Hands-On Mathematics for Deep Learning

Graph Laplacian

Earlier in this chapter, in the Adjacency matrix section, we learned about the adjacency matrix and how we can use it to tell what the structure of a graph is. However, there are other ways of representing graphs in matrix form.

Now, let's suppose we have an undirected, unweighted graph. Then, its Laplacian matrix will be a symmetric n × n matrix, L, whose elements are as follows:

Here, . We can also write this as follows: