laplacian matrices of graphs - cs.yale. · pdf filelaplacian matrices of graphs: spectral and...
TRANSCRIPT
![Page 1: Laplacian Matrices of Graphs - cs.yale. · PDF fileLaplacian Matrices of Graphs: Spectral and Electrical Theory Daniel A. Spielman ... Spectral graph theory A very fast survey Trailer](https://reader031.vdocuments.net/reader031/viewer/2022022500/5aa4690d7f8b9a517d8be0b8/html5/thumbnails/1.jpg)
Laplacian Matrices of Graphs: Spectral and Electrical Theory
Daniel A. Spielman Dept. of Computer Science
Program in Applied Mathematics Yale University
Toronto, Sep. 28, 2011
![Page 2: Laplacian Matrices of Graphs - cs.yale. · PDF fileLaplacian Matrices of Graphs: Spectral and Electrical Theory Daniel A. Spielman ... Spectral graph theory A very fast survey Trailer](https://reader031.vdocuments.net/reader031/viewer/2022022500/5aa4690d7f8b9a517d8be0b8/html5/thumbnails/2.jpg)
Outline
Introduction to graphs
Physical metaphors
Laplacian matrices
Spectral graph theory
A very fast survey
Trailer for lectures 2 and 3
![Page 3: Laplacian Matrices of Graphs - cs.yale. · PDF fileLaplacian Matrices of Graphs: Spectral and Electrical Theory Daniel A. Spielman ... Spectral graph theory A very fast survey Trailer](https://reader031.vdocuments.net/reader031/viewer/2022022500/5aa4690d7f8b9a517d8be0b8/html5/thumbnails/3.jpg)
Graphs and Networks V: a set of vertices (nodes) E: a set of edges an edge is a pair of vertices
Dan
Donna Allan
Gary Maria
Nikhil
Shang-Hua
Difficult to draw when big
![Page 4: Laplacian Matrices of Graphs - cs.yale. · PDF fileLaplacian Matrices of Graphs: Spectral and Electrical Theory Daniel A. Spielman ... Spectral graph theory A very fast survey Trailer](https://reader031.vdocuments.net/reader031/viewer/2022022500/5aa4690d7f8b9a517d8be0b8/html5/thumbnails/4.jpg)
Examples of Graphs
![Page 5: Laplacian Matrices of Graphs - cs.yale. · PDF fileLaplacian Matrices of Graphs: Spectral and Electrical Theory Daniel A. Spielman ... Spectral graph theory A very fast survey Trailer](https://reader031.vdocuments.net/reader031/viewer/2022022500/5aa4690d7f8b9a517d8be0b8/html5/thumbnails/5.jpg)
Examples of Graphs
8
5
1
7 3 2
9 4 6
10
![Page 6: Laplacian Matrices of Graphs - cs.yale. · PDF fileLaplacian Matrices of Graphs: Spectral and Electrical Theory Daniel A. Spielman ... Spectral graph theory A very fast survey Trailer](https://reader031.vdocuments.net/reader031/viewer/2022022500/5aa4690d7f8b9a517d8be0b8/html5/thumbnails/6.jpg)
Examples of Graphs
8
5
1
7 3 2
9 4 6
10
![Page 7: Laplacian Matrices of Graphs - cs.yale. · PDF fileLaplacian Matrices of Graphs: Spectral and Electrical Theory Daniel A. Spielman ... Spectral graph theory A very fast survey Trailer](https://reader031.vdocuments.net/reader031/viewer/2022022500/5aa4690d7f8b9a517d8be0b8/html5/thumbnails/7.jpg)
How to understand a graph
Use physical metaphors Edges as rubber bands Edges as resistors
Examine processes Diffusion of gas Spilling paint
Identify structures Communities
![Page 8: Laplacian Matrices of Graphs - cs.yale. · PDF fileLaplacian Matrices of Graphs: Spectral and Electrical Theory Daniel A. Spielman ... Spectral graph theory A very fast survey Trailer](https://reader031.vdocuments.net/reader031/viewer/2022022500/5aa4690d7f8b9a517d8be0b8/html5/thumbnails/8.jpg)
How to understand a graph
Use physical metaphors Edges as rubber bands Edges as resistors
Examine processes Diffusion of gas Spilling paint
Identify structures Communities
![Page 9: Laplacian Matrices of Graphs - cs.yale. · PDF fileLaplacian Matrices of Graphs: Spectral and Electrical Theory Daniel A. Spielman ... Spectral graph theory A very fast survey Trailer](https://reader031.vdocuments.net/reader031/viewer/2022022500/5aa4690d7f8b9a517d8be0b8/html5/thumbnails/9.jpg)
Nail down some ver8ces, let rest se<le
View edges as rubber bands or ideal linear springs spring constant 1 (for now)
Graphs as Spring Networks
![Page 10: Laplacian Matrices of Graphs - cs.yale. · PDF fileLaplacian Matrices of Graphs: Spectral and Electrical Theory Daniel A. Spielman ... Spectral graph theory A very fast survey Trailer](https://reader031.vdocuments.net/reader031/viewer/2022022500/5aa4690d7f8b9a517d8be0b8/html5/thumbnails/10.jpg)
Nail down some ver8ces, let rest se<le
View edges as rubber bands or ideal linear springs spring constant 1 (for now)
Graphs as Spring Networks
When stretched to length poten8al energy is
�
�2/2
![Page 11: Laplacian Matrices of Graphs - cs.yale. · PDF fileLaplacian Matrices of Graphs: Spectral and Electrical Theory Daniel A. Spielman ... Spectral graph theory A very fast survey Trailer](https://reader031.vdocuments.net/reader031/viewer/2022022500/5aa4690d7f8b9a517d8be0b8/html5/thumbnails/11.jpg)
Nail down some ver8ces, let rest se<le.
Physics: posi8on minimizes total poten8al energy
subject to boundary constraints (nails)
x(a)a
1
2
�
(a,b)∈E
(x(a)− x(b))2
Graphs as Spring Networks
![Page 12: Laplacian Matrices of Graphs - cs.yale. · PDF fileLaplacian Matrices of Graphs: Spectral and Electrical Theory Daniel A. Spielman ... Spectral graph theory A very fast survey Trailer](https://reader031.vdocuments.net/reader031/viewer/2022022500/5aa4690d7f8b9a517d8be0b8/html5/thumbnails/12.jpg)
Nail down some ver8ces, let rest se<le
Energy minimized when free ver8ces are averages of neighbors
x(a)a
�x(a) =1
da
�
(a,b)∈E
�x(b)
Graphs as Spring Networks
is degree of , number of attached edges da a
![Page 13: Laplacian Matrices of Graphs - cs.yale. · PDF fileLaplacian Matrices of Graphs: Spectral and Electrical Theory Daniel A. Spielman ... Spectral graph theory A very fast survey Trailer](https://reader031.vdocuments.net/reader031/viewer/2022022500/5aa4690d7f8b9a517d8be0b8/html5/thumbnails/13.jpg)
If nail down a face of a planar 3-‐connected graph, get a planar embedding!
Tutte’s Theorem ‘63
![Page 14: Laplacian Matrices of Graphs - cs.yale. · PDF fileLaplacian Matrices of Graphs: Spectral and Electrical Theory Daniel A. Spielman ... Spectral graph theory A very fast survey Trailer](https://reader031.vdocuments.net/reader031/viewer/2022022500/5aa4690d7f8b9a517d8be0b8/html5/thumbnails/14.jpg)
3-‐connected: cannot break graph by cuRng 2 edges
Tutte’s Theorem ‘63
![Page 15: Laplacian Matrices of Graphs - cs.yale. · PDF fileLaplacian Matrices of Graphs: Spectral and Electrical Theory Daniel A. Spielman ... Spectral graph theory A very fast survey Trailer](https://reader031.vdocuments.net/reader031/viewer/2022022500/5aa4690d7f8b9a517d8be0b8/html5/thumbnails/15.jpg)
3-‐connected: cannot break graph by cuRng 2 edges
Tutte’s Theorem ‘63
![Page 16: Laplacian Matrices of Graphs - cs.yale. · PDF fileLaplacian Matrices of Graphs: Spectral and Electrical Theory Daniel A. Spielman ... Spectral graph theory A very fast survey Trailer](https://reader031.vdocuments.net/reader031/viewer/2022022500/5aa4690d7f8b9a517d8be0b8/html5/thumbnails/16.jpg)
Tutte’s Theorem ‘63 3-‐connected: cannot break graph by cuRng 2 edges
![Page 17: Laplacian Matrices of Graphs - cs.yale. · PDF fileLaplacian Matrices of Graphs: Spectral and Electrical Theory Daniel A. Spielman ... Spectral graph theory A very fast survey Trailer](https://reader031.vdocuments.net/reader031/viewer/2022022500/5aa4690d7f8b9a517d8be0b8/html5/thumbnails/17.jpg)
Tutte’s Theorem ‘63 3-‐connected: cannot break graph by cuRng 2 edges
![Page 18: Laplacian Matrices of Graphs - cs.yale. · PDF fileLaplacian Matrices of Graphs: Spectral and Electrical Theory Daniel A. Spielman ... Spectral graph theory A very fast survey Trailer](https://reader031.vdocuments.net/reader031/viewer/2022022500/5aa4690d7f8b9a517d8be0b8/html5/thumbnails/18.jpg)
Graphs as Resistor Networks
View edges as resistors connecting vertices
Apply voltages at some vertices. Measure induced voltages and current flow.
1V
0V
![Page 19: Laplacian Matrices of Graphs - cs.yale. · PDF fileLaplacian Matrices of Graphs: Spectral and Electrical Theory Daniel A. Spielman ... Spectral graph theory A very fast survey Trailer](https://reader031.vdocuments.net/reader031/viewer/2022022500/5aa4690d7f8b9a517d8be0b8/html5/thumbnails/19.jpg)
Graphs as Resistor Networks
View edges as resistors connecting vertices
Apply voltages at some vertices. Measure induced voltages and current flow.
Current flow measures strength of connection between endpoints.
More short disjoint paths lead to higher flow.
![Page 20: Laplacian Matrices of Graphs - cs.yale. · PDF fileLaplacian Matrices of Graphs: Spectral and Electrical Theory Daniel A. Spielman ... Spectral graph theory A very fast survey Trailer](https://reader031.vdocuments.net/reader031/viewer/2022022500/5aa4690d7f8b9a517d8be0b8/html5/thumbnails/20.jpg)
Graphs as Resistor Networks
View edges as resistors connecting vertices
Apply voltages at some vertices. Measure induced voltages and current flow.
1V
0V
0.5V
0.5V
0.625V 0.375V
![Page 21: Laplacian Matrices of Graphs - cs.yale. · PDF fileLaplacian Matrices of Graphs: Spectral and Electrical Theory Daniel A. Spielman ... Spectral graph theory A very fast survey Trailer](https://reader031.vdocuments.net/reader031/viewer/2022022500/5aa4690d7f8b9a517d8be0b8/html5/thumbnails/21.jpg)
Graphs as Resistor Networks
View edges as resistors connecting vertices
Apply voltages at some vertices. Measure induced voltages and current flow.
Induced voltages minimize �
(a,b)∈E
(v(a)− v(b))2
Subject to fixed voltages (by battery)
![Page 22: Laplacian Matrices of Graphs - cs.yale. · PDF fileLaplacian Matrices of Graphs: Spectral and Electrical Theory Daniel A. Spielman ... Spectral graph theory A very fast survey Trailer](https://reader031.vdocuments.net/reader031/viewer/2022022500/5aa4690d7f8b9a517d8be0b8/html5/thumbnails/22.jpg)
Learning on Graphs [Zhu-‐Ghahramani-‐Lafferty ’03]
Infer values of a function at all vertices from known values at a few vertices.
Minimize
Subject to known values
�
(a,b)∈E
(x(a)− x(b))2
0
1
![Page 23: Laplacian Matrices of Graphs - cs.yale. · PDF fileLaplacian Matrices of Graphs: Spectral and Electrical Theory Daniel A. Spielman ... Spectral graph theory A very fast survey Trailer](https://reader031.vdocuments.net/reader031/viewer/2022022500/5aa4690d7f8b9a517d8be0b8/html5/thumbnails/23.jpg)
Learning on Graphs [Zhu-‐Ghahramani-‐Lafferty ’03]
Infer values of a function at all vertices from known values at a few vertices.
Minimize
Subject to known values
�
(a,b)∈E
(x(a)− x(b))2
0
1 0.5
0.5
0.625 0.375
![Page 24: Laplacian Matrices of Graphs - cs.yale. · PDF fileLaplacian Matrices of Graphs: Spectral and Electrical Theory Daniel A. Spielman ... Spectral graph theory A very fast survey Trailer](https://reader031.vdocuments.net/reader031/viewer/2022022500/5aa4690d7f8b9a517d8be0b8/html5/thumbnails/24.jpg)
The Laplacian quadratic form �
(a,b)∈E
(x(a)− x(b))2
![Page 25: Laplacian Matrices of Graphs - cs.yale. · PDF fileLaplacian Matrices of Graphs: Spectral and Electrical Theory Daniel A. Spielman ... Spectral graph theory A very fast survey Trailer](https://reader031.vdocuments.net/reader031/viewer/2022022500/5aa4690d7f8b9a517d8be0b8/html5/thumbnails/25.jpg)
xTLx =�
(a,b)∈E
(x(a)− x(b))2
The Laplacian matrix of a graph
![Page 26: Laplacian Matrices of Graphs - cs.yale. · PDF fileLaplacian Matrices of Graphs: Spectral and Electrical Theory Daniel A. Spielman ... Spectral graph theory A very fast survey Trailer](https://reader031.vdocuments.net/reader031/viewer/2022022500/5aa4690d7f8b9a517d8be0b8/html5/thumbnails/26.jpg)
The Laplacian matrix of a graph
To minimize subject to boundary constraints, set derivative to zero.
Solve equation of form
xTLx =�
(a,b)∈E
(x(a)− x(b))2
Lx = b
![Page 27: Laplacian Matrices of Graphs - cs.yale. · PDF fileLaplacian Matrices of Graphs: Spectral and Electrical Theory Daniel A. Spielman ... Spectral graph theory A very fast survey Trailer](https://reader031.vdocuments.net/reader031/viewer/2022022500/5aa4690d7f8b9a517d8be0b8/html5/thumbnails/27.jpg)
Weighted Graphs
Edge assigned a non-negative real weight measuring strength of connection spring constant 1/resistance
wa,b ∈ R
xTLx =�
(a,b)∈E
wa,b(x(a)− x(b))2
(a, b)
![Page 28: Laplacian Matrices of Graphs - cs.yale. · PDF fileLaplacian Matrices of Graphs: Spectral and Electrical Theory Daniel A. Spielman ... Spectral graph theory A very fast survey Trailer](https://reader031.vdocuments.net/reader031/viewer/2022022500/5aa4690d7f8b9a517d8be0b8/html5/thumbnails/28.jpg)
Weighted Graphs
Edge assigned a non-negative real weight measuring strength of connection spring constant 1/resistance
wa,b ∈ R
xTLx =�
(a,b)∈E
wa,b(x(a)− x(b))2
(a, b)
I’ll show the matrix entries tomorrow
![Page 29: Laplacian Matrices of Graphs - cs.yale. · PDF fileLaplacian Matrices of Graphs: Spectral and Electrical Theory Daniel A. Spielman ... Spectral graph theory A very fast survey Trailer](https://reader031.vdocuments.net/reader031/viewer/2022022500/5aa4690d7f8b9a517d8be0b8/html5/thumbnails/29.jpg)
Measuring boundaries of sets
Boundary: edges leaving a set
S S
![Page 30: Laplacian Matrices of Graphs - cs.yale. · PDF fileLaplacian Matrices of Graphs: Spectral and Electrical Theory Daniel A. Spielman ... Spectral graph theory A very fast survey Trailer](https://reader031.vdocuments.net/reader031/viewer/2022022500/5aa4690d7f8b9a517d8be0b8/html5/thumbnails/30.jpg)
Measuring boundaries of sets
Boundary: edges leaving a set
S S
![Page 31: Laplacian Matrices of Graphs - cs.yale. · PDF fileLaplacian Matrices of Graphs: Spectral and Electrical Theory Daniel A. Spielman ... Spectral graph theory A very fast survey Trailer](https://reader031.vdocuments.net/reader031/viewer/2022022500/5aa4690d7f8b9a517d8be0b8/html5/thumbnails/31.jpg)
Measuring boundaries of sets
Boundary: edges leaving a set
S
0 0
0 0
0 0
1
1 0
1 1
1 1 1
0
0 0
1
S
Characteristic Vector of S:
x(a) =
�1 a in S
0 a not in S
![Page 32: Laplacian Matrices of Graphs - cs.yale. · PDF fileLaplacian Matrices of Graphs: Spectral and Electrical Theory Daniel A. Spielman ... Spectral graph theory A very fast survey Trailer](https://reader031.vdocuments.net/reader031/viewer/2022022500/5aa4690d7f8b9a517d8be0b8/html5/thumbnails/32.jpg)
Measuring boundaries of sets
Boundary: edges leaving a set
S
0 0
0 0
0 0
1
1 0
1 1
1 1 1
0
0 0
1
S
Characteristic Vector of S:
x(a) =
�1 a in S
0 a not in S
xTLx =�
(a,b)∈E
(x(a)− x(b))2 = |boundary(S)|
![Page 33: Laplacian Matrices of Graphs - cs.yale. · PDF fileLaplacian Matrices of Graphs: Spectral and Electrical Theory Daniel A. Spielman ... Spectral graph theory A very fast survey Trailer](https://reader031.vdocuments.net/reader031/viewer/2022022500/5aa4690d7f8b9a517d8be0b8/html5/thumbnails/33.jpg)
Cluster Quality
S
0
0 0
0 0
1
1 0
1 1
1 1 1
0
0 0
1
S
Number of edges leaving S Size of S
=|boundary(S)|
|S|def= Φ(S) (sparsity)
![Page 34: Laplacian Matrices of Graphs - cs.yale. · PDF fileLaplacian Matrices of Graphs: Spectral and Electrical Theory Daniel A. Spielman ... Spectral graph theory A very fast survey Trailer](https://reader031.vdocuments.net/reader031/viewer/2022022500/5aa4690d7f8b9a517d8be0b8/html5/thumbnails/34.jpg)
Cluster Quality
S
0
0 0
0 0
1
1 0
1 1
1 1 1
0
0 0
1
S
Number of edges leaving S Size of S
=|boundary(S)|
|S|def= Φ(S)
The Rayleigh Quotient of with respect to x L
(sparsity)
=xTLx
xTx=
�(a,b)∈E(x(a)− x(b))2
�a x(a)
2
![Page 35: Laplacian Matrices of Graphs - cs.yale. · PDF fileLaplacian Matrices of Graphs: Spectral and Electrical Theory Daniel A. Spielman ... Spectral graph theory A very fast survey Trailer](https://reader031.vdocuments.net/reader031/viewer/2022022500/5aa4690d7f8b9a517d8be0b8/html5/thumbnails/35.jpg)
Spectral Graph Theory A n-by-n symmetric matrix has n real eigenvalues and eigenvectors such that
λ1 ≤ λ2 · · · ≤ λn
v1, ..., vn
Lvi = λivi
![Page 36: Laplacian Matrices of Graphs - cs.yale. · PDF fileLaplacian Matrices of Graphs: Spectral and Electrical Theory Daniel A. Spielman ... Spectral graph theory A very fast survey Trailer](https://reader031.vdocuments.net/reader031/viewer/2022022500/5aa4690d7f8b9a517d8be0b8/html5/thumbnails/36.jpg)
Spectral Graph Theory A n-by-n symmetric matrix has n real eigenvalues and eigenvectors such that
λ1 ≤ λ2 · · · ≤ λn
v1, ..., vn
These eigenvalues and eigenvectors tell us a lot about a graph!
Theorems Algorithms Heuristics
Lvi = λivi
![Page 37: Laplacian Matrices of Graphs - cs.yale. · PDF fileLaplacian Matrices of Graphs: Spectral and Electrical Theory Daniel A. Spielman ... Spectral graph theory A very fast survey Trailer](https://reader031.vdocuments.net/reader031/viewer/2022022500/5aa4690d7f8b9a517d8be0b8/html5/thumbnails/37.jpg)
The Rayleigh Quotient and Eigenvalues A n-by-n symmetric matrix has n real eigenvalues and eigenvectors such that
λ1 ≤ λ2 · · · ≤ λn
v1, ..., vn
Courant-Fischer Theorem:
λ1 = minx �=0
xTLx
xTxv1 = argmin
x �=0
xTLx
xTx
Lvi = λivi
![Page 38: Laplacian Matrices of Graphs - cs.yale. · PDF fileLaplacian Matrices of Graphs: Spectral and Electrical Theory Daniel A. Spielman ... Spectral graph theory A very fast survey Trailer](https://reader031.vdocuments.net/reader031/viewer/2022022500/5aa4690d7f8b9a517d8be0b8/html5/thumbnails/38.jpg)
The Courant Fischer Theorem
![Page 39: Laplacian Matrices of Graphs - cs.yale. · PDF fileLaplacian Matrices of Graphs: Spectral and Electrical Theory Daniel A. Spielman ... Spectral graph theory A very fast survey Trailer](https://reader031.vdocuments.net/reader031/viewer/2022022500/5aa4690d7f8b9a517d8be0b8/html5/thumbnails/39.jpg)
The Courant Fischer Theorem
λk = minx⊥v1,...,vk−1
xTLx
xTx
![Page 40: Laplacian Matrices of Graphs - cs.yale. · PDF fileLaplacian Matrices of Graphs: Spectral and Electrical Theory Daniel A. Spielman ... Spectral graph theory A very fast survey Trailer](https://reader031.vdocuments.net/reader031/viewer/2022022500/5aa4690d7f8b9a517d8be0b8/html5/thumbnails/40.jpg)
The first eigenvalue
Setting for all x(a) = 1 a
We find and λ1 = 0 v1 = 1
λ1 = minx �=0
xTLx
xTx
= minx �=0
�(a,b)∈E(x(a)− x(b))2
�x�2
![Page 41: Laplacian Matrices of Graphs - cs.yale. · PDF fileLaplacian Matrices of Graphs: Spectral and Electrical Theory Daniel A. Spielman ... Spectral graph theory A very fast survey Trailer](https://reader031.vdocuments.net/reader031/viewer/2022022500/5aa4690d7f8b9a517d8be0b8/html5/thumbnails/41.jpg)
The second eigenvalue
if and only if G is connected λ2 > 0
Proof: if G is not connected, are two functions with Rayleigh quotient zero
1 1
0
0
0 0
1 1
![Page 42: Laplacian Matrices of Graphs - cs.yale. · PDF fileLaplacian Matrices of Graphs: Spectral and Electrical Theory Daniel A. Spielman ... Spectral graph theory A very fast survey Trailer](https://reader031.vdocuments.net/reader031/viewer/2022022500/5aa4690d7f8b9a517d8be0b8/html5/thumbnails/42.jpg)
The second eigenvalue
if and only if G is connected λ2 > 0
Proof: if G is not connected, are two functions with Rayleigh quotient zero
0 0
1
1
1 1
0 0
![Page 43: Laplacian Matrices of Graphs - cs.yale. · PDF fileLaplacian Matrices of Graphs: Spectral and Electrical Theory Daniel A. Spielman ... Spectral graph theory A very fast survey Trailer](https://reader031.vdocuments.net/reader031/viewer/2022022500/5aa4690d7f8b9a517d8be0b8/html5/thumbnails/43.jpg)
The second eigenvalue
if and only if G is connected λ2 > 0
Proof: if G is connected,
-
+
x ⊥ 1�
a x(a) = 0means
So must be an edge (a,b) for which x(a) < x(b) and so (x(a)− x(b))2 > 0
![Page 44: Laplacian Matrices of Graphs - cs.yale. · PDF fileLaplacian Matrices of Graphs: Spectral and Electrical Theory Daniel A. Spielman ... Spectral graph theory A very fast survey Trailer](https://reader031.vdocuments.net/reader031/viewer/2022022500/5aa4690d7f8b9a517d8be0b8/html5/thumbnails/44.jpg)
The second eigenvalue
if and only if G is connected λ2 > 0
Proof: if G is connected,
- - +
+
x ⊥ 1�
a x(a) = 0means
So must be an edge (a,b) for which x(a) < x(b) and so (x(a)− x(b))2 > 0
![Page 45: Laplacian Matrices of Graphs - cs.yale. · PDF fileLaplacian Matrices of Graphs: Spectral and Electrical Theory Daniel A. Spielman ... Spectral graph theory A very fast survey Trailer](https://reader031.vdocuments.net/reader031/viewer/2022022500/5aa4690d7f8b9a517d8be0b8/html5/thumbnails/45.jpg)
The second eigenvalue
if and only if G is connected λ2 > 0
Fiedler (‘73) called “the algebraic connectivity of a graph” The further from 0, the more connected.
![Page 46: Laplacian Matrices of Graphs - cs.yale. · PDF fileLaplacian Matrices of Graphs: Spectral and Electrical Theory Daniel A. Spielman ... Spectral graph theory A very fast survey Trailer](https://reader031.vdocuments.net/reader031/viewer/2022022500/5aa4690d7f8b9a517d8be0b8/html5/thumbnails/46.jpg)
Cheeger’s Inequality [Cheeger ‘70]
[Alon-Milman ‘85, Jerrum-Sinclair ‘89, Diaconis-Stroock ‘91]
1. is big if and only if G does not have good clusters.
2. If is small, can use to find a good cluster. v2
![Page 47: Laplacian Matrices of Graphs - cs.yale. · PDF fileLaplacian Matrices of Graphs: Spectral and Electrical Theory Daniel A. Spielman ... Spectral graph theory A very fast survey Trailer](https://reader031.vdocuments.net/reader031/viewer/2022022500/5aa4690d7f8b9a517d8be0b8/html5/thumbnails/47.jpg)
Cheeger’s Inequality [Cheeger ‘70]
[Alon-Milman ‘85, Jerrum-Sinclair ‘89, Diaconis-Stroock ‘91]
1. is big if and only if G does not have good clusters.
λ2/2 ≤ min|S|≤n/2
Φ(S) ≤�
2dλ2
When every vertex has d edges,
Φ(S) =|boundary(S)|
|S|
![Page 48: Laplacian Matrices of Graphs - cs.yale. · PDF fileLaplacian Matrices of Graphs: Spectral and Electrical Theory Daniel A. Spielman ... Spectral graph theory A very fast survey Trailer](https://reader031.vdocuments.net/reader031/viewer/2022022500/5aa4690d7f8b9a517d8be0b8/html5/thumbnails/48.jpg)
Cheeger’s Inequality [Cheeger ‘70]
[Alon-Milman ‘85, Jerrum-Sinclair ‘89, Diaconis-Stroock ‘91]
1. is big if and only if G does not have good clusters.
2. If is small, can use to find a good cluster. v2
In a moment…
![Page 49: Laplacian Matrices of Graphs - cs.yale. · PDF fileLaplacian Matrices of Graphs: Spectral and Electrical Theory Daniel A. Spielman ... Spectral graph theory A very fast survey Trailer](https://reader031.vdocuments.net/reader031/viewer/2022022500/5aa4690d7f8b9a517d8be0b8/html5/thumbnails/49.jpg)
Spectral Graph Drawing [Hall ‘70]
31 2
4
56 7
8 9
Arbitrary Drawing
![Page 50: Laplacian Matrices of Graphs - cs.yale. · PDF fileLaplacian Matrices of Graphs: Spectral and Electrical Theory Daniel A. Spielman ... Spectral graph theory A very fast survey Trailer](https://reader031.vdocuments.net/reader031/viewer/2022022500/5aa4690d7f8b9a517d8be0b8/html5/thumbnails/50.jpg)
Spectral Graph Drawing Plot vertex at draw edges as straight lines
[Hall ‘70]
(v2(a), v3(a))a
31 2
4
56 7
8 9
12
4
5
6
9
3
8
7Arbitrary Drawing
Spectral Drawing
![Page 51: Laplacian Matrices of Graphs - cs.yale. · PDF fileLaplacian Matrices of Graphs: Spectral and Electrical Theory Daniel A. Spielman ... Spectral graph theory A very fast survey Trailer](https://reader031.vdocuments.net/reader031/viewer/2022022500/5aa4690d7f8b9a517d8be0b8/html5/thumbnails/51.jpg)
A Graph
![Page 52: Laplacian Matrices of Graphs - cs.yale. · PDF fileLaplacian Matrices of Graphs: Spectral and Electrical Theory Daniel A. Spielman ... Spectral graph theory A very fast survey Trailer](https://reader031.vdocuments.net/reader031/viewer/2022022500/5aa4690d7f8b9a517d8be0b8/html5/thumbnails/52.jpg)
Drawing of the graph using v2, v3
Plot vertex at a (v2(a), v3(a))
![Page 53: Laplacian Matrices of Graphs - cs.yale. · PDF fileLaplacian Matrices of Graphs: Spectral and Electrical Theory Daniel A. Spielman ... Spectral graph theory A very fast survey Trailer](https://reader031.vdocuments.net/reader031/viewer/2022022500/5aa4690d7f8b9a517d8be0b8/html5/thumbnails/53.jpg)
![Page 54: Laplacian Matrices of Graphs - cs.yale. · PDF fileLaplacian Matrices of Graphs: Spectral and Electrical Theory Daniel A. Spielman ... Spectral graph theory A very fast survey Trailer](https://reader031.vdocuments.net/reader031/viewer/2022022500/5aa4690d7f8b9a517d8be0b8/html5/thumbnails/54.jpg)
The Airfoil Graph, original coordinates
![Page 55: Laplacian Matrices of Graphs - cs.yale. · PDF fileLaplacian Matrices of Graphs: Spectral and Electrical Theory Daniel A. Spielman ... Spectral graph theory A very fast survey Trailer](https://reader031.vdocuments.net/reader031/viewer/2022022500/5aa4690d7f8b9a517d8be0b8/html5/thumbnails/55.jpg)
The Airfoil Graph, spectral coordinates
![Page 56: Laplacian Matrices of Graphs - cs.yale. · PDF fileLaplacian Matrices of Graphs: Spectral and Electrical Theory Daniel A. Spielman ... Spectral graph theory A very fast survey Trailer](https://reader031.vdocuments.net/reader031/viewer/2022022500/5aa4690d7f8b9a517d8be0b8/html5/thumbnails/56.jpg)
The Airfoil Graph, spectral coordinates
![Page 57: Laplacian Matrices of Graphs - cs.yale. · PDF fileLaplacian Matrices of Graphs: Spectral and Electrical Theory Daniel A. Spielman ... Spectral graph theory A very fast survey Trailer](https://reader031.vdocuments.net/reader031/viewer/2022022500/5aa4690d7f8b9a517d8be0b8/html5/thumbnails/57.jpg)
Spectral drawing of Streets in Rome
![Page 58: Laplacian Matrices of Graphs - cs.yale. · PDF fileLaplacian Matrices of Graphs: Spectral and Electrical Theory Daniel A. Spielman ... Spectral graph theory A very fast survey Trailer](https://reader031.vdocuments.net/reader031/viewer/2022022500/5aa4690d7f8b9a517d8be0b8/html5/thumbnails/58.jpg)
Spectral drawing of Erdos graph: edge between co-authors of papers
![Page 59: Laplacian Matrices of Graphs - cs.yale. · PDF fileLaplacian Matrices of Graphs: Spectral and Electrical Theory Daniel A. Spielman ... Spectral graph theory A very fast survey Trailer](https://reader031.vdocuments.net/reader031/viewer/2022022500/5aa4690d7f8b9a517d8be0b8/html5/thumbnails/59.jpg)
Dodecahedron
Best embedded by first three eigenvectors
![Page 60: Laplacian Matrices of Graphs - cs.yale. · PDF fileLaplacian Matrices of Graphs: Spectral and Electrical Theory Daniel A. Spielman ... Spectral graph theory A very fast survey Trailer](https://reader031.vdocuments.net/reader031/viewer/2022022500/5aa4690d7f8b9a517d8be0b8/html5/thumbnails/60.jpg)
Condition for eigenvector
Spectral graph drawing: Tutte justification
Gives for all
λ small says near average of neighbors
�x(a) =1
da − λ
�
(a,b)∈E
�x(b)
�x(a)
a
![Page 61: Laplacian Matrices of Graphs - cs.yale. · PDF fileLaplacian Matrices of Graphs: Spectral and Electrical Theory Daniel A. Spielman ... Spectral graph theory A very fast survey Trailer](https://reader031.vdocuments.net/reader031/viewer/2022022500/5aa4690d7f8b9a517d8be0b8/html5/thumbnails/61.jpg)
Spectral graph drawing: Tutte justification
For planar graphs:
λ2 ≤ 8d/n [S-Teng ‘96]
λ3 ≤ O(d/n) [Kelner-Lee-Price-Teng ‘09]
Condition for eigenvector
Gives for all
λ small says near average of neighbors
�x(a) =1
da − λ
�
(a,b)∈E
�x(b)
�x(a)
a
![Page 62: Laplacian Matrices of Graphs - cs.yale. · PDF fileLaplacian Matrices of Graphs: Spectral and Electrical Theory Daniel A. Spielman ... Spectral graph theory A very fast survey Trailer](https://reader031.vdocuments.net/reader031/viewer/2022022500/5aa4690d7f8b9a517d8be0b8/html5/thumbnails/62.jpg)
Small eigenvalues are not enough
Plot vertex at a (v3(a), v4(a))
![Page 63: Laplacian Matrices of Graphs - cs.yale. · PDF fileLaplacian Matrices of Graphs: Spectral and Electrical Theory Daniel A. Spielman ... Spectral graph theory A very fast survey Trailer](https://reader031.vdocuments.net/reader031/viewer/2022022500/5aa4690d7f8b9a517d8be0b8/html5/thumbnails/63.jpg)
Spectral Graph Partitioning
for some
[Donath-Hoffman ‘72, Barnes ‘82, Hagen-Kahng ‘92]
S = {a : v2(a) ≤ t} t
![Page 64: Laplacian Matrices of Graphs - cs.yale. · PDF fileLaplacian Matrices of Graphs: Spectral and Electrical Theory Daniel A. Spielman ... Spectral graph theory A very fast survey Trailer](https://reader031.vdocuments.net/reader031/viewer/2022022500/5aa4690d7f8b9a517d8be0b8/html5/thumbnails/64.jpg)
Spectral Graph Partitioning [Donath-Hoffman ‘72, Barnes ‘82, Hagen-Kahng ‘92]
Cheeger’s Inequality says there is a so that
Φ(S) ≤�
2dλ2
for some S = {a : v2(a) ≤ t} t
t
![Page 65: Laplacian Matrices of Graphs - cs.yale. · PDF fileLaplacian Matrices of Graphs: Spectral and Electrical Theory Daniel A. Spielman ... Spectral graph theory A very fast survey Trailer](https://reader031.vdocuments.net/reader031/viewer/2022022500/5aa4690d7f8b9a517d8be0b8/html5/thumbnails/65.jpg)
Major topics in spectral graph theory Graph Isomorphism: determining if two graphs are the same
Independent sets: large sets of vertices containing no edges
Graph Coloring: so that edges connect different colors
![Page 66: Laplacian Matrices of Graphs - cs.yale. · PDF fileLaplacian Matrices of Graphs: Spectral and Electrical Theory Daniel A. Spielman ... Spectral graph theory A very fast survey Trailer](https://reader031.vdocuments.net/reader031/viewer/2022022500/5aa4690d7f8b9a517d8be0b8/html5/thumbnails/66.jpg)
Major topics in spectral graph theory Graph Isomorphism Independent sets Graph Coloring Behavior under graph transformations Random Walks and Diffusion PageRank and Hits Colin de Verdière invariant Special Graphs from groups from meshes Machine learning Image processing
![Page 67: Laplacian Matrices of Graphs - cs.yale. · PDF fileLaplacian Matrices of Graphs: Spectral and Electrical Theory Daniel A. Spielman ... Spectral graph theory A very fast survey Trailer](https://reader031.vdocuments.net/reader031/viewer/2022022500/5aa4690d7f8b9a517d8be0b8/html5/thumbnails/67.jpg)
Solving linear equations in Laplacians
For energy minimization and computation of eigenvectors and eigenvalues
Can do it in time nearly-linear in the number of edges in the graph!
A powerful computational primitive.
![Page 68: Laplacian Matrices of Graphs - cs.yale. · PDF fileLaplacian Matrices of Graphs: Spectral and Electrical Theory Daniel A. Spielman ... Spectral graph theory A very fast survey Trailer](https://reader031.vdocuments.net/reader031/viewer/2022022500/5aa4690d7f8b9a517d8be0b8/html5/thumbnails/68.jpg)
Maximum flow problem
Send as much stuff as possible from s to t. At most one unit can go through each edge.
s t
![Page 69: Laplacian Matrices of Graphs - cs.yale. · PDF fileLaplacian Matrices of Graphs: Spectral and Electrical Theory Daniel A. Spielman ... Spectral graph theory A very fast survey Trailer](https://reader031.vdocuments.net/reader031/viewer/2022022500/5aa4690d7f8b9a517d8be0b8/html5/thumbnails/69.jpg)
Maximum flow problem
s t
1 1
1 1 1
1 1
Send as much stuff as possible from s to t. At most one unit can go through each edge.
![Page 70: Laplacian Matrices of Graphs - cs.yale. · PDF fileLaplacian Matrices of Graphs: Spectral and Electrical Theory Daniel A. Spielman ... Spectral graph theory A very fast survey Trailer](https://reader031.vdocuments.net/reader031/viewer/2022022500/5aa4690d7f8b9a517d8be0b8/html5/thumbnails/70.jpg)
Maximum flow problem
s t
1 1
Standard approach: incrementally add flow paths
Send as much stuff as possible from s to t. At most one unit can go through each edge.
![Page 71: Laplacian Matrices of Graphs - cs.yale. · PDF fileLaplacian Matrices of Graphs: Spectral and Electrical Theory Daniel A. Spielman ... Spectral graph theory A very fast survey Trailer](https://reader031.vdocuments.net/reader031/viewer/2022022500/5aa4690d7f8b9a517d8be0b8/html5/thumbnails/71.jpg)
Maximum flow problem
s t
1 1
1 1
Standard approach: incrementally add flow paths
Send as much stuff as possible from s to t. At most one unit can go through each edge.
![Page 72: Laplacian Matrices of Graphs - cs.yale. · PDF fileLaplacian Matrices of Graphs: Spectral and Electrical Theory Daniel A. Spielman ... Spectral graph theory A very fast survey Trailer](https://reader031.vdocuments.net/reader031/viewer/2022022500/5aa4690d7f8b9a517d8be0b8/html5/thumbnails/72.jpg)
Maximum flow problem
s t
1 1 1
1
1
1
1
Standard approach: incrementally add flow paths
Send as much stuff as possible from s to t. At most one unit can go through each edge.
![Page 73: Laplacian Matrices of Graphs - cs.yale. · PDF fileLaplacian Matrices of Graphs: Spectral and Electrical Theory Daniel A. Spielman ... Spectral graph theory A very fast survey Trailer](https://reader031.vdocuments.net/reader031/viewer/2022022500/5aa4690d7f8b9a517d8be0b8/html5/thumbnails/73.jpg)
Maximum flow problem
s t 1
1
Standard approach: incrementally add flow paths Issue: sometimes requires backtracking
1
Send as much stuff as possible from s to t. At most one unit can go through each edge.
![Page 74: Laplacian Matrices of Graphs - cs.yale. · PDF fileLaplacian Matrices of Graphs: Spectral and Electrical Theory Daniel A. Spielman ... Spectral graph theory A very fast survey Trailer](https://reader031.vdocuments.net/reader031/viewer/2022022500/5aa4690d7f8b9a517d8be0b8/html5/thumbnails/74.jpg)
Maximum flow problem, electrical approach
1. Try the electrical flow.
s t
[Christiano-Kelner-Madry-S-Teng ‘11]
0V
2.18 V
0.82 0.82 0.55
1.09
1.09
1.09
1.09
0.27 0.27
![Page 75: Laplacian Matrices of Graphs - cs.yale. · PDF fileLaplacian Matrices of Graphs: Spectral and Electrical Theory Daniel A. Spielman ... Spectral graph theory A very fast survey Trailer](https://reader031.vdocuments.net/reader031/viewer/2022022500/5aa4690d7f8b9a517d8be0b8/html5/thumbnails/75.jpg)
Maximum flow problem, electrical approach
1. Try the electrical flow. 2. Increase resistance when too much flow
s t
[Christiano-Kelner-Madry-S-Teng ‘11]
0V
2.18 V
0.82 0.82 0.55
1.09
1.09
1.09
1.09
0.27 0.27
![Page 76: Laplacian Matrices of Graphs - cs.yale. · PDF fileLaplacian Matrices of Graphs: Spectral and Electrical Theory Daniel A. Spielman ... Spectral graph theory A very fast survey Trailer](https://reader031.vdocuments.net/reader031/viewer/2022022500/5aa4690d7f8b9a517d8be0b8/html5/thumbnails/76.jpg)
Maximum flow problem, electrical approach
1. Try the electrical flow. 2. Increase resistance when too much flow
s t
[Christiano-Kelner-Madry-S-Teng ‘11]
0V
2.10 V
0.90 0.90 0.60
1.05
1.05
1.05
1.05
0.30 0.30
![Page 77: Laplacian Matrices of Graphs - cs.yale. · PDF fileLaplacian Matrices of Graphs: Spectral and Electrical Theory Daniel A. Spielman ... Spectral graph theory A very fast survey Trailer](https://reader031.vdocuments.net/reader031/viewer/2022022500/5aa4690d7f8b9a517d8be0b8/html5/thumbnails/77.jpg)
Solving linear equations in Laplacians
For energy minimization and computation of eigenvectors and eigenvalues
Can do it in time nearly-linear in the number of edges in the graph!
Key ideas: how to approximate a graph by a tree or by a very sparse graph random matrix theory numerical linear algebra
![Page 78: Laplacian Matrices of Graphs - cs.yale. · PDF fileLaplacian Matrices of Graphs: Spectral and Electrical Theory Daniel A. Spielman ... Spectral graph theory A very fast survey Trailer](https://reader031.vdocuments.net/reader031/viewer/2022022500/5aa4690d7f8b9a517d8be0b8/html5/thumbnails/78.jpg)
Approximating Graphs
for all
1
1 + �≤ xTLHx
xTLGx≤ 1 + �
x
A graph H is an ∊ -approximation of G if
![Page 79: Laplacian Matrices of Graphs - cs.yale. · PDF fileLaplacian Matrices of Graphs: Spectral and Electrical Theory Daniel A. Spielman ... Spectral graph theory A very fast survey Trailer](https://reader031.vdocuments.net/reader031/viewer/2022022500/5aa4690d7f8b9a517d8be0b8/html5/thumbnails/79.jpg)
Approximating Graphs
for all
1
1 + �≤ xTLHx
xTLGx≤ 1 + �
x
To solve linear equations quickly, approximate G by a simpler graph H
A graph H is an ∊ -approximation of G if
![Page 80: Laplacian Matrices of Graphs - cs.yale. · PDF fileLaplacian Matrices of Graphs: Spectral and Electrical Theory Daniel A. Spielman ... Spectral graph theory A very fast survey Trailer](https://reader031.vdocuments.net/reader031/viewer/2022022500/5aa4690d7f8b9a517d8be0b8/html5/thumbnails/80.jpg)
Approximating Graphs
for all
1
1 + �≤ xTLHx
xTLGx≤ 1 + �
x
A very strong notion of approximation Preserves all electrical and spectral properties
A graph H is an ∊ -approximation of G if
![Page 81: Laplacian Matrices of Graphs - cs.yale. · PDF fileLaplacian Matrices of Graphs: Spectral and Electrical Theory Daniel A. Spielman ... Spectral graph theory A very fast survey Trailer](https://reader031.vdocuments.net/reader031/viewer/2022022500/5aa4690d7f8b9a517d8be0b8/html5/thumbnails/81.jpg)
Approximating Graphs
for all 1
1 + �≤ xTLHx
xTLGx≤ 1 + �x
Theorem [Batson-S-Srivastava ‘09] Every graph G has an ∊ -approximation H with edges |V | (2 + �)2/�2
A graph H is an ∊ -approximation of G if
![Page 82: Laplacian Matrices of Graphs - cs.yale. · PDF fileLaplacian Matrices of Graphs: Spectral and Electrical Theory Daniel A. Spielman ... Spectral graph theory A very fast survey Trailer](https://reader031.vdocuments.net/reader031/viewer/2022022500/5aa4690d7f8b9a517d8be0b8/html5/thumbnails/82.jpg)
Approximating Graphs
for all 1
1 + �≤ xTLHx
xTLGx≤ 1 + �x
Theorem [Batson-S-Srivastava ‘09] Every graph G has an ∊ -approximation H with edges
A powerful technique in linear algebra many applications
|V | (2 + �)2/�2
A graph H is an ∊ -approximation of G if
![Page 83: Laplacian Matrices of Graphs - cs.yale. · PDF fileLaplacian Matrices of Graphs: Spectral and Electrical Theory Daniel A. Spielman ... Spectral graph theory A very fast survey Trailer](https://reader031.vdocuments.net/reader031/viewer/2022022500/5aa4690d7f8b9a517d8be0b8/html5/thumbnails/83.jpg)
To learn more
Lectures 2 and 3: More precision More notation Similar sophistication
![Page 84: Laplacian Matrices of Graphs - cs.yale. · PDF fileLaplacian Matrices of Graphs: Spectral and Electrical Theory Daniel A. Spielman ... Spectral graph theory A very fast survey Trailer](https://reader031.vdocuments.net/reader031/viewer/2022022500/5aa4690d7f8b9a517d8be0b8/html5/thumbnails/84.jpg)
To learn more
See my lecture notes from “Spectral Graph Theory” and “Graphs and Networks”