this is for holding javascript data
ZoƩ Christoff edited untitled.tex
about 8 years ago
Commit id: 92c709cd560d5812e847e697a1b6157161ee67fd
deletions | additions
diff --git a/untitled.tex b/untitled.tex
index 82a24b6..ace8e09 100644
--- a/untitled.tex
+++ b/untitled.tex
...
The above gives a characterization of the class of influence profiles on which all opinion profiles converge. But we can also give a full characterization of convergence, i.e, we can also characterize the class of opinion profiles which converge for graphs that do \emph{not} belong on this class:
\begin{proposition}{}
Let $\G=(G_{p_1},\dots,G_{p_m})$ be an influence
profile and $\O=(O_1,\dots,O_n)$ be an opinion profile. Then the following are equivalent:
\begin{itemize}
\item[] The BDP converges for
an opinion profile $\O$ on $\G$.
\item[] For some $l\in \{1,\ldots,m\}$, there is a set of agents $C\subseteq\N$ such that: $C$ is a cycle in $G_{p_l}$ and there are two agents $i,j\in C$ such that $O_i(p_l)\neq O_j(p_l)$.
\end{itemize}
\end{proposition}
...
Below is yet another way to put it:
\begin{proposition}{}
Let $\G=(G_{p_1},\dots,G_{p_m})$ be an influence profile and
$\O$ $\O=(O_1,\dots,O_n)$ be an opinion profile. Then the following are equivalent:
\begin{itemize}
\item[] The BDP converges for $\O$ on $\G$.
\item[] For all $j\in \{1,\ldots,m\}$, $G_{p_j}$ contains no cycle of length $\geq 2$ or for some $l\in \{1,\ldots,m\}$, there is a set of agents $C\subseteq\N$ such that: $C$ is a cycle in $G_{p_l}$ and there are two agents $i,j\in C$ such that $O_i(p_l)\neq O_j(p_l)$.