Samples of iid Lognormals: Approximations for Characteristics of Minima

We concentrate on characteristics of minima X-N from samples of iid lognormals X-1i approximate to 2pLND. We demonstrate that the distribution of X-N for 2 N 1,000 may be fitted more accurately by 2pLND than by the limiting Gumbel distribution. An extended power model is established to represent the quotients C-N/C-1, where C-N is the mean, standard deviation, or p-quantile of X-N and C-1 is the corresponding characteristic of X-1i. Our empirical comparisons show that this model provides not only more accurate estimates than alternating approximations but it is also much simpler than its competitors.