26
Eigenvalues and geometric representations of graphs László Lovász Microsoft Research One Microsoft Way, Redmond, WA 98052 [email protected]

Eigenvalues and geometric representations of graphs L á szl ó Lov á sz Microsoft Research

Embed Size (px)

DESCRIPTION

Eigenvalues and geometric representations of graphs L á szl ó Lov á sz Microsoft Research One Microsoft Way, Redmond, WA 98052 [email protected]. Every 3-connected planar graph is the skeleton of a convex 3-polytope. Steinitz 1922. 3-connected planar graph. - PowerPoint PPT Presentation

Citation preview

Page 1: Eigenvalues  and geometric representations  of graphs L á szl ó  Lov á sz  Microsoft Research

Eigenvalues and

geometric representations

of graphs

László Lovász

Microsoft Research

One Microsoft Way, Redmond, WA 98052

[email protected]

Page 2: Eigenvalues  and geometric representations  of graphs L á szl ó  Lov á sz  Microsoft Research

Steinitz 1922

Every 3-connected planar graphis the skeleton of a convex 3-polytope.

3-connected planar graph

Page 3: Eigenvalues  and geometric representations  of graphs L á szl ó  Lov á sz  Microsoft Research

Representation by special polyhedra

Every 3-connected planar graph

is the skeleton of a convex polytope

such that every edge

touches the unit sphere

Koebe-Andreev-Thurston

Page 4: Eigenvalues  and geometric representations  of graphs L á szl ó  Lov á sz  Microsoft Research

From polyhedra to circles

horizon

Page 5: Eigenvalues  and geometric representations  of graphs L á szl ó  Lov á sz  Microsoft Research

From polyhedra to the polar

Page 6: Eigenvalues  and geometric representations  of graphs L á szl ó  Lov á sz  Microsoft Research

Coin representation

Every planar graph can be represented by touching circles

Koebe (1936)

Discrete Riemann Mapping Theorem

Page 7: Eigenvalues  and geometric representations  of graphs L á szl ó  Lov á sz  Microsoft Research

Representation by orthogonal circles:

A planar triangulation can be represented byorthogonal circles

no separating 3- or 4-cycles Andreev

Thurston

/ 2ija

Page 8: Eigenvalues  and geometric representations  of graphs L á szl ó  Lov á sz  Microsoft Research

The Colin de Verdière number

G: connected graph

Roughly: multiplicity of second largest eigenvalue

of adjacency matrix

But: non-degeneracy condition on weightings

Largest has multiplicity 1.

But: maximize over weighting the edges and diagonal entries

Page 9: Eigenvalues  and geometric representations  of graphs L á szl ó  Lov á sz  Microsoft Research

Mii arbitrary

Strong Arnold Property

( ) max corank ( )G M

normalization

M=(Mij): symmetric VxV matrix•

Mij

<0, if ijE

0, if ,ij E i j •

M has =1 negative eigenvalue•

( )ijX X symmetric, 0 for andijX ij E i j

00MX X •

The Colin de Verdière number of a graph

Page 10: Eigenvalues  and geometric representations  of graphs L á szl ó  Lov á sz  Microsoft Research

Basic Properties

μ(G) is minor monotone

deleting and contracting edges

μk is polynomial timedecidable for fixed k

for μ>2, μ(G) is invariant under subdivision

for μ>3, μ(G) is invariant under Δ-Y transformation

Page 11: Eigenvalues  and geometric representations  of graphs L á szl ó  Lov á sz  Microsoft Research

μ(G)1 G is a path

Special values

μ(G)3 G is a planar

Colin de Verdière, using pde’sVan der Holst, elementary proof

μ(G)2 G is outerplanar

Page 12: Eigenvalues  and geometric representations  of graphs L á szl ó  Lov á sz  Microsoft Research

0x 0x 0x

supp ( ), supp ( )xx are connected.

Van der Holst’s lemma

Courant’s Nodal Theorem

0Mx

supp( ) minimalx

Page 13: Eigenvalues  and geometric representations  of graphs L á szl ó  Lov á sz  Microsoft Research

0

0 0

0

0

0

_

_

+

+

_0

+

_

G planar corank of M is at most 3.

Page 14: Eigenvalues  and geometric representations  of graphs L á szl ó  Lov á sz  Microsoft Research

representation of G in μ

Nullspace representation:

0ij jj

M u

1

2

n

u

u

u

11 12 1

21 22 2

1 2

...

...

...n n n

x x x

x x x

x x x

basis of nullspace of M1 2 .. :.x x x

corank of M is at most 3 G planar .

Page 15: Eigenvalues  and geometric representations  of graphs L á szl ó  Lov á sz  Microsoft Research

Van der Holst’s Lemma, geometric form

like convex polytopes?

or…

connected

Page 16: Eigenvalues  and geometric representations  of graphs L á szl ó  Lov á sz  Microsoft Research

G 3-connected planar

nullspace representation,scaled to unit vectors,gives embedding in S2

L-Schrijver

G 3-connected planar

nullspace representationcan be scaled to convex polytope L

Page 17: Eigenvalues  and geometric representations  of graphs L á szl ó  Lov á sz  Microsoft Research

nullspace representationplanar embedding

Page 18: Eigenvalues  and geometric representations  of graphs L á szl ó  Lov á sz  Microsoft Research

P P*

Colin de Verdière matrix M

Steinitz representationP

( )uvMp q u v

u

v

q

p

Page 19: Eigenvalues  and geometric representations  of graphs L á szl ó  Lov á sz  Microsoft Research

μ(G)1 G is a path

Special values

μ(G)3 G is a planar

Colin de Verdière, using pde’sVan der Holst, elementary proof

μ(G)2 G is outerplanar

μ(G)4 G is linklessly embeddable in 3-space

L - Schrijver

Page 20: Eigenvalues  and geometric representations  of graphs L á szl ó  Lov á sz  Microsoft Research

Linklessly embedable graphs

homological, homotopical,…equivalent

embedable in 3 without linked cycles

Apex graph

Page 21: Eigenvalues  and geometric representations  of graphs L á szl ó  Lov á sz  Microsoft Research

G linklessly embedable

G has no minor in the “Petersen family”

Robertson – Seymour - Thomas

Page 22: Eigenvalues  and geometric representations  of graphs L á szl ó  Lov á sz  Microsoft Research

G 4-connected

linkless embed. nullspace representation gives

linkless embedding in 3

?

G path nullspace representation gives

embedding in 1

properly normalized

G 2-connected

outerplanar nullspace representation gives

outerplanar embedding in 2

G 3-connected

planar nullspace representation gives

planar embedding in 2, and also

Steinitz representationL-Schrijver; L

Page 23: Eigenvalues  and geometric representations  of graphs L á szl ó  Lov á sz  Microsoft Research

μ(G)1 G is a path

μ(G)n-4 complement G is planar_

~

Kotlov-L-Vempala

Special values

μ(G)3 G is a planar

Colin de Verdière, using pde’sVan der Holst, elementary proof

μ(G)2 G is outerplanar

μ(G)4 G is linklessly embeddable in 3-space

L - SchrijverKoebe-Andreevrepresentation

Page 24: Eigenvalues  and geometric representations  of graphs L á szl ó  Lov á sz  Microsoft Research

The Gram representation

1 1 1 1( )TA Q M Q Q MQ J pos semidefinite

, : diag( )M Q

1, ( );

1, ( ).Ti j

ij E Gu u

ij E G

if

if

Kotlov – L - Vempala

1( )T nij i j iA u u u

Gram representation

Page 25: Eigenvalues  and geometric representations  of graphs L á szl ó  Lov á sz  Microsoft Research

Properties of the Gram representation

ui is a vertex of P

1: conv( ,..., )nP u u | | 1iu exceptional

Assume: G has no twin nodes, and | | 1iu

( ) i juij E G u is an edge of P

0 int P

If G has no twin nodes, and μ(G)n-4, then

is planar.G

Page 26: Eigenvalues  and geometric representations  of graphs L á szl ó  Lov á sz  Microsoft Research