A counterexample to the existence of a general central limit theorem for pairwise independent identically distributed random variables (Q2660453)

The classical central limit states that the standardized sum of a sequence of \(n\) mutually independent and identically distributed random variables with finite variance converges in distribution to a standard Gaussian distribution, i.e., we have \((S_{n}-n\mu )/\sigma \sqrt{n}\Longrightarrow Z\sim N(0,1)\). In the paper the authors construct a sequence of pairwise independent random variables having a common distribution \(F\). Under mild conditions on \(F\) they prove that \((S_{n}-n\mu )/\sigma \sqrt{n} \Longrightarrow \sqrt{1-r^{2}}Z+rU(l-1)\), where \(r\) and \(l\) are explicitly given and where \(U(l-1)\) is independent of \(Z\) and has a standardized chi-square distribution with parameter \(l-1\). The proof involves a detailed analysis of the characteristic function of the sum.