0
$\begingroup$

Let $(\Omega,\mathcal{F},\mathbb{P})$ be a probability space, $\{X_n\}_{n\in \mathbb{N}}$ is a sequence of random variables in $L^1$.which converges to $X$ in probability and a constant $M>0 $ such that $||X_n||_1\leq M $.

Show that : $$ \mathbb {E}(|X|)\leq \liminf_{n\rightarrow\infty}(\mathbb {E}(|X_n|))\leq \sup_{n}(\mathbb {E}(|X_n|)) $$

I am thinking of applying the Fatou theorem to $\{|X_n|\}_{n\in \mathbb{N}}$, but I cannot think of a bound.

$\endgroup$

2 Answers 2

0
$\begingroup$

You don't require any bound on $E|X_n|$ for this. You can prove it using Fatou's Lemma.

Suppose $E|X| >\lim \inf E|X_n|$. Then we can find a real number $c$ and integers $n_1<n_2<...$ such that $E|X| >c >E|X_{n_k}|$ for all $k$. Since $X_{n_k} \to X$ in probability there is a subsequence, call it $Y_j$ which converges to $X$ almost everywhere. Bu then $E|X| >c >E|Y_j|$ for all $j$ and $Y_j$ converges to $X$ almost everywhere. This is a contradiction to Fatou's Lemma.

$\endgroup$
0
$\begingroup$

Choose a subsequence $(n_k)_{j\geq 1}$ such that

$$\liminf_{n\to\infty} \mathbb{E}[|X_n|] = \lim_{k\to\infty} \mathbb{E}[|X_{n_k}|]$$

holds. Since $X_{n_k} \to X$ in probability, we may find a further subsequence $(n'_j)_{j\geq 1} \subseteq (n_k)_{k\geq 1}$ such that $X_{n'_j} \to X$ almost surely as $j\to\infty$. Then by the Fatou's lemma,

$$ \liminf_{n\to\infty} \mathbb{E}[|X_n|] = \lim_{j\to\infty} \mathbb{E}[|X_{n'_j}|] \geq \mathbb{E}\Bigl[ \liminf_{j\to\infty} |X_{n'_j}| \Bigr] = \mathbb{E}[|X|]. $$

$\endgroup$
2
  • $\begingroup$ Convergence in $L^{1}$ is not in the hypothesis. $\endgroup$ Commented Nov 5, 2019 at 23:58
  • $\begingroup$ @KaboMurphy, Thank you for pointing out my mistake. I was careless when parsing OP's question. I fixed my answer accordingly. $\endgroup$ Commented Nov 5, 2019 at 23:59

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .