1: \begin{definition}\label{def:c_g_by_observable} (Coarse-graining defined by an observable)

2: Let $\hat{A}=\sum_ia_i\P_{a_i}$, where $a_i\neq a_j$ for $i\neq j$, be the spectral decomposition of an observable\footnote{More generally, we can use any Hermitian matrix $\hat{A}$ to define a coarse-graining since every Hermitian matrix has a spectral decomposition. Later in Theorems~\ref{thm:bounded} and \ref{thm:bounded_multiple} we will use coarse-graining $\C_\R$ given by the density matrix $\R$ although the density matrix is not usually considered as to be an observable in a physical sense.} $\hat{A}$. We define coarse-graining given by the observable $\hat{A}$ as ${\C_{\hat{A}}=\{\P_{a_i}\}_{a_i}}$. We say that coarse-graining $\C$ commutes with an observable $\hat{A}$ if $[\P_k,\hat{A}]=0$ for all $\P_k\in\C$.

3: \end{definition}

4: