Magic Sequence

Problem

A magic sequence of length n is a sequence of integers x0 . . xn-1 between 0 and n-1, such that for all i in 0 to n-1, the number i occurs exactly xi times in the sequence. For instance, 6,2,1,0,0,0,1,0,0,0 is a magic sequence since 0 occurs 6 times in it, 1 occurs twice, ... – CSPLib

Code

 1 require 'rubygems'
 2 require 'gecoder'
 3 
 4 # Solves the magic sequence problem.
 5 class MagicSequence
 6   include Gecode::Mixin
 7 
 8   # n is the length of the sequence.
 9   def initialize(n)
10     # The i:th variable represents the value of the i:th element in the 
11     # sequence.
12     sequence_is_an int_var_array(n, 0...n)
13     
14     # The basic requirement to qualify as a magic sequence.
15     n.times{ |i| sequence.count(i).must == sequence[i] }
16   
17     # The following are implied constraints. They do not affect which 
18     # assignments are solutions, but they do help prune the search space 
19     # quicker.
20     
21     # The sum must be n. This follows from that there are exactly n elements and
22     # that the sum of all elements are the number of occurrences in total, i.e. 
23     # the number of elements.
24     sequence.sum.must == n
25     
26     # sum(seq[i] * (i-1)) must equal 0 because sum(seq[i]) = n as seen above
27     # and sum(i*seq[i]) is just another way to compute sum(seq[i]). So we get 
28     # sum(seq[i] * (i-1)) = sum(seq[i]) - sum(i*seq[i]) = n - n = 0
29     sequence.zip((-1...n).to_a).map{ |element, c| element*c }.sum.must == 0
30     
31     branch_on sequence, :variable => :smallest_degree, :value => :split_max
32   end
33   
34   def to_s
35     sequence.values.join(', ')
36   end
37 end
38 
39 class Array
40   # Sums all the elements in the array using #+ .
41   def sum
42     inject{ |sum, element| sum + element }
43   end
44 end
45 
46 puts MagicSequence.new(500).solve!.to_s

Output

496, 2, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0

Notes

This might be a bit boring problem to use constraint programming on as the pattern for sequences is obvious for n >= 7, making it trivial to construct an efficient algorithm.