Sufficient condition for $L^1$-convergence of random variables

54 Views Asked by At

Assume that $X_n$ is a sequence of non-negative random variables such that

$$ \Bbb E[X_n] \to c $$

for some constant $c > 0$. From this we want to deduce the $L^1$-convergence

$$ \Bbb E[|X_n-c|] \to 0. $$

Now, finding a bound for this term turned out to be quite the challenge and I'm not really sure how I should prove this, or whether this is even true. Triangle inequality, reverse triangle inequality, none of these things work. I'm at a loss here. Can someone help me out?

1

There are 1 best solutions below

0
On BEST ANSWER

Example for measure space, that the claim is not true:

Take $X_n = 1_{[n, n+1)}$ in a $(\mathbb{R}, \mathcal{B}(\mathbb{R}), \lambda)$ measure space.

$\mathbb{E}(X_n) = 1 \underset{n \to \infty}{\to} 1$

$\mathbb{E}(\lvert X_n - 1 \rvert) = \infty \neq 0$

Example in probability space: Let $(\Omega, \Sigma) = ([0,1], \mathcal{B}([0,1]))$ with Lebesgue measure on it.

Let $X_n = n \cdot 1_{\left[0, \frac{1}{n}\right)}$.

$\mathbb{E}(X_n) = 1 \underset{n \to \infty}{\to} 1$, but

$\mathbb{E}(\lvert X_n - 1 \rvert) \to 2 \neq 0$.