Bounds for characteristic functions of polynomials in asymptotically normal random variables

Let \(X: \Omega\to \mathbb{R}^k\) be a random variable on a probability space \((\Omega,B,P)\) and let \(X_1,X_2, \dots\) be a sequence of independent copies of \(X\). Denote \(S_n= {1\over n} \sum^n_{k=1} X_k\). The characteristic function (Fourier transform) \[ g_n (t,a) =\mathbb{E} \exp \bigl\{it f(S_n+a) \bigr\},\quad a\in \mathbb{R}^k, \] is studied, where \[ f(x)= \sum_{ m_1=0}^m \cdots \sum^m_{m_k=0} \alpha(m_1, \dots, m_k)x_1^{m_1} \dots x_k^{m_k}, \quad m \geq 2,\;x= (x_1, \dots, x_k)\in \mathbb{R}^k. \] In case \(k=1\) and \(f\) a polynomial \(f(x)= \alpha x^m+ \dots\), for any non-degenerated r.v. \(X\) and any \(A>0\), \(\varepsilon>0\), \(m\geq 2\) constants \(c_1\) and \(c_2\) exist (of course, depending on \(A,\varepsilon,m\) and distribution of \(X)\) such that \[ \bigl|g_n(t,a) \bigr|\leq c_1 \bigl(n^{-A} +|\alpha t|^{-1/m2^m} \bigr) \] provided that \(c_2|\alpha t|\leq n^{(m-1)/(2-\varepsilon)}\). It is shown, that the exponent \(1/m2^m\) of \(|t|^{-1}\) can be replaced by a larger number in certain cases. Bounds for \(g\) in Gaussian case in \(\mathbb{R}^k\) are also obtained.

Recommendations

Cited in

(13)

This page was built for publication: Bounds for characteristic functions of polynomials in asymptotically normal random variables

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4341898)