The Random Walk Laplacian
We show how the random walk laplacian $L_{rw} = I - P$ is related to the Graph Laplacian $L = D - A$.
In the previous posts we discussed the graph laplacian
\[L = D - A\]where $D$ and $A$ are the degree and adjacency matrices respectively. The random walk Laplacian however is expressed as
\[L_{rw} = I - P\]where $P$ is the random-walk transition matrix on the graph.
While this may seem puzzling at first glance, $L_{rw}$ turns out to simply be $L$ normalized by the degrees of the nodes.
\[\begin{aligned} L_{rw} &= D^{-1}L \\ &= D^{-1}(D - A) \\ &= D^{-1}D - D^{-1}A \\ &= I - P \end{aligned}\]Some other properties about $L_{rw}$:
- $L_{rw}$ has the same eigenvectors as $P$. If $λ$ is an eigenvalue of $L_{rw}$, then $1 - λ$ is an eigenvalue of $P$:
- The stationary distribution of $x$ is defined as $x = xP$. This can be now rewritten as $xL_{rw} = 0$:
- There is, unfortunately, no immediate relationship between the eigenvectors of $L$ and $L_{rw}$. (Relevant Math StackExchange)
This post is an elaboration of a Math StackExchange answer by Misha Lavrov.
This page was last modified on January 3, 2026.