Can graph spectrums be derived from incident matrices?

[1Truthseeker]

Will the set of eigenvalues of an incident matrix derive an equivalent notion of a graph spectrum as it does with an adjacency matrix?

Specifically:

Let s_a be the set of eigenvalues of an adjacency matrix for graph G.

And,

Let s_i be the set of eigenvalues of an incident matrix for graph G.

What are the differences between s_a and s_i? Could these both be considered spectrums of the graph? If not, why?

Thanks much!

 

[spikedmath]

What is your definition for incident matrix? If it's the usual definition, then it's likely not a square (nxn) matrix so it doesn't make sense to talk about eigenvalues. You can however look at its singular values which I think are related to eigenvalues of the line graph of G.