as staring into a car's gearbox when you are lost (the gears caused you to move into the place where you are lost thus they are somehow the cause...). Or, you take a step back, and consider related program fragments (called slices) that share certain properties with your original program. For termination, a failure-slice helps you to better understand what is at stake. In your case consider:
list_reverse([],[]) :- false.
list_reverse([X],[X]) :- false.
list_reverse(Ls,[R|Rs]) :-
last_elem(Ls,R), false,
without_last_elem(Ls,Next),
list_reverse(Next,Rs).
last_elem([E],E) :- false.
last_elem([_|Xs],E) :-
last_elem(Xs,E), false.
?- list_reverse(X,[1,2,3]), false.
Since this failure slice does not terminate, also your original program doesn't terminate! And, it is much easier to reason here in this smaller fragment. If you want to fix the problem, you need to modify something in the visible part. Otherwise you will keep being stuck in a loop.
Note that none of the facts is part of the loop. Thus they are irrelevant for non-termination.
Also note that in list_reverse/2 the variable Rs is never used in the visible part. Thus Rs has no influence on termination! Please note that this is a proof of that property already. Does this mean that the second argument of list_reverse/2 has no influence on termination? What do you think?
The last_elem/2 can keep constructing larger lists, that all should be rejected. But you thus get stuck in an infinite loop.
We can make a function that works with accumulator, and iterates over both the two lists concurrently. That means that once the left or right list is exhausted, no more recursive calls will be made:
reverse(L1, L2) :-
reverse(L1, [], L2, L2).
reverse([], L, L, []).
reverse([H|T], L1, R, [_|T2]) :-
reverse(T, [H|L1], R, T2).
Here the [H|T] and [_|T2] pattern thus both pop the first item of the list, and we only match if both lists are exhausted.

Prolog - proof tree misses possibilities

I have the following Prolog Program:
p(f(X), Y) :- p(g(X), g(Y)).
p(g(X), Y) :- p(f(Y), f(X)).
p(f(a), g(b)).
The prolog proof tree has to be drawn for the predicate p(X, Y).
Question:
Why is Y matched to Y1/Y and not to Y/Y1 and why is Y used further on?
if I match a predicate (e.g. p(X, Y)), I get a new predicate (e.g. p(g(X1), g(Y))) - why contains p(g(X1), g(Y)) just one subtree? I mean, shouldn't it have 3 because the knowledgebase contains 3 statements - instead of just 1?
And why is at each layer of the tree matched with something like X2/X1 and so on ? and not with the predicate before ?
Shouldn't it be g(X1)/fX5, g(Y1)/Y5 ?
Note: Maybe it seems that I have never done a tutorial or something. But I did.. I appreciate every help.
To be honest, I have rarely seen a worse method to explain Prolog than what you show here.
Yes, I expect the author meant Y/Y1 instead of Y1/Y in both cases, otherwise the notation would be quite inconsistent.
As to your other questions: You are facing the usual problems that arise when taking such an extremely operational view of Prolog. The core issue is that this method doesn't scale: You do not have the mental capacity to carry this approach through. Don't take this personal: Humans in general are bad at keeping all details of an execution tree that grows exponentially in mind. This makes the whole approach extremely cumbersome and error-prone. For comparison, consider why human grandmasters have stopped competing against chess computers already many years ago. In this concrete case, note for example that the rightmost branch does not even arise in actual Prolog execution, but the graph wrongly suggests that it does!
Part of the problem here is a confusion in terminology: Please note that Prolog uses unification (not "matching", which is one-sided unification). When you unify a goal with a clause head and the unification succeeds, then you get bindings for variables. You continue with these bindings in place.
To make the whole approach remotely feasible, consider fragments of your program.
For example, suppose I only give you the following fact:
p(f(a), g(b)).
And you then query:
?- p(X, Y).
X = f(a),
Y = g(b).
This answers shows the bindings for X and Y. First make sure you understand this, and understand the difference between these bindings and a "new predicate" (which does not arise!).
Also, there are no "statements", but 3 clauses, which are logical alternatives.
Now, again to simplify the whole task, consider the following fragment of your program, in which I only look at the two rules:
p(f(X), Y) :- p(g(X), g(Y)).
p(g(X), Y) :- p(f(Y), f(X)).
Already with this program, we get:
?- p(X, Y).
nontermination
Adding a further pure clause cannot prevent this nontermination. Thus, I recommend you start with this reduced version of your program, and consider it in more depth.
From there, you can add the remaining fact again, and consider the differences.
Very good questions!
Why is Y matched to Y1/Y and not to Y/Y1 and why is Y used further on?
The naming here seems a little arbitrary in that they could have used Y/Y1 but then would need to use Y1 further on. In this case, they chose Y1/Y and use Y further on. Although the author of this expression tree was inconsistent in their convention, I wouldn't be too concerned about the naming as much as whether they follow the variable correctly down the tree.
if I match a predicate (e.g. p(X, Y)), I get a new predicate (e.g. p(g(X1), g(Y))) - why contains p(g(X1), g(Y)) just one subtree? I mean, should'nt it have 3 because the knowledgebase contains 3 statements - instead of just 1?
First a word on term versus predicate. A term is only a predicate in the context of Head :- Body in which case Head is a term that forms the head of a predicate clause. If a term is an argument to a predicate (for example, p(g(X1), g(Y)), the g(X1) and g(Y) are not predicates. They are just terms.
More specifically in this case, the term p(g(X1), g(Y)) only has one subtree because it only matches the head of one of the 3 predicate clauses which is the one with the head p(g(X), Y) (it matches with X = X1 and Y = g(Y)). The other two can't match since they're of the form p(f(...), ...) and the f(...) term cannot match the g(X1) term.
And why is at each layer of the tree matched with something like X2/X1 and so on ? and not with the predicate before ?
Shouldn't it be g(X1)/fX5, g(Y1)/Y5 ?
I'm not sure I'm following this question, but the principle to follow is that the tree is attempting to use the same variable name if it applies to the same variable in memory, whereas a different variable name (e.g., X1 versus X) is used if it's a different X. For example, if I have foo(X, Y) :- <some code>, bar(f(X), Y). and I have bar(X, Y) :- blah(X), ... then the X referred to in the bar predicate is different than the X referred to in the foo predicate. So we might say, in the call to foo(X, Y) we're calling bar(f(X), Y), or alternatively, bar(X1, Y) where X1 = f(X).

Prolog: temporary list storage

I'm new to Prolog and I'm stuck on a predicate that I'm trying to do. The aim of it is to recurse through a list of quads [X,Y,S,P] with a given P, when the quad has the same P it stores it in a temporary list. When it comes across a new P, it looks to see if the temporary list is greater than length 2, if it is then stores the temporary list in the output list, if less than 2 deletes the quad, and then starts the recursion again the new P.
Heres my code:
deleteUP(_,[],[],[]).
deleteUP(P,[[X,Y,S,P]|Rest],Temp,Output):-
!,
appends([X,Y,S,P],Temp,Temp),
deleteUP(P,[Rest],Temp,Output).
deleteUP(NextP,[[X,Y,S,P]|Rest],Temp,Output):-
NextP =\= P,
listlen(Temp,Z),
Z > 1, !,
appends(Temp,Output,Output),
deleteUP(NextP,[_|Rest],Temp,Output).
listlen([], 0).
listlen([_|T],N) :-
listlen(T,N1),
N is N1 + 1.
appends([],L,L).
appends([H|T],L,[H|Result]):-
appends(T,L,Result).
Thanks for any help!
Your problem description talks about storing, recursing and starting. That is a very imperative, procedural description. Try to focus first on what the relation should describe. Actually, I still have not understood what minimal length of 2 is about.
Consider to use the predefined append/3 and length/2 in place of your own definitions. But actually, both are not needed in your example.
You might want to use a dedicated structure q(X,Y,S,P) in place of the list [X,Y,S,P].
The goal appends([X,Y,S,P],Temp,Temp) shows that you assume that the logical variable Temp can be used like a variable in an imperative language. But this is not the case. By default SWI creates here a very odd structure called an "infinite tree". Forget this for the moment.
?- append([X,Y,S,P],Temp,Temp).
Temp = [X, Y, S, P|Temp].
There is a safe way in SWI to avoid such cases and to detect (some of) such errors automatically. Switch on the occurs check!
?- set_prolog_flag(occurs_check,error).
true.
?- append([X,Y,S,P],Temp,Temp).
sto. % ERROR: lists:append/3: Cannot unify _G392 with [_G395,_G398,_G401,_G404|_G392]: would create an infinite tree
The goal =\=/2 means arithmetical inequality, you might prefer dif/2 instead.
Avoid the ! - it is not needed in this case.
length(L, N), N > 1 is often better expressed as L = [_,_|_].
The major problem, however, is what the third and fourth argument should be. You really need to clarify that first.
Prolog variables can't be 'modified', as you are attempting calling appends: you need a fresh variables to place results. Note this code is untested...
deleteUP(_,[],[],[]).
deleteUP(P,[[X,Y,S,P]|Rest],Temp,Output):-
!,
appends([X,Y,S,P],Temp,Temp1),
deleteUP(P, Rest, Temp1,Output). % was deleteUP(P,[Rest],Temp,Output).
deleteUP(NextP,[[X,Y,S,P]|Rest],Temp,Output1):-
% NextP =\= P, should be useless given the test in clause above
listlen(Temp,Z),
Z > 1, !, % else ?
deleteUP(NextP,[_|Rest],Temp,Output),
appends(Temp,Output,Output1).

Prolog successor notation yields incomplete result and infinite loop

I start to learn Prolog and first learnt about the successor notation.
And this is where I find out about writing Peano axioms in Prolog.
See page 12 of the PDF:
sum(0, M, M).
sum(s(N), M, s(K)) :-
sum(N,M,K).
prod(0,M,0).
prod(s(N), M, P) :-
prod(N,M,K),
sum(K,M,P).
I put the multiplication rules into Prolog. Then I do the query:
?- prod(X,Y,s(s(s(s(s(s(0))))))).
Which means finding the factor of 6 basically.
Here are the results.
X = s(0),
Y = s(s(s(s(s(s(0)))))) ? ;
X = s(s(0)),
Y = s(s(s(0))) ? ;
X = s(s(s(0))),
Y = s(s(0)) ? ;
infinite loop
This result has two problems:
Not all results are shown, note that the result X=6,Y=1 is missing.
It does not stop unless I Ctrl+C then choose abort.
So... my questions are:
WHY is that? I tried switching "prod" and "sum" around. The resulting code gives me all results. And again, WHY is that? It still dead-loops though.
HOW to resolve that?
I read the other answer on infinite loop. But I'd appreciate someone answer basing on this scenario. It greatly helps me.
If you want to study termination properties in depth, programs using successor-arithmetics are an ideal study object: You know a priori what they should describe, so you can concentrate on the more technical details. You will need to understand several notions.
Universal termination
The easiest way to explain it, is to consider Goal, false. This terminates iff Goal terminates universally. That is: Looking at tracers is the most ineffective way - they will show you only a single execution path. But you need to understand all of them at once! Also never look at answers when you want universal termination, they will only distract you. You have seen it above: You got three neat and correct answers, only then your program loops. So better "turn off" answers with false. This removes all distraction.
Failure slice
The next notion you need is that of a failure slice. Take a pure monotonic logic program and throw in some goals false. If the resulting failure slice does not terminate (universally), also the original program won't. In your exemple, consider:
prod(0,M,0) :- false.
prod(s(N), M, P) :-
prod(N,M,K), false,
sum(K,M,P).
These false goals help to remove irrelevant adornments in your program: The remaining part shows you clearly, why prod(X,Y,s(s(s(s(s(s(0))))))). does not terminate. It does not terminate, because that fragment does not care about P at all! You are hoping that the third argument will help to make prod/3 terminate, but the fragment shows you it is all in vain, since P does not occur in any goal. No need for chatty tracers.
Often it is not so easy to find minimal failure slices. But once you found one, it is next to trivial to determine its termination or rather non-termination properties. After some time you can use your intuition to imagine a slice, and then you can use your reason to check if that slice is of relevance or not.
What is so remarkable about the notion of a failure slice is this: If you want to improve the program, you have to modify your program in the part visible in above fragment! As long as you do not change it, the problem will persist. A failure slice is thus a very relevant part of your program.
Termination inference
That is the final thing you need: A termination inferencer (or analyzer) like cTI will help you to identify the termination condition rapidly. Look at the inferred termination conditions of prod/3 and the improved prod2/3 here!
Edit: And since this was a homework question I have not posted the final solution. But to make it clear, here are the termination conditions obtained so far:
prod(A,B,C)terminates_if b(A),b(B).
prod2(A,B,C)terminates_if b(A),b(B);b(A),b(C).
So the new prod2/3 is strictly better than the original program!
Now, it is up to you to find the final program. Its termination condition is:
prod3(A,B,C)terminates_if b(A),b(B);b(C).
To start with, try to find the failure slice for prod2(A,B,s(s(s(s(s(s(0)))))))! We expect it to terminate, but it still does not. So take the program and add manuallyfalse goals! The remaining part will show you the key!
As a final hint: You need to add one extra goal and one fact.
Edit: Upon request, here is the failure slice for prod2(A,B,s(s(s(s(s(s(0))))))):
prod2(0,_,0) :- false.
prod2(s(N), M, P) :-
sum(M, K, P),
prod2(N,M,K), false.
sum(0, M, M).
sum(s(N), M, s(K)) :- false,
sum(N,M,K).
Please note the significantly simplified definition of sum/3. It only says: 0 plus anything is anything. No more. As a consequence even the more specialized prod2(A,0,s(s(s(s(s(s(0))))))) will loop whileprod2(0,X,Y) elegantly terminates ...
The first question (WHY) is fairly easy to spot, specially if know about left recursion. sum(A,B,C) binds A and B when C is bound, but the original program prod(A,B,C) doesn't use that bindings, and instead recurse with still A,B unbound.
If we swap sum,prod we get 2 useful bindings from sum for the recursive call:
sum(M, K, P)
Now M is bound, and will be used to terminate the left-recursion. We can swap N and M, because we know that product is commutative.
sum(0, M, M).
sum(s(N), M, s(K)) :-
sum(N, M, K).
prod3(0, _, 0).
prod3(s(N), M, P) :-
sum(M, K, P),
prod3(M, N, K).
Note that if we swap M,K (i.e. sum(K,M,P)), when prod3 is called with P unknown we again have a non terminating loop, but in sum.
?- prod3(X,Y,s(s(s(s(s(s(0))))))).
X = s(s(s(s(s(s(0)))))),
Y = s(0) ;
X = s(s(s(0))),
Y = s(s(0)) ;
X = s(s(0)),
Y = s(s(s(0))) ;
X = s(0),
Y = s(s(s(s(s(s(0)))))) ;
false.
OT I'm perplexed by cTI report: prod3(A,B,C)terminates_if b(A),b(B);b(A),b(C).

How to count the number of children of parents in prolog (without using lists) in prolog?

I have the following problem. I have a certain number of facts such as:
parent(jane,dick).
parent(michael,dick).
And I want to have a predicate such as:
numberofchildren(michael,X)
so that if I call it like that it shows X=1.
I've searched the web and everyone puts the children into lists, is there a way not to use lists?
Counting number of solutions requires some extra logical tool (it's inherently non monotonic). Here a possible solution:
:- dynamic count_solutions_store/1.
count_solutions(Goal, N) :-
assert(count_solutions_store(0)),
repeat,
( call(Goal),
retract(count_solutions_store(SoFar)),
Updated is SoFar + 1,
assert(count_solutions_store(Updated)),
fail
; retract(count_solutions_store(T))
),
!, N = T.
I can only see two ways to solve this.
The first, which seems easier, is to get all the solutions in a list and count it. I'm not sure why you dislike this option. Are you worried about efficiency or something? Or just an assignment?
The problem is that without using a meta-logical predicate like setof/3 you're going to have to allow Prolog to bind the values the usual way. The only way to loop if you're letting Prolog do that is with failure, as in something like this:
numberofchildren(Person, N) :- parent(Person, _), N is N+1.
This isn't going to work though; first you're going to get arguments not sufficiently instantiated. Then you're going to fix that and get something like this:
?- numberofchildren(michael, N)
N = 1 ;
N = 1 ;
N = 1 ;
no.
The reason is that you need Prolog to backtrack if it's going to go through the facts one by one, and each time it backtracks, it unbinds whatever it bound since the last choice point. The only way I know of to pass data across this barrier is with the dynamic store:
:- dynamic(numberofchildrensofar/1).
numberofchildren(Person, N) :-
asserta(numberofchildrensofar(0)),
numberofchildren1(Person),
numberofchildrensofar(N), !.
numberofchildren1(Person) :-
parent(Person, _),
retract(numberofchildrensofar(N)),
N1 is N + 1,
asserta(numberofchildrensofar(N1),
!, fail.
numberofchildren1(_).
I haven't tested this, because I think it's fairly disgusting, but it could probably be made to work if it doesn't. :)
Anyway, I strongly recommend you take the list option if possible.

Resources