1: \begin{definition}

2: Assume $|\lambda|<C_\nu$. The \textit{limiting probability space} $(\Omega,\Sigma,P_\lambda)$ is defined as follows.

3: 

4: 1. As before, $\Omega=\left\lbrace 

5: \omega\mid\omega:\mathbb{Z}^\nu\rightarrow \{-1,1\}\right\rbrace.$ 

6: 

7: 2. The sigma-algebra $\Sigma$ of events consists of all countable unions of the sets: 

8: \[M_{T,A}=\{

9: \omega\in\Omega\mid\omega(t_1)=a_1,\ldots,\omega(t_m)=a_m\}\]

10: for all pairs of sets $T=\{t_1,t_2,\ldots,t_m\}\subset\mathbb{Z}^\nu$ and $A=\{a_1,a_2,\ldots,a_m\}\subseteq \{-1,1\}$, $m\geqslant 0$. It is sufficient to take only $A\subseteq \{-1,1\}$ because for other $A\subset\mathbb{R}$, $M_{T,A}=\varnothing$.

11: 

12: 3. For any pair of $T=\{t_1,t_2,\ldots,t_m\}\subset\mathbb{Z}^\nu$ and $A=\{a_1,a_2,\ldots,a_m\}\subseteq \{-1,1\}$ we define the probability of $M_{T,A}$ by the following:

13: \begin{equation}

14: P_\lambda(M_{T,A})=\frac{(-1)^k}{2^m}\sum_{T'\subseteq T}f(Q_{T'})\prod_{\{i:t_i\in T\setminus T'\}}a_i,

15: \label{eq:limit_prob}

16: \end{equation}

17: where $f$ is defined in Theorem \ref{theorem:therm_limit}.2) and $k$ is the number of $a_i\in A$ that equal -1.

18: 

19: The formula \eqref{eq:limit_prob} is given in \cite{MM91}; it generates the probability measure  $P_\lambda$ on all events in $\Sigma$, which is called the \textit{limiting Gibbs measure}.

20: 

21: This completes the definition of the limiting probability space.

22: \label{def:probability_space}

23: \end{definition}

24: