graph shift operator

Data
Graph Shift Opterator

We can define more general diffusion processes by defining the matrix SRn×n such that

Sij0(i,j)E

(can only equal 0 along the diagonal?)

This matrix is called the graph shift operator (GSO). We way z=Sx is a (graph) "shift" or diffusion of x by S.

The most common operators are the adjacency matrix and the graph laplacian

Note that we can always recover the local implementations:

zi=jSijxj=jNiSijxj
Examples of Graph Shift Operators

Mentions

File
2025-01-22 graphs lecture 1
2025-01-27 graphs lecture 2
2025-01-29 graphs lecture 3
2025-02-03 graphs lecture 4
2025-02-05 graphs lecture 5
2025-02-12 graphs lecture 7
2025-02-19 graphs lecture 9
2025-02-25 equivariant lecture 4
2025-03-03 graphs lecture 11
2025-03-05 graphs lecture 12
2025-03-10 graphs lecture 13
2025-03-26 lecture 15
2025-03-31 lecture 16
2025-04-02 lecture 17
additive perturbations
conditions for finding a convolutional graph filter
convolutional filter bank
convolutional graph filters are permutation equivariant
convolutional graph filters are shift equivariant
cycle homomorphism density is given by the trace of the adjacency matrix
discriminability of a graph filter
filter permutation invariance
fixed coefficients yield the same spectral response for both graphon and graph convolutions
graph attention model
graph automorphism
graph convolution
graph fourier transform
graph shift operators converge to graphon shift operators
Graph Signals and Graph Signal Processing
Improved Image Classification with Manifold Neural Networks
inverse graph fourier transform
linear graph filter
Lipschitz filters are stable to additive perturbations
operator dilation
spectral graph filter
spectral representation of a convolutional graph filter
spectral representation of graphon convolutions
stable graph filter
the spectral representation of a graph filter is independent from the graph