Convergence Almost Surely does not imply convergence of moments
# Statement
Let $(\Omega, \mathcal{M}, \mathbb{P})$ be a Probability Space. There exists a sequence of Random Variables $(X_n){n=1}^{\infty}$ on $\Omega$ s.t. $X{n} \to X$ but $\mathbb{E}|X_{n}| \not\to \mathbb{E}|X|$.
# Proof
We construct the example. Let $([0,1], \mathcal{B}([0,1]), \lambda)$ be our probability space. Suppose $X_{n}$ is defined as $$X_{n} = 0*1_{x \leq 1 - \frac{1}{n}} + 7n1_{x > 1 - \frac{1}{n}}$$
Then $X_{n} \to 0$ almost surely. To see this, observe that for $s \in [0, 1)$, there exists $N \in \mathbb{N}$ s.t. $1 - \frac{1}{N} > s$ (by the Archimedean property of the Natural Numbers). For all $n \geq N$, we have $\frac{1}{n} \leq \frac{1}{N}$ so $$X_{n}(s) = 0*1_{x \leq 1 - \frac{1}{n}}(s) + 7n1_{x > 1 - \frac{1}{n}}(s) = 0$$
Thus $\mathbb{P}[X_{n} \to 0] = \lambda([0, 1)) = 1 \checkmark$ .
On the other hand, observe that
$$\mathbb{E}|X_{n}| = 0*\lambda([0, 1 - \frac{1}{n}]) + 7n \lambda\left(\left(1-\frac{1}{n}, 1\right)\right) = 0 + \frac{7n}{n} = 7$$
While
$$\mathbb{E}|X| = 0$$
Since $7 \not\to 0$ we have that the Moments do not converge. $\blacksquare$
# Remarks
- In this example, all random variables are non-negative, so the construction also shows that almost sure convergence does not imply expectation converges.
- Any form of convergence implied by almost sure (Convergence in Distribution, Convergence in Probability) will also have this counterexample.
- This is an example where Fatou’s Lemma is strict.
- This example also works as a general example for generic Measure Spaces.