This paper by Sepandar Kamvar, Taher Haveliwala, and Gene Golub describes an algorithm to speed up the computation of PageRank using the fact that pages converge at different rates.
This paper by Sepandar Kamvar and Taher Haveliwala proves analytically the second eigenvalue of the Google Matrix, which has implications for the PageRank algorithm.
Abstract, slide show, references, from First Steps on the Road to Reinventing Computing, by Ian Piumarta, Viewpoints Research Institute. Stanford EE Computer Systems Colloquium, 2007 Feb 14.