- 26 January 2016: Complete rewrite of the section about the Structural Implementation of Fibonacci.
In my 100 days of Fibonacci project I looked at functional and imperative language. There are however many more programming paradigms today we are going to dive into one of the most used logic programming languages.
Prolog is heavily used in the field of artificial intelligence. This is due to the natural way of formulating computations in terms of predicates. This allows for concise expression of tree structures that certain data might exhibit.
In this blog post, i show how to implement Fibonacci in the directly recursive style. As we will see this gives problems when we want to exploit some Prolog features.
Day 4 - Prolog
In Prolog, I defined a predicate that holds both the argument and the result. A predicate is something that can only be true or false. Therefore, the notion of return value is explicit parameters to the predicates in Prolog.
% Implementation using direct recursion.fib_direct(0, 0).fib_direct(1, 1).fib_direct(N, R) :- X1 is N-1, X2 is N-2, fib_direct(X1, A), fib_direct(X2, B), R is A + B.
In above the base cases are defined as simple predicates. The first argument is the nth Fibonacci number and the second is the result.
fib_direct(0, 0). we denote that the predicate predicate
fib_direct is true when applied 0 for both arguments. In
(SWI Prolog interpreter) we can write
fib_direct(0, X). to make
it derive the second argument given the first argument.
$ swipl fib.pl % fib.pl compiled 0.00 sec, 4 clausesWelcome to SWI-Prolog ...For help, use ?- help(Topic). or ?- apropos(Word).?- fib_direct(0, R).R = 0 .
The third definition of the
fib_direct predicate is the one that
allows us to calculate Fibonacci for arbitrary numbers. We can again
make prolog derive the result by supplying N
...?- fib_direct(10, R).R = 55 .
A problem here is that we use the
is/2 notation. It enforces evaluation
of the right side and puts the result in the variable on the left side.
This is something we want to avoid in logical programming, as we need the
right side to be defined. The reason is that we can not choose what to derive anymore.
Yesterday I put up a little challenge to try to figure out where 55 came from. This challenge is ideal to solve in Prolog as we can use the same code to derive what n is for a given Fibonacci number. We would have done something like.
?- fib_direct(N, 55).ERROR: fib_direct/2: Arguments are not sufficiently instantiated
But as you can see it fails. This is because of the use of
is-operator does explicit evaluation of the right-hand side
before it check whether it equals the left-hand side.
A Structural Implementation of Fibonacci
To remove the use of that operator I have implemented Peano natural numbers and defined what addition is. This allows us to implement Fibonacci in a purely structural manner.
Fibonacci is a very simple function that requires 3 constructions: a data type (natural numbers), data aggregation (addition), and a way to define the recursive dependencies. That last thing is done through recursively defined predicates in Prolog.
% Definiition of the Peano numbersnat(0).nat(s(_)).% Readable natsnat_to_int(0, 0).nat_to_int(s(N), R) :- nat_to_int(N, R1), R is R1 + 1.int_to_nat(0, 0).int_to_nat(X, s(R)) :- X1 is X-1, int_to_nat(X1, R).
Above is the data type. The first three lines are somewhat unnecessary as Prologs typing discipline is implicit.
The next two block display how we convert between the arithmetic integers and the naturals. We use pattern matching an a recursively defined predicate. This construction is quite idiomatic for this problem.
Net we have an implementation of addition. Again we use pattern matching in two defined predicates. The first predicate is the base case while the second is the induction case.
% Definition of addition ( add(A, B, Result ))add(0, B, B).add(s(A), B, R) :- add(A, s(B), R).
To understand what is happening one can think of two stacks of plates where we move all plates piecewise from the left stack to the right. When the left stack is empty, we stop and let the right one be the result.
The last part in this implementation is the definition of Fibonacci. Note that we implicitly subtract the number by 2 when we match in the induction case. This is necessary as we don't have any functionality for subtracting. General subtraction is not necessary as we only need to subtract by 2. In the body of the predicate we add by 1, to get the subtraction of 1.
fib_peano(0, 0).fib_peano(s(0), s(0)).fib_peano(s(s(N)), R) :- fib_peano(N, A), fib_peano(s(N), B), add(A, B, R).% Wrapper functions dealing with all the conversionsfib_peano_wrapper_get_res(A, B) :- int_to_nat(A, X), fib_peano(X, Y), nat_to_int(Y, B).fib_peano_wrapper_get_arg(A, B) :- int_to_nat(B, Y), fib_peano(X, Y), nat_to_int(X, A).
The two last blocks are for reading arithmetic arguments. They simply
convert a number to the Peano structure before it run the
predicate. Lastly it converts back to arithmetic integers to make it
easier to read the output.
First we take a look at how it would look without conversion from integers. It is apparent that it is rather cumbersome to practically work with numbers in this format.
?- fib_peano(s(s(s(s(s(s(s(s(s(s(0)))))))))), R), nat_to_int(R, X).R = s(s(s(s(s(s(s(s(s(s(...)))))))))),X = 55 .
The last piece of code for this post demonstrates how the wrappers work. the first example simply returns the 10th Fibonacci number. The second example is more interesting. It lets us derive which index the Fibonacci number 55 has.
?- fib_peano_wrapper_get_res(10, B).B = 55 .?- fib_peano_wrapper_get_arg(A, 55).A = 10 .
The last piece of code also solves the puzzle from yesterday where I asked what n is for $fib(n) = 55$.
The time and space complexity is generally not good when using this representation. It is usually only used when reasoning about programs, i.e. proving different properties about programs. The idea about structuring problems in terms of nothing but predicates is very usable, and is indeed one of the reasons why Prolog is widely used.
As usual the files are available on Github.
Programming Through Predicates
Logic programming has its core in predicates. We define logical propositions and make the subsystem derive what we need. This works very well when we work on structures like lists and trees.
The Fibonacci function, however, works on natural numbers which we usually represent as integers for direct translation to machine instructions.
As seen above it was indeed possible to implement naturals structurally. On a more general note, it is actually possible to model all problems in a structural manner. This is exploited in amongst other languages Coq.
Actually, all problems may be expressed in structures. This we will get back to when discussing theorem provers.
Today I looked at Prolog. I implemented the Fibonacci function in a logic context. We saw that We didn't exploit Prologs ability to derive the argument to Fibonacci. To circumvent that, I built a structural implementation of natural, using Peano, so we didn't have to make an explicit evaluation. This implementation had the ability to derive what number a given Fibonacci number is, even though it is not practical.