What's Still Missing/Unachievable Let's consider a random walk on grap Construct n *n adjacency matrix M Normalize W=D-1/2MD-1/2(D: degree matrix) One step random walk: p+l= Wpt Stationary distribution follows: p= Wp PageRankWhat’s Still Missing/Unachievable? • Let’s consider a random walk on graph – Construct 𝑛 ∗ 𝑛 adjacency matrix 𝐌 – Normalize 𝐖 = 𝐃 −𝟏/𝟐𝐌𝐃 −𝟏/𝟐 (𝐃: degree matrix)) – One step random walk: 𝐩 𝑡+1 = 𝐖𝐩 𝑡 – Stationary distribution follows: 𝐩 = 𝐖𝐩 17