New to discourse, so sorry if this isn’t the right format. For some specific application, I need to perform the Power Iteration method to find the stationary distribution of a random walk on a graph. I found that it is possible, but not recommended, to perform this method in
IGPersonalizedPageRank. Will this be available in the future? Why is it deprecated?
I can in fact run it (just doing
IGPersonalizedPageRank[myGraph, myNode, myDecayParam, Method -> "PowerIteration"]), and I get a result, but I also get some errors:
As a side note, the result I get with this code differs from the result I get with no method specified. Also, my graph edges have no weights, so the error about weights is weird, is this referring to the edge weights? As I said, what I need for my application is the “intermediate” steps in the iteration process, so that’s the reason I would be interested in having this method implemented, as well as an option to allow for edge weights (even though in my current examples edges have no weights, I could use that in the future).
In summary, I would like to know if it is planned to implement/improve/curate the iteration method. Also, I would appreciate some intuition behind the sub-options usage (Number of Iterations, and Epsilon). I understand Epsilon is some threshold for convergence, so, is it necessary to give both parameters or one/the-other?