Suppose $\{X_t\}_{t \in \mathbb{Z}} \subseteq \mathbb{R}^d$ is a $\alpha$- or $\rho$-mixing process. Let $h (x, y) : \mathbb{R}^d \times \mathbb{R}^d \rightarrow \mathbb{R}$ be the symmetric kernel function. Then \begin{equation*} V_n = \frac{1}{n^2} \sum_{s = 1}^n \sum_{t = 1}^n h(X_s, X_t) \end{equation*} is a $V$-statistic of order 2.

I do find an article that involves this question. Leucht and Neumann show that $V_n - \frac{1}{n} \sum_{k = 1}^{\infty} \lambda_k Z_k^2 = o_p \left( \frac{1}{n}\right)$ where $\{ Z_k\}_{k = 1}^{\infty}$ is a sequence of independent standard normal random variables. But this representation may not be practicable when using it: the eigenvalue $\lambda_k$ is the solution of $\mathrm{E} \left[h (x, X_{0}) \Phi_k (X_{0})\right]=\lambda_k \Phi_k(x)$ with $(\Phi_{k})_{k}$ are associated orthonormal eigenfunctions satisfying \begin{equation*} h(x, y)=\sum_{k} \lambda_{k} \Phi_{k}(x) \Phi_{k}(y), \qquad \mathrm{E}\left[\Phi_{j}\left(X_{0}\right) \Phi_{k}\left(X_{0}\right)\right]=\delta_{j k}, \end{equation*} it is extremely difficult to find those $\lambda_k$! However, these exists some simple methods to find the asymptotic expression for $V_n$ when $d = 1$, and we do not need to find these $\lambda_k$!

Could anyone provide me any other literature regarding the situation $d > 1$ with a practical method to analyze $V_n$? Thank you so much in advance!!


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.