《谱图论》读书笔记（第二章）-CSDN博客

本文链接：https://blog.csdn.net/meteor_0033/article/details/100167285

Chapter 2. 等周问题

2.1 历史(History)

2.2 图的Cheeger常数(The Cheeger constant of a graph)

2.3 图的边缘展开(The edge expansion of a graph)

2.4 图的顶点展开(The vertex expansion of a graph)

Chapter 2. 等周问题

2.1 历史(History)

One of the earliest problems in geometry was the isoperimetric problem, which was considered by the ancient Greeks. The problem is to find, among all closed curves of a given length, the one which encloses the maximum area. The basic isoperimetric problem for graphs is essentially the same. Namely, remove as little of the graph as possible to separate out a subset of vertices of some desired “size”. Here the size of a subset of vertices may mean the number of vertices, the number of edges, or some other appropriate measure defined on graphs. A typical case is to remove as few edges as possible to disconnect the graph into two parts of almost equal size. Such problems are usually called separator problems and are particularly useful in a number of areas including recursive algorithms, network design, and parallel architectures for computers, for example [181].

In a graph, a subset of edges which disconnects the graph is called a cut. Cuts arise naturally in the study of connectivity of graphs where the sizes of the disconnected parts are not of concern. Isoperimetric problems examine optimal relations between the size of the cut and the sizes of the separated parts. Many different names are used for various versions of isoperimetric problems (such as the conductance of a graph, the isoperimetric number, etc.) The concepts are all quite similar, but the differences are due to the varying definitions of cuts and sizes.

We distinguish two types of cuts. A vertex-cut is a subset of vertices whose removal disconnects the graph. Similarly, an edge-cut is a subset of edges whose removal separates the graph. The size of a subset of vertices depends on either the number of vertices or the number of edges. Therefore, there are several combinations.

Roughly speaking, isoperimetric problems involving edge-cuts correspond in a natural way to Cheeger constants in spectral geometry. The formulation and the proof techniques are very similar. Cheeger constants were studied in the thesis of Cheeger [48], but they can be traced back to Polyá and Szegö [214]. We will follow tradition and call the discrete versions by the same names, such as the Cheeger constant and the Cheeger inequalities.

2.2 图的Cheeger常数(The Cheeger constant of a graph)

Before we discuss isoperimetric problems for graphs, let us first consider a measure on subsets of vertices. The typical measure just Before we discuss isoperimetric problems for graphs, let us first consider a measure on subsets of vertices. The typical measure just assigns weight 1 to each vertex, so the measure of a subset is its number of vertices. However, this implies that all vertices have the same measure. For some problems, this is appropriate only for regular graphs and does not work for general graphs. The measure we will use here takes into consideration the degree at a vertex. For a subset of the vertices of , we define vol 、the volume of , to be the sum of the degrees of the vertices in :assigns weight 1 to each vertex, so the measure of a subset is its number of vertices. However, this implies that all vertices have the same measure. For some problems, this is appropriate only for regular graphs and does not work for general graphs. The measure we will use here takes into consideration the degree at a vertex. For a subset of the vertices of , we define vol 、the volume of , to be the sum of the degrees of the vertices in :

for

NNext, we define the edge boundary of S to consist of all edges with exactly one endpoint in :ext, we define the edge boundary of to consist of all edges with exactly one endpoint in :

Let denote the complement of , i.e., . Clearly, where denotes the set of edges with one endpoint in and one endpoint in . Similarly, we can define the vertex boundary of to be the set of all vertices not in but adjacent to some vertex in , i.e.,

We are ready to pose the following questions:

Problem 1: For a fixed number , find a subset with such that the edge boundary contains as few edges as possible.

Problem 2: For a fixed number , find a subset S with such that the vertex boundary contains as few vertices as possible.

Cheeger constants are meant to answer exactly the questions above. For a subset S⊂V , we define

2.1

The Cheeger constant of a graph G is defined to be

2.2

In some sense, the problem of determining the Cheeger constant is equivalent to solving Problem 1, since

We remark that is connected if and only if . We will only consider connected graphs. In a similar manner, we define the analogue of (2.1) for “vertex expansion” (instead of “edge expansion”). For a subset , we define

gGS=vol δSmin vol S ,vol S 2.3

and

2.4

For regular graphs, we have

We define for a graph (not necessarily regular)

and

We remark that is the corresponding Cheeger constant when the measure for each vertex is taken to be 1. More general measures will be considered later in Section 2.5. We note that both and are concerned with the vertex expansion of a graph and are useful for many problems.

2.3 图的边缘展开(The edge expansion of a graph)

Lemma 2.1.

Theorem 2.2.

For a connected graph ,

Theorem 2.3.

For any connected graph G, we always have

Corollary 2.4.

In a graph G with eigenfunction f associated with , we define, for each v

And

Then

EXAMPLE 2.5. For a path , the Cheeger constant is . As shown in Example 1.4, the eigenvalue of is . this shown that the cheeger inequality in Theorem 2.2 is best possible up to within a constant factor.

EXAMPLE 2.6. For an -cube , the Cheeger constant is which is equal to (see Example 1.6) . Therefore the inequality in Lemma 2.1 is sharp to within a constant factor.

2.4 图的顶点展开(The vertex expansion of a graph)

Theorem 2.5.

For a connected graph ,

where denotes the maximum degree of .

Example 2.8. For an -cube, the vertex isoperimetric problem has been well studied. According to the Kruskal-Katona theorem , for a subset of vertices, for has at least vertices. Therefore, we have , for even.