Condition for convergence of maxima of random triangular arrays (Q2488438)

A triangular array of r.v.s \(\{X_{kn},k\leq n,n\geq 1\}\) is considered with i.i.d. \(X_{kn}\) for fixed \(n\). Let the CDF of \(X_{kn}\) be \(F_n\). Then \[ {\mathbf P}\left\{\Bigl(\sup_{1\leq i\leq n} X_{in}-b_n\Bigr)/a_n\leq x\right\} =(F_n(a_n x+b_n))^n. \] Locally uniform convergence of \((F_n(a_n x+b_n))^n\) to \(G(x)\) and convergence of the first and the second derivatives are investigated. Sufficient conditions for convergence of normalizing sequences \(a_n\) and \(b_n\) are derived. The Burr distribution is considered as an example.