adam marcus (crisply, yale) daniel spielman (yale) nikhil...
TRANSCRIPT
![Page 1: Adam Marcus (Crisply, Yale) Daniel Spielman (Yale) Nikhil …cs.yale.edu/homes/spielman/TALKS/erdos3raman.pdf · 2014-05-27 · Nikhil Srivastava (MSR India) Expander Graphs Sparse,](https://reader033.vdocuments.net/reader033/viewer/2022043012/5fa9fe9adf2c9913832b65eb/html5/thumbnails/1.jpg)
Ramanujan Graphs of Every Degree
Adam Marcus (Crisply, Yale) Daniel Spielman (Yale)
Nikhil Srivastava (MSR India)
![Page 2: Adam Marcus (Crisply, Yale) Daniel Spielman (Yale) Nikhil …cs.yale.edu/homes/spielman/TALKS/erdos3raman.pdf · 2014-05-27 · Nikhil Srivastava (MSR India) Expander Graphs Sparse,](https://reader033.vdocuments.net/reader033/viewer/2022043012/5fa9fe9adf2c9913832b65eb/html5/thumbnails/2.jpg)
Expander Graphs
Sparse, regular well-connected graphs with many properties of random graphs.
Random walks mix quickly.
Every set of vertices has many neighbors.
Pseudo-random generators.
Error-correcting codes.
Sparse approximations of complete graphs.
![Page 3: Adam Marcus (Crisply, Yale) Daniel Spielman (Yale) Nikhil …cs.yale.edu/homes/spielman/TALKS/erdos3raman.pdf · 2014-05-27 · Nikhil Srivastava (MSR India) Expander Graphs Sparse,](https://reader033.vdocuments.net/reader033/viewer/2022043012/5fa9fe9adf2c9913832b65eb/html5/thumbnails/3.jpg)
Spectral Expanders Let G be a graph and A be its adjacency matrix
a
c
d
e b
0 1 0 0 1 1 0 1 0 1 0 1 0 1 0 0 0 1 0 1 1 1 0 1 0
Eigenvalues �1 � �2 � · · · � �n
If d-regular (every vertex has d edges), �1 = d
“trivial”
![Page 4: Adam Marcus (Crisply, Yale) Daniel Spielman (Yale) Nikhil …cs.yale.edu/homes/spielman/TALKS/erdos3raman.pdf · 2014-05-27 · Nikhil Srivastava (MSR India) Expander Graphs Sparse,](https://reader033.vdocuments.net/reader033/viewer/2022043012/5fa9fe9adf2c9913832b65eb/html5/thumbnails/4.jpg)
Spectral Expanders If bipartite (all edges between two parts/colors)
eigenvalues are symmetric about 0
“trivial”
a
c d
f
b
e
If d-regular and bipartite, �n = �d
0 0 0 1 0 1 0 0 0 1 1 0 0 0 0 0 1 1 1 1 0 0 0 0 0 1 1 0 0 0 1 0 1 0 0 0
![Page 5: Adam Marcus (Crisply, Yale) Daniel Spielman (Yale) Nikhil …cs.yale.edu/homes/spielman/TALKS/erdos3raman.pdf · 2014-05-27 · Nikhil Srivastava (MSR India) Expander Graphs Sparse,](https://reader033.vdocuments.net/reader033/viewer/2022043012/5fa9fe9adf2c9913832b65eb/html5/thumbnails/5.jpg)
Spectral Expanders
0 [ ] -‐d d
G is a good spectral expander if all non-trivial eigenvalues are small
![Page 6: Adam Marcus (Crisply, Yale) Daniel Spielman (Yale) Nikhil …cs.yale.edu/homes/spielman/TALKS/erdos3raman.pdf · 2014-05-27 · Nikhil Srivastava (MSR India) Expander Graphs Sparse,](https://reader033.vdocuments.net/reader033/viewer/2022043012/5fa9fe9adf2c9913832b65eb/html5/thumbnails/6.jpg)
Bipartite Complete Graph Adjacency matrix has rank 2, so all non-trivial eigenvalues are 0
a
c d
f
b
e
0 0 0 1 1 1 0 0 0 1 1 1 0 0 0 1 1 1 1 1 1 0 0 0 1 1 1 0 0 0 1 1 1 0 0 0
![Page 7: Adam Marcus (Crisply, Yale) Daniel Spielman (Yale) Nikhil …cs.yale.edu/homes/spielman/TALKS/erdos3raman.pdf · 2014-05-27 · Nikhil Srivastava (MSR India) Expander Graphs Sparse,](https://reader033.vdocuments.net/reader033/viewer/2022043012/5fa9fe9adf2c9913832b65eb/html5/thumbnails/7.jpg)
Spectral Expanders
0 [ ] -‐d d
Challenge: construct infinite families of fixed degree
G is a good spectral expander if all non-trivial eigenvalues are small
![Page 8: Adam Marcus (Crisply, Yale) Daniel Spielman (Yale) Nikhil …cs.yale.edu/homes/spielman/TALKS/erdos3raman.pdf · 2014-05-27 · Nikhil Srivastava (MSR India) Expander Graphs Sparse,](https://reader033.vdocuments.net/reader033/viewer/2022043012/5fa9fe9adf2c9913832b65eb/html5/thumbnails/8.jpg)
Spectral Expanders
G is a good spectral expander if all non-trivial eigenvalues are small
0 [ ] -‐d d
Challenge: construct infinite families of fixed degree
Alon-Boppana ‘86: Cannot beat 2pd� 1
) ( 2pd� 1�2
pd� 1
![Page 9: Adam Marcus (Crisply, Yale) Daniel Spielman (Yale) Nikhil …cs.yale.edu/homes/spielman/TALKS/erdos3raman.pdf · 2014-05-27 · Nikhil Srivastava (MSR India) Expander Graphs Sparse,](https://reader033.vdocuments.net/reader033/viewer/2022043012/5fa9fe9adf2c9913832b65eb/html5/thumbnails/9.jpg)
G is a Ramanujan Graph if absolute value of non-trivial eigs
0 [ ] -‐d d
) (
Ramanujan Graphs: 2pd� 1
2pd� 1
2pd� 1�2
pd� 1
![Page 10: Adam Marcus (Crisply, Yale) Daniel Spielman (Yale) Nikhil …cs.yale.edu/homes/spielman/TALKS/erdos3raman.pdf · 2014-05-27 · Nikhil Srivastava (MSR India) Expander Graphs Sparse,](https://reader033.vdocuments.net/reader033/viewer/2022043012/5fa9fe9adf2c9913832b65eb/html5/thumbnails/10.jpg)
G is a Ramanujan Graph if absolute value of non-trivial eigs
0 [ ] -‐d d
) (
Ramanujan Graphs: 2pd� 1
2pd� 1
Margulis, Lubotzky-‐Phillips-‐Sarnak’88: Infinite sequences of Ramanujan graphs exist for
2pd� 1�2
pd� 1
d = prime + 1
![Page 11: Adam Marcus (Crisply, Yale) Daniel Spielman (Yale) Nikhil …cs.yale.edu/homes/spielman/TALKS/erdos3raman.pdf · 2014-05-27 · Nikhil Srivastava (MSR India) Expander Graphs Sparse,](https://reader033.vdocuments.net/reader033/viewer/2022043012/5fa9fe9adf2c9913832b65eb/html5/thumbnails/11.jpg)
G is a Ramanujan Graph if absolute value of non-trivial eigs
0 [ ] -‐d d
) (
Ramanujan Graphs: 2pd� 1
2pd� 1
Friedman’08: A random d-‐regular graph is almost Ramanujan : 2
pd� 1 + ✏
2pd� 1�2
pd� 1
![Page 12: Adam Marcus (Crisply, Yale) Daniel Spielman (Yale) Nikhil …cs.yale.edu/homes/spielman/TALKS/erdos3raman.pdf · 2014-05-27 · Nikhil Srivastava (MSR India) Expander Graphs Sparse,](https://reader033.vdocuments.net/reader033/viewer/2022043012/5fa9fe9adf2c9913832b65eb/html5/thumbnails/12.jpg)
Ramanujan Graphs of Every Degree
Theorem: there are infinite families of bipartite Ramanujan graphs of every degree.
![Page 13: Adam Marcus (Crisply, Yale) Daniel Spielman (Yale) Nikhil …cs.yale.edu/homes/spielman/TALKS/erdos3raman.pdf · 2014-05-27 · Nikhil Srivastava (MSR India) Expander Graphs Sparse,](https://reader033.vdocuments.net/reader033/viewer/2022043012/5fa9fe9adf2c9913832b65eb/html5/thumbnails/13.jpg)
Theorem: there are infinite families of bipartite Ramanujan graphs of every degree. And, are infinite families of (c,d)-biregular Ramanujan graphs, having non-trivial eigs bounded by p
d� 1 +pc� 1
Ramanujan Graphs of Every Degree
![Page 14: Adam Marcus (Crisply, Yale) Daniel Spielman (Yale) Nikhil …cs.yale.edu/homes/spielman/TALKS/erdos3raman.pdf · 2014-05-27 · Nikhil Srivastava (MSR India) Expander Graphs Sparse,](https://reader033.vdocuments.net/reader033/viewer/2022043012/5fa9fe9adf2c9913832b65eb/html5/thumbnails/14.jpg)
Find an operation that doubles the size of a graph without creating large eigenvalues.
0 [ ] -‐d d
Bilu-Linial ‘06 Approach
2pd� 1�2
pd� 1
( )
![Page 15: Adam Marcus (Crisply, Yale) Daniel Spielman (Yale) Nikhil …cs.yale.edu/homes/spielman/TALKS/erdos3raman.pdf · 2014-05-27 · Nikhil Srivastava (MSR India) Expander Graphs Sparse,](https://reader033.vdocuments.net/reader033/viewer/2022043012/5fa9fe9adf2c9913832b65eb/html5/thumbnails/15.jpg)
0 [ ] -‐d d
Bilu-Linial ‘06 Approach
2pd� 1�2
pd� 1
Find an operation that doubles the size of a graph without creating large eigenvalues.
( )
![Page 16: Adam Marcus (Crisply, Yale) Daniel Spielman (Yale) Nikhil …cs.yale.edu/homes/spielman/TALKS/erdos3raman.pdf · 2014-05-27 · Nikhil Srivastava (MSR India) Expander Graphs Sparse,](https://reader033.vdocuments.net/reader033/viewer/2022043012/5fa9fe9adf2c9913832b65eb/html5/thumbnails/16.jpg)
2-lifts of graphs
a
c
d
e b
![Page 17: Adam Marcus (Crisply, Yale) Daniel Spielman (Yale) Nikhil …cs.yale.edu/homes/spielman/TALKS/erdos3raman.pdf · 2014-05-27 · Nikhil Srivastava (MSR India) Expander Graphs Sparse,](https://reader033.vdocuments.net/reader033/viewer/2022043012/5fa9fe9adf2c9913832b65eb/html5/thumbnails/17.jpg)
2-lifts of graphs
a
c
d
e b
a
c
d
e b
duplicate every vertex
![Page 18: Adam Marcus (Crisply, Yale) Daniel Spielman (Yale) Nikhil …cs.yale.edu/homes/spielman/TALKS/erdos3raman.pdf · 2014-05-27 · Nikhil Srivastava (MSR India) Expander Graphs Sparse,](https://reader033.vdocuments.net/reader033/viewer/2022043012/5fa9fe9adf2c9913832b65eb/html5/thumbnails/18.jpg)
2-lifts of graphs
duplicate every vertex
a0
c0
d0
e0
b0
a1
c1
d1
e1
b1
![Page 19: Adam Marcus (Crisply, Yale) Daniel Spielman (Yale) Nikhil …cs.yale.edu/homes/spielman/TALKS/erdos3raman.pdf · 2014-05-27 · Nikhil Srivastava (MSR India) Expander Graphs Sparse,](https://reader033.vdocuments.net/reader033/viewer/2022043012/5fa9fe9adf2c9913832b65eb/html5/thumbnails/19.jpg)
2-lifts of graphs
a0
c0
d0
e0
b0
a1
d1
e1
b1
c1
for every pair of edges: leave on either side (parallel), or make both cross
![Page 20: Adam Marcus (Crisply, Yale) Daniel Spielman (Yale) Nikhil …cs.yale.edu/homes/spielman/TALKS/erdos3raman.pdf · 2014-05-27 · Nikhil Srivastava (MSR India) Expander Graphs Sparse,](https://reader033.vdocuments.net/reader033/viewer/2022043012/5fa9fe9adf2c9913832b65eb/html5/thumbnails/20.jpg)
2-lifts of graphs
for every pair of edges: leave on either side (parallel), or make both cross
a0
c0
d0
e0
b0
a1
d1
e1
b1
c1
![Page 21: Adam Marcus (Crisply, Yale) Daniel Spielman (Yale) Nikhil …cs.yale.edu/homes/spielman/TALKS/erdos3raman.pdf · 2014-05-27 · Nikhil Srivastava (MSR India) Expander Graphs Sparse,](https://reader033.vdocuments.net/reader033/viewer/2022043012/5fa9fe9adf2c9913832b65eb/html5/thumbnails/21.jpg)
2-lifts of graphs
0 1 0 0 1 1 0 1 0 1 0 1 0 1 0 0 0 1 0 1 1 1 0 1 0
![Page 22: Adam Marcus (Crisply, Yale) Daniel Spielman (Yale) Nikhil …cs.yale.edu/homes/spielman/TALKS/erdos3raman.pdf · 2014-05-27 · Nikhil Srivastava (MSR India) Expander Graphs Sparse,](https://reader033.vdocuments.net/reader033/viewer/2022043012/5fa9fe9adf2c9913832b65eb/html5/thumbnails/22.jpg)
2-lifts of graphs
0 1 0 0 1 0 0 0 0 0 1 0 1 0 1 0 0 0 0 0 0 1 0 1 0 0 0 0 0 0 0 0 1 0 1 0 0 0 0 0 1 1 0 1 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 1 0 0 0 0 0 1 0 1 0 1 0 0 0 0 0 0 1 0 1 0 0 0 0 0 0 0 0 1 0 1 0 0 0 0 0 1 1 0 1 0
![Page 23: Adam Marcus (Crisply, Yale) Daniel Spielman (Yale) Nikhil …cs.yale.edu/homes/spielman/TALKS/erdos3raman.pdf · 2014-05-27 · Nikhil Srivastava (MSR India) Expander Graphs Sparse,](https://reader033.vdocuments.net/reader033/viewer/2022043012/5fa9fe9adf2c9913832b65eb/html5/thumbnails/23.jpg)
2-lifts of graphs
0 0 0 0 1 0 1 0 0 0 0 0 1 0 1 1 0 0 0 0 0 1 0 0 0 0 0 0 1 0 0 0 0 0 1 0 0 1 0 0 1 1 0 1 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 1 1 0 0 0 0 0 0 1 0 1 0 0 0 1 0 0 1 0 0 0 0 0 1 0 0 0 0 0 0 1 0 0 0 0 0 1 1 0 1 0
![Page 24: Adam Marcus (Crisply, Yale) Daniel Spielman (Yale) Nikhil …cs.yale.edu/homes/spielman/TALKS/erdos3raman.pdf · 2014-05-27 · Nikhil Srivastava (MSR India) Expander Graphs Sparse,](https://reader033.vdocuments.net/reader033/viewer/2022043012/5fa9fe9adf2c9913832b65eb/html5/thumbnails/24.jpg)
Eigenvalues of 2-lifts (Bilu-Linial)
Given a 2-lift of G, create a signed adjacency matrix As with a -1 for crossing edges and a 1 for parallel edges
0 -‐1 0 0 1 -‐1 0 1 0 1 0 1 0 -‐1 0 0 0 -‐1 0 1 1 1 0 1 0
a0
c0
d0
e0
b0
a1
d1
e1
b1
c1
![Page 25: Adam Marcus (Crisply, Yale) Daniel Spielman (Yale) Nikhil …cs.yale.edu/homes/spielman/TALKS/erdos3raman.pdf · 2014-05-27 · Nikhil Srivastava (MSR India) Expander Graphs Sparse,](https://reader033.vdocuments.net/reader033/viewer/2022043012/5fa9fe9adf2c9913832b65eb/html5/thumbnails/25.jpg)
Eigenvalues of 2-lifts (Bilu-Linial)
Theorem: The eigenvalues of the 2-lift are the union of the eigenvalues of A (old)
and the eigenvalues of As (new)
0 -‐1 0 0 1 -‐1 0 1 0 1 0 1 0 -‐1 0 0 0 -‐1 0 1 1 1 0 1 0
a0
c0
d0
e0
b0
a1
d1
e1
b1
c1
![Page 26: Adam Marcus (Crisply, Yale) Daniel Spielman (Yale) Nikhil …cs.yale.edu/homes/spielman/TALKS/erdos3raman.pdf · 2014-05-27 · Nikhil Srivastava (MSR India) Expander Graphs Sparse,](https://reader033.vdocuments.net/reader033/viewer/2022043012/5fa9fe9adf2c9913832b65eb/html5/thumbnails/26.jpg)
Eigenvalues of 2-lifts (Bilu-Linial)
Conjecture: Every d-regular graph has a 2-lift in which all the new eigenvalues
have absolute value at most 2pd� 1
Theorem: The eigenvalues of the 2-lift are the union of the eigenvalues of A (old)
and the eigenvalues of As (new)
![Page 27: Adam Marcus (Crisply, Yale) Daniel Spielman (Yale) Nikhil …cs.yale.edu/homes/spielman/TALKS/erdos3raman.pdf · 2014-05-27 · Nikhil Srivastava (MSR India) Expander Graphs Sparse,](https://reader033.vdocuments.net/reader033/viewer/2022043012/5fa9fe9adf2c9913832b65eb/html5/thumbnails/27.jpg)
Eigenvalues of 2-lifts (Bilu-Linial)
Conjecture: Every d-regular graph has a 2-lift in which all the new eigenvalues
have absolute value at most 2pd� 1
Would give infinite families of Ramanujan Graphs: start with the complete graph, and keep lifting.
![Page 28: Adam Marcus (Crisply, Yale) Daniel Spielman (Yale) Nikhil …cs.yale.edu/homes/spielman/TALKS/erdos3raman.pdf · 2014-05-27 · Nikhil Srivastava (MSR India) Expander Graphs Sparse,](https://reader033.vdocuments.net/reader033/viewer/2022043012/5fa9fe9adf2c9913832b65eb/html5/thumbnails/28.jpg)
Eigenvalues of 2-lifts (Bilu-Linial)
Conjecture: Every d-regular graph has a 2-lift in which all the new eigenvalues
have absolute value at most 2pd� 1
We prove this in the bipartite case.
a 2-lift of a bipartite graph is bipartite
![Page 29: Adam Marcus (Crisply, Yale) Daniel Spielman (Yale) Nikhil …cs.yale.edu/homes/spielman/TALKS/erdos3raman.pdf · 2014-05-27 · Nikhil Srivastava (MSR India) Expander Graphs Sparse,](https://reader033.vdocuments.net/reader033/viewer/2022043012/5fa9fe9adf2c9913832b65eb/html5/thumbnails/29.jpg)
Eigenvalues of 2-lifts (Bilu-Linial)
Theorem: Every d-regular graph has a 2-lift in which all the new eigenvalues
have absolute value at most 2pd� 1
Trick: eigenvalues of bipartite graphs are symmetric about 0, so only need to bound largest
![Page 30: Adam Marcus (Crisply, Yale) Daniel Spielman (Yale) Nikhil …cs.yale.edu/homes/spielman/TALKS/erdos3raman.pdf · 2014-05-27 · Nikhil Srivastava (MSR India) Expander Graphs Sparse,](https://reader033.vdocuments.net/reader033/viewer/2022043012/5fa9fe9adf2c9913832b65eb/html5/thumbnails/30.jpg)
Eigenvalues of 2-lifts (Bilu-Linial)
Theorem: Every d-regular bipartite graph has a 2-lift in which all the new eigenvalues
have absolute value at most 2pd� 1
![Page 31: Adam Marcus (Crisply, Yale) Daniel Spielman (Yale) Nikhil …cs.yale.edu/homes/spielman/TALKS/erdos3raman.pdf · 2014-05-27 · Nikhil Srivastava (MSR India) Expander Graphs Sparse,](https://reader033.vdocuments.net/reader033/viewer/2022043012/5fa9fe9adf2c9913832b65eb/html5/thumbnails/31.jpg)
First idea: a random 2-lift
Specify a lift by
Pick s uniformly at random
s 2 {±1}m
![Page 32: Adam Marcus (Crisply, Yale) Daniel Spielman (Yale) Nikhil …cs.yale.edu/homes/spielman/TALKS/erdos3raman.pdf · 2014-05-27 · Nikhil Srivastava (MSR India) Expander Graphs Sparse,](https://reader033.vdocuments.net/reader033/viewer/2022043012/5fa9fe9adf2c9913832b65eb/html5/thumbnails/32.jpg)
First idea: a random 2-lift
Specify a lift by
Pick s uniformly at random
s 2 {±1}m
Are graphs for which this usually fails
![Page 33: Adam Marcus (Crisply, Yale) Daniel Spielman (Yale) Nikhil …cs.yale.edu/homes/spielman/TALKS/erdos3raman.pdf · 2014-05-27 · Nikhil Srivastava (MSR India) Expander Graphs Sparse,](https://reader033.vdocuments.net/reader033/viewer/2022043012/5fa9fe9adf2c9913832b65eb/html5/thumbnails/33.jpg)
First idea: a random 2-lift
Specify a lift by
Pick s uniformly at random
s 2 {±1}m
Are graphs for which this usually fails
Bilu and Linial proved G almost Ramanujan, implies new eigenvalues usually small. Improved by Puder and Agarwal-Kolla-Madan
![Page 34: Adam Marcus (Crisply, Yale) Daniel Spielman (Yale) Nikhil …cs.yale.edu/homes/spielman/TALKS/erdos3raman.pdf · 2014-05-27 · Nikhil Srivastava (MSR India) Expander Graphs Sparse,](https://reader033.vdocuments.net/reader033/viewer/2022043012/5fa9fe9adf2c9913832b65eb/html5/thumbnails/34.jpg)
The expected polynomial
Consider Es[ �As(x) ]
![Page 35: Adam Marcus (Crisply, Yale) Daniel Spielman (Yale) Nikhil …cs.yale.edu/homes/spielman/TALKS/erdos3raman.pdf · 2014-05-27 · Nikhil Srivastava (MSR India) Expander Graphs Sparse,](https://reader033.vdocuments.net/reader033/viewer/2022043012/5fa9fe9adf2c9913832b65eb/html5/thumbnails/35.jpg)
The expected polynomial
Consider Es[ �As(x) ]
max-root (�As(x)) 2pd� 1
Conclude there is an s so that
Prove max-root
⇣Es[ �As(x) ]
⌘ 2
pd� 1
Prove is an interlacing family �As(x)
![Page 36: Adam Marcus (Crisply, Yale) Daniel Spielman (Yale) Nikhil …cs.yale.edu/homes/spielman/TALKS/erdos3raman.pdf · 2014-05-27 · Nikhil Srivastava (MSR India) Expander Graphs Sparse,](https://reader033.vdocuments.net/reader033/viewer/2022043012/5fa9fe9adf2c9913832b65eb/html5/thumbnails/36.jpg)
The expected polynomial
Es[ �As(x) ] = µG(x)
Theorem (Godsil-Gutman ‘81):
the matching polynomial of G
![Page 37: Adam Marcus (Crisply, Yale) Daniel Spielman (Yale) Nikhil …cs.yale.edu/homes/spielman/TALKS/erdos3raman.pdf · 2014-05-27 · Nikhil Srivastava (MSR India) Expander Graphs Sparse,](https://reader033.vdocuments.net/reader033/viewer/2022043012/5fa9fe9adf2c9913832b65eb/html5/thumbnails/37.jpg)
The matching polynomial (Heilmann-Lieb ‘72)
mi = the number of matchings with i edges
µG(x) =X
i�0
x
n�2i(�1)imi
![Page 38: Adam Marcus (Crisply, Yale) Daniel Spielman (Yale) Nikhil …cs.yale.edu/homes/spielman/TALKS/erdos3raman.pdf · 2014-05-27 · Nikhil Srivastava (MSR India) Expander Graphs Sparse,](https://reader033.vdocuments.net/reader033/viewer/2022043012/5fa9fe9adf2c9913832b65eb/html5/thumbnails/38.jpg)
µG(x) = x
6 � 7x4 + 11x2 � 2
![Page 39: Adam Marcus (Crisply, Yale) Daniel Spielman (Yale) Nikhil …cs.yale.edu/homes/spielman/TALKS/erdos3raman.pdf · 2014-05-27 · Nikhil Srivastava (MSR India) Expander Graphs Sparse,](https://reader033.vdocuments.net/reader033/viewer/2022043012/5fa9fe9adf2c9913832b65eb/html5/thumbnails/39.jpg)
µG(x) = x
6 � 7x4 + 11x2 � 2
one matching with 0 edges
![Page 40: Adam Marcus (Crisply, Yale) Daniel Spielman (Yale) Nikhil …cs.yale.edu/homes/spielman/TALKS/erdos3raman.pdf · 2014-05-27 · Nikhil Srivastava (MSR India) Expander Graphs Sparse,](https://reader033.vdocuments.net/reader033/viewer/2022043012/5fa9fe9adf2c9913832b65eb/html5/thumbnails/40.jpg)
µG(x) = x
6 � 7x4 + 11x2 � 2
7 matchings with 1 edge
![Page 41: Adam Marcus (Crisply, Yale) Daniel Spielman (Yale) Nikhil …cs.yale.edu/homes/spielman/TALKS/erdos3raman.pdf · 2014-05-27 · Nikhil Srivastava (MSR India) Expander Graphs Sparse,](https://reader033.vdocuments.net/reader033/viewer/2022043012/5fa9fe9adf2c9913832b65eb/html5/thumbnails/41.jpg)
µG(x) = x
6 � 7x4 + 11x2 � 2
![Page 42: Adam Marcus (Crisply, Yale) Daniel Spielman (Yale) Nikhil …cs.yale.edu/homes/spielman/TALKS/erdos3raman.pdf · 2014-05-27 · Nikhil Srivastava (MSR India) Expander Graphs Sparse,](https://reader033.vdocuments.net/reader033/viewer/2022043012/5fa9fe9adf2c9913832b65eb/html5/thumbnails/42.jpg)
µG(x) = x
6 � 7x4 + 11x2 � 2
![Page 43: Adam Marcus (Crisply, Yale) Daniel Spielman (Yale) Nikhil …cs.yale.edu/homes/spielman/TALKS/erdos3raman.pdf · 2014-05-27 · Nikhil Srivastava (MSR India) Expander Graphs Sparse,](https://reader033.vdocuments.net/reader033/viewer/2022043012/5fa9fe9adf2c9913832b65eb/html5/thumbnails/43.jpg)
Proof that Es[ �As(x) ] = µG(x)
Es[ det(xI �As) ]
x ±1 0 0 ±1 ±1! ±1 x ±1 0 0 0! 0 ±1 x ±1 0 0! 0 0 ±1 x ±1 0! ±1 0 0 ±1 x ±1! ±1 0 0 0 ±1 x!
Expand using permutations
![Page 44: Adam Marcus (Crisply, Yale) Daniel Spielman (Yale) Nikhil …cs.yale.edu/homes/spielman/TALKS/erdos3raman.pdf · 2014-05-27 · Nikhil Srivastava (MSR India) Expander Graphs Sparse,](https://reader033.vdocuments.net/reader033/viewer/2022043012/5fa9fe9adf2c9913832b65eb/html5/thumbnails/44.jpg)
Proof that Es[ �As(x) ] = µG(x)
Es[ det(xI �As) ]
x ±1 0 0 ±1 ±1! ±1 x ±1 0 0 0! 0 ±1 x ±1 0 0! 0 0 ±1 x ±1 0! ±1 0 0 ±1 x ±1! ±1 0 0 0 ±1 x!
Expand using permutations
same edge: same value
![Page 45: Adam Marcus (Crisply, Yale) Daniel Spielman (Yale) Nikhil …cs.yale.edu/homes/spielman/TALKS/erdos3raman.pdf · 2014-05-27 · Nikhil Srivastava (MSR India) Expander Graphs Sparse,](https://reader033.vdocuments.net/reader033/viewer/2022043012/5fa9fe9adf2c9913832b65eb/html5/thumbnails/45.jpg)
Proof that Es[ �As(x) ] = µG(x)
Es[ det(xI �As) ]
x ±1 0 0 ±1 ±1! ±1 x ±1 0 0 0! 0 ±1 x ±1 0 0! 0 0 ±1 x ±1 0! ±1 0 0 ±1 x ±1! ±1 0 0 0 ±1 x!
Expand using permutations
same edge: same value
![Page 46: Adam Marcus (Crisply, Yale) Daniel Spielman (Yale) Nikhil …cs.yale.edu/homes/spielman/TALKS/erdos3raman.pdf · 2014-05-27 · Nikhil Srivastava (MSR India) Expander Graphs Sparse,](https://reader033.vdocuments.net/reader033/viewer/2022043012/5fa9fe9adf2c9913832b65eb/html5/thumbnails/46.jpg)
Proof that Es[ �As(x) ] = µG(x)
Es[ det(xI �As) ]
x ±1 0 0 ±1 ±1! ±1 x ±1 0 0 0! 0 ±1 x ±1 0 0! 0 0 ±1 x ±1 0! ±1 0 0 ±1 x ±1! ±1 0 0 0 ±1 x!
Expand using permutations
Get 0 if hit any 0s
![Page 47: Adam Marcus (Crisply, Yale) Daniel Spielman (Yale) Nikhil …cs.yale.edu/homes/spielman/TALKS/erdos3raman.pdf · 2014-05-27 · Nikhil Srivastava (MSR India) Expander Graphs Sparse,](https://reader033.vdocuments.net/reader033/viewer/2022043012/5fa9fe9adf2c9913832b65eb/html5/thumbnails/47.jpg)
Proof that Es[ �As(x) ] = µG(x)
Es[ det(xI �As) ]
x ±1 0 0 ±1 ±1! ±1 x ±1 0 0 0! 0 ±1 x ±1 0 0! 0 0 ±1 x ±1 0! ±1 0 0 ±1 x ±1! ±1 0 0 0 ±1 x!
Expand using permutations
Get 0 if take just one entry for any edge
![Page 48: Adam Marcus (Crisply, Yale) Daniel Spielman (Yale) Nikhil …cs.yale.edu/homes/spielman/TALKS/erdos3raman.pdf · 2014-05-27 · Nikhil Srivastava (MSR India) Expander Graphs Sparse,](https://reader033.vdocuments.net/reader033/viewer/2022043012/5fa9fe9adf2c9913832b65eb/html5/thumbnails/48.jpg)
Proof that Es[ �As(x) ] = µG(x)
Es[ det(xI �As) ]
x ±1 0 0 ±1 ±1! ±1 x ±1 0 0 0! 0 ±1 x ±1 0 0! 0 0 ±1 x ±1 0! ±1 0 0 ±1 x ±1! ±1 0 0 0 ±1 x!
Expand using permutations
Only permutations that count are involutions
![Page 49: Adam Marcus (Crisply, Yale) Daniel Spielman (Yale) Nikhil …cs.yale.edu/homes/spielman/TALKS/erdos3raman.pdf · 2014-05-27 · Nikhil Srivastava (MSR India) Expander Graphs Sparse,](https://reader033.vdocuments.net/reader033/viewer/2022043012/5fa9fe9adf2c9913832b65eb/html5/thumbnails/49.jpg)
Proof that Es[ �As(x) ] = µG(x)
Es[ det(xI �As) ]
x ±1 0 0 ±1 ±1! ±1 x ±1 0 0 0! 0 ±1 x ±1 0 0! 0 0 ±1 x ±1 0! ±1 0 0 ±1 x ±1! ±1 0 0 0 ±1 x!
Expand using permutations
Only permutations that count are involutions
![Page 50: Adam Marcus (Crisply, Yale) Daniel Spielman (Yale) Nikhil …cs.yale.edu/homes/spielman/TALKS/erdos3raman.pdf · 2014-05-27 · Nikhil Srivastava (MSR India) Expander Graphs Sparse,](https://reader033.vdocuments.net/reader033/viewer/2022043012/5fa9fe9adf2c9913832b65eb/html5/thumbnails/50.jpg)
Proof that Es[ �As(x) ] = µG(x)
Es[ det(xI �As) ]
x ±1 0 0 ±1 ±1! ±1 x ±1 0 0 0! 0 ±1 x ±1 0 0! 0 0 ±1 x ±1 0! ±1 0 0 ±1 x ±1! ±1 0 0 0 ±1 x!
Expand using permutations
Only permutations that count are involutions
Correspond to matchings
![Page 51: Adam Marcus (Crisply, Yale) Daniel Spielman (Yale) Nikhil …cs.yale.edu/homes/spielman/TALKS/erdos3raman.pdf · 2014-05-27 · Nikhil Srivastava (MSR India) Expander Graphs Sparse,](https://reader033.vdocuments.net/reader033/viewer/2022043012/5fa9fe9adf2c9913832b65eb/html5/thumbnails/51.jpg)
The matching polynomial (Heilmann-Lieb ‘72)
µG(x) =X
i�0
x
n�2i(�1)imi
Theorem (Heilmann-Lieb) all the roots are real
![Page 52: Adam Marcus (Crisply, Yale) Daniel Spielman (Yale) Nikhil …cs.yale.edu/homes/spielman/TALKS/erdos3raman.pdf · 2014-05-27 · Nikhil Srivastava (MSR India) Expander Graphs Sparse,](https://reader033.vdocuments.net/reader033/viewer/2022043012/5fa9fe9adf2c9913832b65eb/html5/thumbnails/52.jpg)
The matching polynomial (Heilmann-Lieb ‘72)
µG(x) =X
i�0
x
n�2i(�1)imi
Theorem (Heilmann-Lieb) all the roots are real and have absolute value at most
2pd� 1
![Page 53: Adam Marcus (Crisply, Yale) Daniel Spielman (Yale) Nikhil …cs.yale.edu/homes/spielman/TALKS/erdos3raman.pdf · 2014-05-27 · Nikhil Srivastava (MSR India) Expander Graphs Sparse,](https://reader033.vdocuments.net/reader033/viewer/2022043012/5fa9fe9adf2c9913832b65eb/html5/thumbnails/53.jpg)
The matching polynomial (Heilmann-Lieb ‘72)
µG(x) =X
i�0
x
n�2i(�1)imi
Theorem (Heilmann-Lieb) all the roots are real and have absolute value at most
2pd� 1
Implies max-root
⇣Es[ �As(x) ]
⌘ 2
pd� 1
![Page 54: Adam Marcus (Crisply, Yale) Daniel Spielman (Yale) Nikhil …cs.yale.edu/homes/spielman/TALKS/erdos3raman.pdf · 2014-05-27 · Nikhil Srivastava (MSR India) Expander Graphs Sparse,](https://reader033.vdocuments.net/reader033/viewer/2022043012/5fa9fe9adf2c9913832b65eb/html5/thumbnails/54.jpg)
Interlacing
Polynomial
interlaces
if
q(x) =Qd�1
i=1 (x� �i)
p(x) =Qd
i=1(x� ↵i)
↵1 �1 ↵2 · · ·↵d�1 �d�1 ↵d
![Page 55: Adam Marcus (Crisply, Yale) Daniel Spielman (Yale) Nikhil …cs.yale.edu/homes/spielman/TALKS/erdos3raman.pdf · 2014-05-27 · Nikhil Srivastava (MSR India) Expander Graphs Sparse,](https://reader033.vdocuments.net/reader033/viewer/2022043012/5fa9fe9adf2c9913832b65eb/html5/thumbnails/55.jpg)
p1(x)
Common Interlacing
and have a common interlacing if can partition the line into intervals so that each interval contains one root from each poly
�1
�2
�3
p2(x)
�d�1
![Page 56: Adam Marcus (Crisply, Yale) Daniel Spielman (Yale) Nikhil …cs.yale.edu/homes/spielman/TALKS/erdos3raman.pdf · 2014-05-27 · Nikhil Srivastava (MSR India) Expander Graphs Sparse,](https://reader033.vdocuments.net/reader033/viewer/2022043012/5fa9fe9adf2c9913832b65eb/html5/thumbnails/56.jpg)
p1(x)
Common Interlacing
and have a common interlacing if can partition the line into intervals so that each interval contains one root from each poly
�1
�2
�3
p2(x)
�d�1
) ) ) ) ( ( ( (
![Page 57: Adam Marcus (Crisply, Yale) Daniel Spielman (Yale) Nikhil …cs.yale.edu/homes/spielman/TALKS/erdos3raman.pdf · 2014-05-27 · Nikhil Srivastava (MSR India) Expander Graphs Sparse,](https://reader033.vdocuments.net/reader033/viewer/2022043012/5fa9fe9adf2c9913832b65eb/html5/thumbnails/57.jpg)
If p1 and p2 have a common interlacing,
for some i.
max-root (pi) max-root (Ei [ pi ])
Largest root of average
Common Interlacing
�1
�2
�3
�d�1
![Page 58: Adam Marcus (Crisply, Yale) Daniel Spielman (Yale) Nikhil …cs.yale.edu/homes/spielman/TALKS/erdos3raman.pdf · 2014-05-27 · Nikhil Srivastava (MSR India) Expander Graphs Sparse,](https://reader033.vdocuments.net/reader033/viewer/2022043012/5fa9fe9adf2c9913832b65eb/html5/thumbnails/58.jpg)
If p1 and p2 have a common interlacing,
for some i.
max-root (pi) max-root (Ei [ pi ])
Largest root of average
Common Interlacing
�1
�2
�3
�d�1
![Page 59: Adam Marcus (Crisply, Yale) Daniel Spielman (Yale) Nikhil …cs.yale.edu/homes/spielman/TALKS/erdos3raman.pdf · 2014-05-27 · Nikhil Srivastava (MSR India) Expander Graphs Sparse,](https://reader033.vdocuments.net/reader033/viewer/2022043012/5fa9fe9adf2c9913832b65eb/html5/thumbnails/59.jpg)
is an interlacing family
If the polynomials can be placed on the leaves of a tree so that when put average of descendants at nodes siblings have common interlacings
Ei [ p1,i ]
Ei,j [ pi,j ]
Interlacing Family of Polynomials {ps}s2{±1}m
p1,�1 p�1,�1p�1,1
Ei [ p�1,i ]
p1,1
![Page 60: Adam Marcus (Crisply, Yale) Daniel Spielman (Yale) Nikhil …cs.yale.edu/homes/spielman/TALKS/erdos3raman.pdf · 2014-05-27 · Nikhil Srivastava (MSR India) Expander Graphs Sparse,](https://reader033.vdocuments.net/reader033/viewer/2022043012/5fa9fe9adf2c9913832b65eb/html5/thumbnails/60.jpg)
is an interlacing family
Interlacing Family of Polynomials
If the polynomials can be placed on the leaves of a tree so that when put average of descendants at nodes siblings have common interlacings
{ps}s2{±1}m
p1
p;
p1,�1 p�1,�1p�1,1
p�1
p1,1
![Page 61: Adam Marcus (Crisply, Yale) Daniel Spielman (Yale) Nikhil …cs.yale.edu/homes/spielman/TALKS/erdos3raman.pdf · 2014-05-27 · Nikhil Srivastava (MSR India) Expander Graphs Sparse,](https://reader033.vdocuments.net/reader033/viewer/2022043012/5fa9fe9adf2c9913832b65eb/html5/thumbnails/61.jpg)
Interlacing Family of Polynomials There is an s so that
Theorem:
max-root (ps(x)) max-root
⇣Es[ ps(x) ]
⌘
p1,1
p1
p;
p1,�1 p�1,�1p�1,1
p�1
![Page 62: Adam Marcus (Crisply, Yale) Daniel Spielman (Yale) Nikhil …cs.yale.edu/homes/spielman/TALKS/erdos3raman.pdf · 2014-05-27 · Nikhil Srivastava (MSR India) Expander Graphs Sparse,](https://reader033.vdocuments.net/reader033/viewer/2022043012/5fa9fe9adf2c9913832b65eb/html5/thumbnails/62.jpg)
An interlacing family Theorem: Let ps(x) = �As(x)
is an interlacing family {ps}s2{±1}m
![Page 63: Adam Marcus (Crisply, Yale) Daniel Spielman (Yale) Nikhil …cs.yale.edu/homes/spielman/TALKS/erdos3raman.pdf · 2014-05-27 · Nikhil Srivastava (MSR India) Expander Graphs Sparse,](https://reader033.vdocuments.net/reader033/viewer/2022043012/5fa9fe9adf2c9913832b65eb/html5/thumbnails/63.jpg)
Interlacing
p1(x) and have a common interlacing iff p2(x)
�p1(x) + (1� �)p2(x) is real rooted for all 0 � 1
�1
�2
�3
�d�1
![Page 64: Adam Marcus (Crisply, Yale) Daniel Spielman (Yale) Nikhil …cs.yale.edu/homes/spielman/TALKS/erdos3raman.pdf · 2014-05-27 · Nikhil Srivastava (MSR India) Expander Graphs Sparse,](https://reader033.vdocuments.net/reader033/viewer/2022043012/5fa9fe9adf2c9913832b65eb/html5/thumbnails/64.jpg)
To prove interlacing family
Let ps1,. . . ,sk(x) = Esk+1,. . . ,sm
[ ps1,. . . ,sm(x) ]
![Page 65: Adam Marcus (Crisply, Yale) Daniel Spielman (Yale) Nikhil …cs.yale.edu/homes/spielman/TALKS/erdos3raman.pdf · 2014-05-27 · Nikhil Srivastava (MSR India) Expander Graphs Sparse,](https://reader033.vdocuments.net/reader033/viewer/2022043012/5fa9fe9adf2c9913832b65eb/html5/thumbnails/65.jpg)
To prove interlacing family
Let
Need to prove that for all , s1, . . . , sk
�ps1,. . . ,sk,1(x) + (1� �)ps1,. . . ,sk,�1(x)
� 2 [0, 1]
is real rooted
ps1,. . . ,sk(x) = Esk+1,. . . ,sm
[ ps1,. . . ,sm(x) ]
![Page 66: Adam Marcus (Crisply, Yale) Daniel Spielman (Yale) Nikhil …cs.yale.edu/homes/spielman/TALKS/erdos3raman.pdf · 2014-05-27 · Nikhil Srivastava (MSR India) Expander Graphs Sparse,](https://reader033.vdocuments.net/reader033/viewer/2022043012/5fa9fe9adf2c9913832b65eb/html5/thumbnails/66.jpg)
To prove interlacing family
Let
Need to prove that for all , s1, . . . , sk
�ps1,. . . ,sk,1(x) + (1� �)ps1,. . . ,sk,�1(x)
� 2 [0, 1]
s1, . . . , sk
sk+2, . . . , sm
ps1,. . . ,sk(x) = Esk+1,. . . ,sm
[ ps1,. . . ,sm(x) ]
sk+1
are fixed is 1 with probability , -1 with �
are uniformly
1� �
±1
is real rooted
![Page 67: Adam Marcus (Crisply, Yale) Daniel Spielman (Yale) Nikhil …cs.yale.edu/homes/spielman/TALKS/erdos3raman.pdf · 2014-05-27 · Nikhil Srivastava (MSR India) Expander Graphs Sparse,](https://reader033.vdocuments.net/reader033/viewer/2022043012/5fa9fe9adf2c9913832b65eb/html5/thumbnails/67.jpg)
Generalization of Heilmann-Lieb
We prove
Es2{±1}m
[ ps(x) ] is real rooted
for every independent distribution on the entries of s
![Page 68: Adam Marcus (Crisply, Yale) Daniel Spielman (Yale) Nikhil …cs.yale.edu/homes/spielman/TALKS/erdos3raman.pdf · 2014-05-27 · Nikhil Srivastava (MSR India) Expander Graphs Sparse,](https://reader033.vdocuments.net/reader033/viewer/2022043012/5fa9fe9adf2c9913832b65eb/html5/thumbnails/68.jpg)
Generalization of Heilmann-Lieb
We prove
Es2{±1}m
[ ps(x) ] is real rooted
for every independent distribution on the entries of s
![Page 69: Adam Marcus (Crisply, Yale) Daniel Spielman (Yale) Nikhil …cs.yale.edu/homes/spielman/TALKS/erdos3raman.pdf · 2014-05-27 · Nikhil Srivastava (MSR India) Expander Graphs Sparse,](https://reader033.vdocuments.net/reader033/viewer/2022043012/5fa9fe9adf2c9913832b65eb/html5/thumbnails/69.jpg)
Mixed Characteristic Polynomials
For independently chosen random vectors
is their mixed characteristic polynomial.
a1, ..., an
E⇥poly(
Pi aia
Ti )
⇤= µ(A1, ..., An)
Theorem: Mixed characteristic polynomials are real rooted. Proof: Using theory of real stable polynomials.
![Page 70: Adam Marcus (Crisply, Yale) Daniel Spielman (Yale) Nikhil …cs.yale.edu/homes/spielman/TALKS/erdos3raman.pdf · 2014-05-27 · Nikhil Srivastava (MSR India) Expander Graphs Sparse,](https://reader033.vdocuments.net/reader033/viewer/2022043012/5fa9fe9adf2c9913832b65eb/html5/thumbnails/70.jpg)
Mixed Characteristic Polynomials
For independently chosen random vectors
is their mixed characteristic polynomial.
a1, ..., an
E⇥poly(
Pi aia
Ti )
⇤= µ(A1, ..., An)
Obstacle: our matrix is a sum of random rank-2 matrices
0 -‐1 -‐1 0
0 1 1 0
or
![Page 71: Adam Marcus (Crisply, Yale) Daniel Spielman (Yale) Nikhil …cs.yale.edu/homes/spielman/TALKS/erdos3raman.pdf · 2014-05-27 · Nikhil Srivastava (MSR India) Expander Graphs Sparse,](https://reader033.vdocuments.net/reader033/viewer/2022043012/5fa9fe9adf2c9913832b65eb/html5/thumbnails/71.jpg)
Mixed Characteristic Polynomials
For independently chosen random vectors
is their mixed characteristic polynomial.
a1, ..., an
E⇥poly(
Pi aia
Ti )
⇤= µ(A1, ..., An)
Solution: add to the diagonal
1 -‐1 -‐1 1
1 1 1 1
or
![Page 72: Adam Marcus (Crisply, Yale) Daniel Spielman (Yale) Nikhil …cs.yale.edu/homes/spielman/TALKS/erdos3raman.pdf · 2014-05-27 · Nikhil Srivastava (MSR India) Expander Graphs Sparse,](https://reader033.vdocuments.net/reader033/viewer/2022043012/5fa9fe9adf2c9913832b65eb/html5/thumbnails/72.jpg)
Generalization of Heilmann-Lieb
We prove
Es2{±1}m
[ ps(x) ] is real rooted
for every independent distribution on the entries of s
Implies is an interlacing family �As(x)
![Page 73: Adam Marcus (Crisply, Yale) Daniel Spielman (Yale) Nikhil …cs.yale.edu/homes/spielman/TALKS/erdos3raman.pdf · 2014-05-27 · Nikhil Srivastava (MSR India) Expander Graphs Sparse,](https://reader033.vdocuments.net/reader033/viewer/2022043012/5fa9fe9adf2c9913832b65eb/html5/thumbnails/73.jpg)
Generalization of Heilmann-Lieb
We prove
Es2{±1}m
[ ps(x) ] is real rooted
for every independent distribution on the entries of s
max-root (�As(x)) 2pd� 1
Conclude there is an s so that
Implies is an interlacing family �As(x)
![Page 74: Adam Marcus (Crisply, Yale) Daniel Spielman (Yale) Nikhil …cs.yale.edu/homes/spielman/TALKS/erdos3raman.pdf · 2014-05-27 · Nikhil Srivastava (MSR India) Expander Graphs Sparse,](https://reader033.vdocuments.net/reader033/viewer/2022043012/5fa9fe9adf2c9913832b65eb/html5/thumbnails/74.jpg)
Universal Covers
The universal cover of a graph G is a tree T of which G is a quotient. vertices map to vertices edges map to edges homomorphism on neighborhoods Is the tree of non-backtracking walks in G.
The universal cover of a d-regular graph is the infinite d-regular tree.
![Page 75: Adam Marcus (Crisply, Yale) Daniel Spielman (Yale) Nikhil …cs.yale.edu/homes/spielman/TALKS/erdos3raman.pdf · 2014-05-27 · Nikhil Srivastava (MSR India) Expander Graphs Sparse,](https://reader033.vdocuments.net/reader033/viewer/2022043012/5fa9fe9adf2c9913832b65eb/html5/thumbnails/75.jpg)
Quotients of Trees
d-regular Ramanujan as quotient of infinite d-ary tree Spectral radius and norm of inf d-ary tree are
2pd� 1
![Page 76: Adam Marcus (Crisply, Yale) Daniel Spielman (Yale) Nikhil …cs.yale.edu/homes/spielman/TALKS/erdos3raman.pdf · 2014-05-27 · Nikhil Srivastava (MSR India) Expander Graphs Sparse,](https://reader033.vdocuments.net/reader033/viewer/2022043012/5fa9fe9adf2c9913832b65eb/html5/thumbnails/76.jpg)
Godsil’s Proof of Heilmann-Lieb
T(G,v) : the path tree of G at v vertices are paths in G starting at v edges to paths differing in one step
![Page 77: Adam Marcus (Crisply, Yale) Daniel Spielman (Yale) Nikhil …cs.yale.edu/homes/spielman/TALKS/erdos3raman.pdf · 2014-05-27 · Nikhil Srivastava (MSR India) Expander Graphs Sparse,](https://reader033.vdocuments.net/reader033/viewer/2022043012/5fa9fe9adf2c9913832b65eb/html5/thumbnails/77.jpg)
Godsil’s Proof of Heilmann-Lieb
a
c
d
e b
a
ab
e
a
e
ab
e
ab
c
ab
d
e
a
![Page 78: Adam Marcus (Crisply, Yale) Daniel Spielman (Yale) Nikhil …cs.yale.edu/homes/spielman/TALKS/erdos3raman.pdf · 2014-05-27 · Nikhil Srivastava (MSR India) Expander Graphs Sparse,](https://reader033.vdocuments.net/reader033/viewer/2022043012/5fa9fe9adf2c9913832b65eb/html5/thumbnails/78.jpg)
Godsil’s Proof of Heilmann-Lieb
T(G,v) : the path tree of G at v vertices are paths in G starting at v edges to paths differing in one step
Theorem: The matching polynomial divides the characteristic polynomial of T(G,v)
![Page 79: Adam Marcus (Crisply, Yale) Daniel Spielman (Yale) Nikhil …cs.yale.edu/homes/spielman/TALKS/erdos3raman.pdf · 2014-05-27 · Nikhil Srivastava (MSR India) Expander Graphs Sparse,](https://reader033.vdocuments.net/reader033/viewer/2022043012/5fa9fe9adf2c9913832b65eb/html5/thumbnails/79.jpg)
Godsil’s Proof of Heilmann-Lieb
T(G,v) : the path tree of G at v vertices are paths in G starting at v edges to paths differing in one step
Theorem: The matching polynomial divides the characteristic polynomial of T(G,v)
Is a subgraph of infinite tree, so has smaller spectral radius
![Page 80: Adam Marcus (Crisply, Yale) Daniel Spielman (Yale) Nikhil …cs.yale.edu/homes/spielman/TALKS/erdos3raman.pdf · 2014-05-27 · Nikhil Srivastava (MSR India) Expander Graphs Sparse,](https://reader033.vdocuments.net/reader033/viewer/2022043012/5fa9fe9adf2c9913832b65eb/html5/thumbnails/80.jpg)
Quotients of Trees
(c,d)-biregular bipartite Ramanujan as quotient of infinite (c,d)-ary tree Spectral radius
pd� 1 +
pc� 1
For (c,d)-regular bipartite Ramanujan graphs
pd� 1 +
pc� 1
![Page 81: Adam Marcus (Crisply, Yale) Daniel Spielman (Yale) Nikhil …cs.yale.edu/homes/spielman/TALKS/erdos3raman.pdf · 2014-05-27 · Nikhil Srivastava (MSR India) Expander Graphs Sparse,](https://reader033.vdocuments.net/reader033/viewer/2022043012/5fa9fe9adf2c9913832b65eb/html5/thumbnails/81.jpg)
Questions
Non-bipartite Ramanujan Graphs of every degree?
Efficient constructions?
Explicit constructions?