Latest on accelerating pagerank compuatation...

[Subbarao Kambhampati <rao@asu.edu>]

Here is a paper that appeared in the last WWW that talks about how the 
iterative page-rank computation can be speeded up by adjusting the page 
rank vector every iteration by subtracting off estimates for non-primary 
eigen vectors  (remember that the PR corresponds to the primary eigen 
vector). [I am guessing that it makes sense in that it further reduces the 
components of the vector in the non-primary directions, thus making it 
converge to primary direction faster.]

Completely optional--and useful only to quench your curiosity (such as you 
may have it ;-)

Rao

http://nlp.stanford.edu/cmanning/papers/extrapolation.pdf