Comments start with a # on the line.
Suppose players 1,..,n and each of them can say "yes" or "no". Each such player i is represented by a Boolean variable xi which can be 1 ("yes") or 0 ("no").
Now consider a set of m ≥-constraints and non-negative integer coefficients wi,j ("weights"), where i is the player and j is the constraint:
One such constraint is called a Weighted Voting Game (WVG) and a satisfying coalition is called a winning coalition (of players). If a coalition is winning only if is satisfies all m constraints, then it is called Multiple Weights Voting Game (MWVG).
To model a MWVG as input for the laboratory the constraints have to be transposed as shown below. The main advantage of this representation is that we can now add the names of the players. You should keep in mind that the input is player-oriented:
As an example, consider the following two constraints:
# Right hand sides: 2 3 # Weights and names. Here: A, B, C 1 3 A 1 2 B 1 1 C
A class line has the form
(x<multiplier>) <weights> (<name>)where (...) indicates an optional part. The multiplier indicates the number of players in the class. For instance, x5 indicates 5 players. If no multiplier is given, x1 is assumed.
The <weights> has to provide a non-negative integer weight for each constraint.
The <name> is an optional name for the player. There are no restrictions for its form. The name ends at the end of the line.
The join command provides a Boolean expression which describes how to join the constraints. The variables are the m constraints numbered from 1 to m. Boolean operations are AND and OR. Negation is not allowed. The parser respects precedence of conjunction over disjunction and recognizes brackets.
Multiple occurrences of a variable are allowed. See also the example of the US Federal Legal System below.
# UN Security Council - First version 39 x5 7 Permanent Member x10 1 Non-permanent MemberIf we do care on the names of the permanent members we could model:
# UN Security Council - Second version 39 7 United States 7 Russia 7 China 7 France 7 United Kingdom x10 1 Non-permanent MemberEven though the weighted representation is well known for this game, the fact that we express two different rules (veto power and "9 out of 15") in a single constraint it not optimal. For example, consider the following model which uses two constraints:
# UN Security Council - Third version 5 9 x5 1 1 Permanent Member x10 0 1 Non-permanent MemberThis model is much more obvious in that the first constraint models veto power of the permanent members and the second models the "9 out of 15" rule.
We now consider a more complicated example, viz. the US Federal Legal System. The players are the 100 members of the senate, the 435 members of the house of representatives, the president of the United States and the vice president of the United States which is the president of the senate. A bill passes if at least one of the following conditions is satisfied:
%join (1 AND 2 AND 3) OR (1 AND 4 AND 5 AND 3) OR (6 AND 7) 1 51 218 1 50 67 290 x100 0 1 0 0 1 1 0 Member of the Senate x435 0 0 1 0 0 0 1 Member of the House of Rep. 1 0 0 0 0 0 0 President 0 0 0 1 0 0 0 Vice-President
∀ S ∈ 2N: ∑i ∈ S wi ≥ Q ⇔ S ∈ W.
In this case, the Q is called the quota and w1,…,wi are called the weights (of the players). For S∈ 2N we define w(S):=∑i∈ S wi. A weighted representation [Q;w1,…,wn] is called homogeneous if for each minimal winning coalition S it holds w(S)=Q. For instance, [8;6,4,3,1] and [5;3,2,2,1] represent the same game, the former representation is not homogeneous while the latter is homogeneous. When we use vectors instead of scalar integers for the weights and for the quota we end up at multiple weighted voting games (Also known as vector-weighted voting games). Operations like + and ≥ are defined component-wise for this. For instance, the game [(2,3);(1,2),(2,1),(0,2),(1,1)] is not weighted.i≽I j ⇔ ∀ S ⊆ N\ {i,j}: S ∪ {j} ∈ W ⇒ S ∪ {i} ∈ W
Two players i,j ∈ N are called symmetric (denoted by i≈Ij) if i≽I j and j ≽I i. A player i ∈ N is called ...∃ S ∈ Wmin, i ∈ S: ∃ T ⊆ N \ S: (∀ j ∈ T: i ≽I j) ∧ (S \ {i}) ∪ T ∈ W.
There is a minimal winning coalition containing i in which i can be replaced by some not more desirable players such that the resulting coalition is winning again.∃ S ∈A : ∀ k=1,...,t : mk = |S ∩ Nk|.
Representions by models are often applied to the sets of the winning, the minimal winning and the shift-minimal winning coalitions. A simlar statement holds for the corresponding sets of the losing coalitions.