The Topology of Graph Configuration Spaces David G.C. Handron Carnegie Mellon University...

29
The Topology of Graph Configuration Spaces David G.C. Handron Carnegie Mellon University [email protected]

Transcript of The Topology of Graph Configuration Spaces David G.C. Handron Carnegie Mellon University...

Page 1: The Topology of Graph Configuration Spaces David G.C. Handron Carnegie Mellon University handron@andrew.cmu.edu.

The Topology of Graph Configuration Spaces

David G.C. Handron

Carnegie Mellon [email protected]

Page 2: The Topology of Graph Configuration Spaces David G.C. Handron Carnegie Mellon University handron@andrew.cmu.edu.

The Topology of Graph Configuration Spaces

1. Configuration Spaces2. Graphs3. Topology4. Morse Theory5. Results

Page 3: The Topology of Graph Configuration Spaces David G.C. Handron Carnegie Mellon University handron@andrew.cmu.edu.

Configuration Spaces

Term configuration space is commonly used to refer to the space of configurations of k distinct points in a manifold M

This is a subspace of

Page 4: The Topology of Graph Configuration Spaces David G.C. Handron Carnegie Mellon University handron@andrew.cmu.edu.

Cyclic Configuration Spaces:

Other Configuration Spaces

Page 5: The Topology of Graph Configuration Spaces David G.C. Handron Carnegie Mellon University handron@andrew.cmu.edu.

Cyclic Configuration Spaces: Not all points must be distinct, only those whose indices differ by one (mod k).

Other Configuration Spaces

Page 6: The Topology of Graph Configuration Spaces David G.C. Handron Carnegie Mellon University handron@andrew.cmu.edu.

Cyclic Configuration Spaces: Not all points must be distinct, only those whose indices differ by one (mod k).

Used (e.g. by Farber and Tabachnikov) to study periodic billiard paths.

Other Configuration Spaces

Page 7: The Topology of Graph Configuration Spaces David G.C. Handron Carnegie Mellon University handron@andrew.cmu.edu.

Cyclic Configuration Spaces: Not all points must be distinct, only those whose indices differ by one (mod k).

Used (e.g. by Farber and Tabachnikov) to study periodic billiard paths.

Path Configuration Spaces:

Other Configuration Spaces

Page 8: The Topology of Graph Configuration Spaces David G.C. Handron Carnegie Mellon University handron@andrew.cmu.edu.

Cyclic Configuration Spaces: Not all points must be distinct, only those whose indices differ by one (mod k).

Used (e.g. by Farber and Tabachnikov) to study periodic billiard paths.

Path Configuration Spaces:

Used by myself to study non-cyclic billiard paths.

Other Configuration Spaces

Page 9: The Topology of Graph Configuration Spaces David G.C. Handron Carnegie Mellon University handron@andrew.cmu.edu.

Graph Configuration Spaces

v1

v2

v3

v4v5

v6

Configuration of points in a manifold

One point for each vertex of a graph

Points corresponding to adjacent vertices must be distinct

Page 10: The Topology of Graph Configuration Spaces David G.C. Handron Carnegie Mellon University handron@andrew.cmu.edu.

Cyclic configuration spaces correspond to graphs that form a loop

Path configuration spaces correspond to graphs that form a continuous path

Page 11: The Topology of Graph Configuration Spaces David G.C. Handron Carnegie Mellon University handron@andrew.cmu.edu.

Graph Theory

A graph G consists of:

(1) a finite set V(G) of vertices, and

(2) a set E(G) of unordered pairs of vertices.

The elements of E(G) are the edges of the graph.

Page 12: The Topology of Graph Configuration Spaces David G.C. Handron Carnegie Mellon University handron@andrew.cmu.edu.

v1

v2

v3

v4v5

v6

G:

V(G)={v1, v2, v3, v4, v5, v6}

E(G)={{v1, v3}, {v2, v3},...}

Page 13: The Topology of Graph Configuration Spaces David G.C. Handron Carnegie Mellon University handron@andrew.cmu.edu.

v1

v2

v3

v4v5

v6

G:

V(G)={v1, v2, v3, v4, v5, v6}

E(G)={{v1, v3}, {v2, v3},...} = {e13, e23, e34, ...}

Page 14: The Topology of Graph Configuration Spaces David G.C. Handron Carnegie Mellon University handron@andrew.cmu.edu.

Subgraphs

A subgraph of a graph G is a graph Hsuch that

(1) Every vertex of H is a vertex of G.(2) Every edge of H is an edge of G.

If V' is a subset of V(G), the induced graph G[V'] includes all the edges of G joining vertices in V'.

Page 15: The Topology of Graph Configuration Spaces David G.C. Handron Carnegie Mellon University handron@andrew.cmu.edu.

Contractions

v1

v2

v3

v4v5

v6

v1

v2

[v3, v4]

v5

v6

G:

G/e34:

We can contract a graph with respect to an edge ...

...by identifying the vertices joined by that edge

Page 16: The Topology of Graph Configuration Spaces David G.C. Handron Carnegie Mellon University handron@andrew.cmu.edu.

Contractions, cont.

v1

v2

v3

v4v5

v6

v2 [v1, v3, v4, v6]

v5

G:

G/{e13, e34, e46}:

We can contract with respect to a set of edges.

Simply identify each pair of vertices.

Page 17: The Topology of Graph Configuration Spaces David G.C. Handron Carnegie Mellon University handron@andrew.cmu.edu.

Partitions

The vertices of a graph G can be partitioned into a collection of disjoint subsets.

This partition determines a subgraph of G.

is the induced subgraph of the partition P.

Page 18: The Topology of Graph Configuration Spaces David G.C. Handron Carnegie Mellon University handron@andrew.cmu.edu.

Partitions and Induced Graphs

For each partition P of a graph G, there is a corresponding contraction:

contract all the edges in G[P].

Page 19: The Topology of Graph Configuration Spaces David G.C. Handron Carnegie Mellon University handron@andrew.cmu.edu.

Examplesv1

v2

v3

v4

[v1, v2]

v3

v4

Two partitions {{v1,v2,v4},{v3}} and {{v1,v2},{v3},{v4}} may induce the same edge set...

...and produce the same quotient.

A partition is connected if each is a connected graph. It can be shown that connected partitions induce the same subgraphs and partitions.

Page 20: The Topology of Graph Configuration Spaces David G.C. Handron Carnegie Mellon University handron@andrew.cmu.edu.

Topology

The goal of this work is to describe topological invarients of a graph-configuration space. This description will involve properties of the graph,

and topological properties of the underlying manifold.

Today, we'll be concerned with the Euler characteristic of these configuration spaces.

Page 21: The Topology of Graph Configuration Spaces David G.C. Handron Carnegie Mellon University handron@andrew.cmu.edu.

Euler Characteristic

The Euler characteristic of a polyhedron is commonly describes as v-e+f.

Page 22: The Topology of Graph Configuration Spaces David G.C. Handron Carnegie Mellon University handron@andrew.cmu.edu.

Euler Characteristic

The Euler characteristic of a polyhedron is commonly describes as v-e+f.

Cube: 8-12+6=2

Page 23: The Topology of Graph Configuration Spaces David G.C. Handron Carnegie Mellon University handron@andrew.cmu.edu.

Euler Characteristic

The Euler characteristic of a polyhedron is commonly describes as v-e+f.

Cube: 8-12+6=2 Tetrahedron: 4-6+4=2

Page 24: The Topology of Graph Configuration Spaces David G.C. Handron Carnegie Mellon University handron@andrew.cmu.edu.

Euler Characteristic

The Euler characteristic of a polyhedron is commonly describes as v-e+f.

Cube: 8-12+6=2 Tetrahedron: 4-6+4=2 Both topologically equivalent

(homeomorphic) to a sphere.

Page 25: The Topology of Graph Configuration Spaces David G.C. Handron Carnegie Mellon University handron@andrew.cmu.edu.

Euler Characteristic of a CW-complex

A CW-complex is similar to a polyhedron. It is constructed out of cells (vertices, edges, faces,

etc.) of varying dimension.

Each cell is attached along its edge to cells of one lower dimension.

If n(i) is the number of cells with dimension i, then

Page 26: The Topology of Graph Configuration Spaces David G.C. Handron Carnegie Mellon University handron@andrew.cmu.edu.

Morse Theory

A Morse function is a smooth function from a manifold M to R

which has non-degenerate critical points.

Page 27: The Topology of Graph Configuration Spaces David G.C. Handron Carnegie Mellon University handron@andrew.cmu.edu.

Non-Degenerate Critical Points

A point p in M is a critical point if df=0. In coordinates this means

A critical point is non-degenerate if the Hessian matrix of second partial derivatives has nonzero determinate.

Page 28: The Topology of Graph Configuration Spaces David G.C. Handron Carnegie Mellon University handron@andrew.cmu.edu.

Index of a Non-Degenerate Critical Point

The index of a non-degenerate critical point is the number of negative eigenvalues of the Hessian

matrix.

I'll switch to the whiteboard to explain what this is all about...

Page 29: The Topology of Graph Configuration Spaces David G.C. Handron Carnegie Mellon University handron@andrew.cmu.edu.

Morse Theory Results

(1) If f is a Morse function on M, then M is homotopy equivalent to a CW-complex with one cell of dimension i for each critical point of f with index i.

(2) A similar result holds for a stratified Morse function on a stratified space.