Density convergence in the Breuer-Major theorem for Gaussian stationary sequences (Q888483)

Based on quotations taken from the article: Let \(X=\{X_k: k\in \{0,1,2,\dots\}\}\) be a centered Gaussian stationary sequence with unit variance. For all \(v\in \{0,\pm 1,\pm 2,\dots\}\) set \(\rho=\operatorname{E}[X_0X_{|v|}]\). Therefore, \(\rho(0)=1\) and \(|\rho(v)|\leq 1\) for all \(v\). Let \(\gamma\) be the standard Gaussian probability measure and \(f\in L^2(\gamma)\) be a fixed deterministic function such that \(\operatorname{E}[f(X_1)]=0\). Expand \(f\) in the orthonormal basis of Hermite polynomials \(\{H_k;~k\geq 0\}\), where \(H_k\) is the \(k\)-th Hermite polynomial given by \[ H_k(x)= (-1)^k e^{x^2/2} \frac{d^k}{dx^k} e^{-x^2/2}. \] In particular, if \(f\) has Hermite rank \(d\geq 1\), it admits the following expansion: \[ f(x)=\sum_{j=d}^\infty a_j H_j(x), \] with \(a_d\not = 0\). Define a weighted sum of the form \[ V_n=\frac{1}{\sqrt{n}}\sum_{k=0}^{n-1}f(X_k). \] Assume that the central limit theorem holds for the sum \(V_n\). Then, the authors prove (the main result of the paper) that the uniform convergence of the density of \(V_n\) towards the standard Gaussian density also holds, under a mild additional assumption involving the causal representation of \(X\).