Let $X_1,X_2,...$ and $Y_1,Y_2,...$ be two sequences of random variables such that $\mathcal{L}(X_n)\xrightarrow{w}\mathcal{L}(X)$ and $|Y_n-X_n|\xrightarrow{w}0$ as $n\rightarrow\infty$ Show that $$\mathcal{L}(Y_n)\xrightarrow{w}\mathcal{L}(X)$$ as $n\rightarrow\infty$.

I have the next attempt:

I know that convergence in probability implies convergence weakly or in distribution, then given $\epsilon>0$ I can say that:


Also since $\mathcal{L}(X_n)\xrightarrow{w}\mathcal{L}(X)$, I can say that $Ef(X_n)\rightarrow Ef(X)$ for a bounded Lipschitz function using the Portamnteau theorem.

I don't know how to use these facts to conclude.


1 Answer 1


The Portmanteau theorem works here. Let $f$ be bounded and Lipschitz, so there exists $B>0$ such that $|f(x)|\le B$ for all $x$, and there exists $K>0$ such that $|f(y)-f(x)|\le K|y-x|$ for all $y$ and $x$.

To show that $Ef(Y_n)\to Ef(X)$, start with the triangle inequality: $$ |Ef(Y_n)-Ef(X)|\le E|f(Y_n)-f(X_n)|+|Ef(X_n)-Ef(X)|\tag1 $$ The second term on the RHS of (1) tends to zero since $X_n$ converges weakly to $X$. To show the first term tends to $0$, pick $\epsilon>0$ and break up the expectation $E|f(Y_n)-f(X_n)|$ according to the cases $|Y_n-X_n|>\epsilon$ and $|Y_n-X_n|\le\epsilon$. In the first case use the fact that $f$ is bounded to see $$ E\left[|f(Y_n)-f(X_n)|I(|Y_n-X_n|>\epsilon)\right]\le 2B\,P(|Y_n-X_n|>\epsilon),\tag2 $$ which tends to zero because $Y_n-X_n$ converges in probability to zero. In the second case use the fact that $f$ is Lipschitz: $$E\left[|f(Y_n)-f(X_n)|I(|Y_n-X_n|\le\epsilon)\right]\le E K|Y_n-X_n|I(|Y_n-X_n|\le\epsilon)\le K\epsilon.\tag3 $$ But $\epsilon$ is arbitrary positive, so the result follows.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.