DM841 - Heuristics and Constraint Programming for Discrete Optimization
[pdf format]



Exercise 1

Recall the definition of domain consistency and relaxed domain consistency such as bound(Z), bound(D) and range consistency. What is the consistency level of the following CSPs?

  P = ⟨ x1 ∈ {1,3}, x2 ∈ {1,3}, x3 ∈ {1,3}, x4 ∈ {1,3},  C≡ alldifferent(x1,x2,x3,x4)⟩



Solution


The weakiest form is bound(Z) consistency. Accordingly we should ask whether the bounds of x
i, i=1,2,3,4 have a bounded support. For x1=1 we can have x2=2, x3=3 but no value would be left for x4. Hence, P is not bound(Z) consistent. Since this is the weakiest form of those asked all the others are also not satisfied.

  P = ⟨ x1 ∈ {1,2,3}, x2 ∈ {2,3}, x3 ∈ {2,3}, x4 ∈ {1,2,3,4},  Calldifferent(x1,x2,x3,x4)⟩



Solution


Same as above, it is not consistent in any of the forms.

 P = ⟨ x1 ∈ {1,3}, x2 ∈ {2}, x3 ∈ {1,2,3},  C≡ alldifferent(x1,x2,x3)⟩



Solution


It is bound(Z). For being bound(D) it must be that for all x
i, i=1,2,3 its bounds belong to a support. This holds true for this case. For being range consistent all values of xi i=1,2,3 must belong to a bounded support. This is not true since x3=2 has no support in x2. Hence it is also not arc consistent.

 P = ⟨ x1 ∈ {1,3}, x2 ∈ {2}, x3 ∈ {1,3},  C≡ alldifferent(x1,x2,x3) ⟩



Solution


It is bound(Z) and bound(D). Since 2 is not anymore in the domain of x
3 then it is also range consistent.

  P = ⟨ x1 ∈ {1,3}, x2 ∈ {1,3}, x3 ∈ {1,3},  C≡ alldifferent(x1,x2,x3)⟩



Solution


bound(Z): yes, bounded support hence we can reintroduce the value 2 in the support variables

bound(D): no.

range: yes, bounded support hence we can reintroduce the value 2 in the support variables