A1133
Title: Statistical inference and eigenvector fluctuations for random graphs with infinite rank kernels
Authors: Minh Tang - North Carolina State University (United States) [presenting]
Joshua Cape - University of Michigan (United States)
Abstract: The problem of estimating the leading eigenvectors for independent edge random graphs generated from a latent position model whose link function is possibly of infinite rank is considered. Error bounds in 2 to infinity norm are derived and row-wise normal approximations for these eigenvectors. These results are applied to the two-sample testing problem in which a pair of vertices have the same latent positions. A test statistic that converges to a weighted sum of independent chi-square is proposed under the null hypothesis.
