1: \begin{definition}\label{def:Improved Outer Bound-vector}

2: Consider a set of correlated sources $Y_s, s\in {\mathcal S}$ with underlying probability distribution $P_{Y_{\mathcal S}}(\cdot)$. From this distribution, construct binary partition random variables

3: $A_{\langle \alpha \rangle}, \langle \alpha \rangle \in \Omega$ from partitions of $\mathcal Y_{\mathcal S}$ as described in Theorem \ref{thm2}. 

4: %

5: Let $\mathcal R'_{\mathrm{cs}}(\overline{\Gamma^*})$ be the set of all link capacity tuples $\mathbf{{C}} = (C_{e}: e \in \mathcal E)$ such that there exists an almost entropic function $ h  \in \overline{\Gamma^*}$ for the set $\sessions \cup \edges \cup \Omega$  satisfying the constraints

6: \begin{align*}

7: h (\alpha, \beta)-H(Y_{s}, s \in \alpha, A_{i}, i \in \beta)&=0\\

8: h ({e}| f \in \sessions\cup\edges \:: f \rightarrow  {e}  )  &= 0\\

9: h (s| s' \in \sessions  : u \in a(s'), f \in\edges : f \rightarrow u   ) &=0 \\

10: h ({e}) &\leq C_{e} 

11: \end{align*}

12: for all $\alpha \subseteq \mathcal S, \beta \subseteq \Omega, s \in \sessions, u \in b(s)$ and $e\in \mathcal E$. Replacing $\overline{\Gamma^*}$ by $\Gamma$ in Definition \ref{def:Improved Outer Bound-vector} we obtain an \emph{explicitly computable outer bound} $\mathcal R'_{\mathrm{cs}}(\Gamma)$.

13: \end{definition}

14: