Moving averages of random vectors with regularly varying tails (Q2703261)

The authors consider a moving average process \(X_t=\sum_{j=-\infty}^{+\infty} C_jZ_j\), where \(Z_j\), \(j=-\infty,\dots,+\infty\), is a sequence of i.i.d. random vectors with common regularly varying distribution \(\mu\) with index \(E\) (\(E\) is a matrix!). \(C_j\) are real matrices. Conditions are obtained under which NEWLINE\[NEWLINEA_n(X_1+\dots+X_n-na_n)\Rightarrow U,NEWLINE\]NEWLINE for some \(a_n\in R^d\) and a regularly varying sequence of matrices \(A_n\), \(U\) being a full-dimensional operator stable distribution. These conditions include the claim that real parts of \(E\) eigenvalues should be \(>1/2\). The limit behavior of the sample covariance matrix of \(X_t\) is also considered.