Issues with basic rules in Prolog

Question

I want to translate my natural code in prolog.

Here is my natural code :

if E < W then
   if C < Max then
      print('increment C')
   else
      print('C unchanging')
if E > W then
   if C > Min then
      print('decrement C')
   else
      print('error')

Here is my code in SWI-Prolog :

tddv_worst(_).
tddv_best(_).
tddv_courant(_).
tddv_estime(_).
etendue_gisement_courant(_).
etendue_gisement_min(_).
etendue_gisement_max(_).

compromis_1(tddv_estime(E),tddv_worst(W),etendue_gisement_courant(C),etendue_gisement_min(Min),etendue_gisement_max(Max)) :-
(E < W , C < Max) -> writeln('Increment C') ; writeln('C unchanging') ;
(E > W , C > Min) -> writeln('Decrement C') ; writeln('Error').

When I write this in the interpreter :

compromis_1(tddv_estime(19),tddv_worst(16),etendue_gisement_courant(2),etendu e_gisement_max(8),etendue_gisement_min(1)).

It displays C unchanging.

It's not supposed to display this.

Can someone explain me why ?

Answer 1

Your implementation isn't really equivalent to the C version you are showing.

Here's your current code:

(E < W , C < Max) -> writeln('Increment C') ;
writeln('C unchanging') ;
(E > W , C > Min) -> writeln('Decrement C') ;
writeln('Error').

Prolog sees this as a series of four clauses related by a logical disjunction (OR -- ;). If the first -> expression fails, it then calls the next disjunctive expression, writeln('C unchanging'). The above code is roughly equivalent to:

if E < W and C < Max then
   print('increment C')
else
   print('C unchanging')

if E > W and C > Min then
   print('decrement C')
else
   print('error')

A Prolog equivalent of your original pseudo-code would be:

(   E < W
->  (   C < Max
    ->  writeln('increment C')
    ;   writeln('C unchanging')
    )
) ;
(   E > W
->  (   C > Min
    ->  writeln('decrement C')
    ;   writeln('Error')
    )
)

In Prolog, this behavior happens to work like the pseudo-code since -> eliminates the choice point if it succeeds and Prolog will not backtrack.