Joint distribution of two normal marginal distributions

553 Views Asked by At

My question is related to the possibility of stating joint convergence in distribution from marginal weak convergence.

Consider two sequences of random vectors $X_n$ and $Y_n$ defined on the probability space $(\Omega, \mathcal{F}, P)$. Let $X,Y$ be other two random vectors on $(\Omega, \mathcal{F}, P)$. All random vectors take values in $\mathbb{R}^k$.

Let "$\rightarrow_d$" denote convergence in distribution.

General statement: $X_n\rightarrow_dX$ and $Y_n\rightarrow_dY$ does not imply $(X_n,Y_n)\rightarrow_d (X,Y)$

Specific case: Assume $X_n\rightarrow_dX\sim N(0,J)$, $Y_n=h^TX_n$ with $h\in \mathbb{R}^k$ and $J$ is a $k\times k$ matrix. Hence, $Y_n=h^TX_n\rightarrow_d h^TX \sim N(0,h^TJh)$. Is it correct to write down $(X_n,Y_n)\rightarrow_d(X,h^TX)$?

Where I found this specific case: I think it is used in the proof of Theorem 9.4 in van der Vaart "Asymptotic Statistics" p. 128 here where the author says

"By Assumption, the sequence $(\Delta_n,h_0^T\Delta_n)$ converges in distribution under $h_0=0$ to a vector $(\Delta,h^T\Delta)$... "

Did I misinterpret the proof? Any hint would be really appreciated