Question

I am new to Prolog and I have some probably simple issue with a piece of code. This is a real world problem that arose last Friday and believe me this is not a CS Homework.

We want to print business cards and these can only be printed in blocks of 900 cards (100 sheets with 9 cards per sheet). The cards for anybody should not be distributed over several blocks. People ordered different amount of cards, E.G:

% "braucht" is german and means "needs"
braucht(anton,400).
braucht(berta,200).
braucht(claudia,400).
braucht(dorothee,100).
braucht(edgar,200).
braucht(frank,400).
braucht(georg,100).

I put together the following definition to find an appropriate block of 900 business cards:

block(0,[]).
block(N,[H|T]) :-
    braucht(H,Nh),
%   \+(member(H,T)),
    D is N - Nh,
    D >= 0,
    block(D,T).

This produces a nice list of blocks of people whose cards fit together on a 900 cards block. But it stops working if I activate the commented line “\+member….” and just gives me a “false”. But I need to assure that nobody is getting more than once on that block. What am I doing wrong here?

Answer 1

It seems that what you want to achieve is to set a constraint that H does not appear in the tail T of the list. However, T is still unbound when you call member/2, so that member(H, T) will succeed and hence \+ member(H,T) will fail.

If you don’t want to use Constraint Programming, but use pure Prolog instead, you should use the check in the other direction and check whether H is already present in the list of people that has been aggregated up to that point. Something like:

block(0, List, List).
block(N, Rest, List) :-
  braucht(H, Nh),
  \+(memberchk(H, Rest)), % will fail when H is already in Rest
  D is N-Nh,
  D >= 0,
  block(D, [H|Rest], List).

The predicate block/3 can be called from a predicate block/2:

block(N, List) :-
  block(N, [], List).