1: \begin{definition}{$A(B,\Sigma)$}

2: 

3: The set $A(B,\Sigma) \subset \{ 1,\ldots,G\}$ denotes indices $g$ for which $b_g$ has nonvanishing

4: coefficient $\alpha_g$ in one of the convex combinations

5: $M_\Sigma(B)=\sum_{g=1,\ldots,G} \alpha_g b_g$ with $\alpha_g \ge 0$,

6: $\sum_{g=1,\ldots,G} \alpha_g =1$. Note that by the assumptions of Theorem \ref{coverage}

7: or Theorem \ref{maintheorem} the $\alpha_g$ are unique for $M_{\Sigma^0}(B^0)$.

8: 

9: \end{definition}

10: