Skip to content

Commit f1191c6

Browse files
committed
Merge branch 'ebunle/main'
2 parents ec94c00 + 8c4810d commit f1191c6

1 file changed

Lines changed: 46 additions & 0 deletions

File tree

blueprint/src/content.tex

Lines changed: 46 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -573,6 +573,52 @@ \subsection{Total Expectation and Probability}
573573
\end{align*}
574574
\end{proof}
575575

576+
577+
%% -- Non-Degeneracy
578+
\subsection{Non-Degeneracy}
579+
\begin{theorem}[Non-degeneracy of $L_1$-norm] \label{thm:non-degeneracy}
580+
Let $(\Omega,\mathcal{F},\PP)$ be a discrete probability space with $\Omega=\{\omega_1,\omega_2,\ldots\}$ countable, and let $p_i=P(\{\omega_i\})\geq0$ with $\sum_i p_i =1$. Let $X$ be a random variable with $X_i=X(\omega_i)$. Then
581+
\[
582+
\E[|X|]=0 \Longleftrightarrow \PP(\{\omega\in\Omega:X(\omega)=0\})=1.
583+
\]
584+
\end{theorem}
585+
586+
\begin{proof}
587+
\textit{Proof of ($\Leftarrow$):} Assume $\PP(X=0)=1$.
588+
If $\PP(X=0)=1$, then $\PP(X\neq0)=0$.
589+
In discrete terms $p_i=0$ for all $\omega_i$ such that $X_i\neq0$.
590+
Now compute $\E[|X|]$ such that
591+
\[
592+
\E[|X|]=\sum_i |X_i| \cdot p_i.
593+
\]
594+
595+
Case 1: $X_i=0\Rightarrow |X_i|p_i=0\cdot p_i=0$.\\
596+
Case 2: $X_i\neq0\Rightarrow p_i=0\Rightarrow |X_i|p_i=|X_i|\cdot0=0$.
597+
Every term is zero, so $\E[|X|]=0$.
598+
599+
\textit{Proof of ($\Rightarrow$):} Assume $\E[|X|]=0$.
600+
We have
601+
\[
602+
\E[|X|]=\sum_i |X_i| \cdot p_i=0,
603+
\]
604+
where $|X_i|p_i\geq0$ for all $i$. For a sum of nonnegative terms to be zero, each term must be zero
605+
\[
606+
|X_i|p_i=0 \quad \text{for all } i.
607+
\]
608+
Thus, for each $i$, either $p_i=0$ or $|X_i|=0$ (i.e., $X_i=0$).
609+
Let $N=\{\omega_i:X_i\neq0\}$.
610+
For $\omega_i\in N$, we have $X_i\neq0\Rightarrow |X_i|\neq0\Rightarrow$ from $|X_i|p_i=0$ we must have $p_i=0$.
611+
Therefore:
612+
\[
613+
\PP(X\neq0)=\sum_{\omega_i\in N} p_i =\sum_{\omega_i\in N} 0=0.
614+
\]
615+
Thus $\P(X=0)=1-\PP(X\neq0)=1$. Hence, we conclude that
616+
\[
617+
\E[|X|]=0 \Longleftrightarrow \PP(X=0)=1.
618+
\]
619+
\end{proof}
620+
621+
576622
\section{Formal Decision Framework}
577623

578624
\subsection{Markov Decision Process}

0 commit comments

Comments
 (0)