pagerank-analyzer

Provide the graph as a sparse adjacency matrix in COO format with values, row indices, and column indices, then call the pageRank tool with a damping factor and convergence epsilon. The sublinear algorithm handles graphs with millions of nodes efficiently.

Personalized PageRank biases the random walk toward a specific preference vector, making it useful for recommendation systems. Pass a personalized vector alongside the adjacency matrix to rank items relative to a particular user's interests.

Yes, computing PageRank over an agent communication graph identifies hub nodes and bottlenecks. The scores guide reconnection strategies that improve load distribution and communication efficiency in distributed swarms.

The computation expects a sparse adjacency matrix, typically in COO format with explicit row count, column count, edge weights, and index arrays. Dense matrices can be converted to sparse representations for memory efficiency.

Sublinear algorithms trade exactness for speed, producing approximations bounded by the epsilon parameter. For small graphs where exact computation is cheap, traditional iterative methods may be simpler and more precise.