×
Study for the day when you don't have to worry about price tags.
--Your friends at LectureNotes
Close

Graph and Network Theory

by Abhishek Apoorv
Type: NoteOffline Downloads: 193Views: 3831Uploaded: 8 months agoAdd to Favourite

Share it with your friends

Suggested Materials

Leave your Comments

Contributors

Abhishek Apoorv
Abhishek Apoorv
Graph and Network Theory Ernesto Estrada Department of Mathematics and Statistics University of Strathclyde, Glasgow Introduction ................................................................................................................................ 2 1 The language of graphs and networks .................................................................................... 3 1.1 Graph operators ................................................................................................................ 3 1.2 General graph concepts .................................................................................................... 5 1.3 Types of graphs ................................................................................................................ 6 2 Graphs in condensed matter physics ....................................................................................... 7 2.1 Tight-binding models ....................................................................................................... 7 2.1.1 Nullity and zero-energy states ................................................................................... 9 2.2 Hubbard model ............................................................................................................... 10 3 Graphs in statistical physics .................................................................................................. 12 4 Feynman graphs .................................................................................................................... 16 4.1 Symanzik polynomials and spanning trees .................................................................... 17 4.2 Symanzik polynomials and the Laplacian matrix .......................................................... 20 4.3 Symanzik polynomials and edge deletion/contraction ................................................... 21 5 Graphs and electrical networks ............................................................................................. 21 6 Graphs and vibrations ........................................................................................................... 23 6.1 Graph vibrational Hamiltonians ..................................................................................... 24 6.2 Network of Classical Oscillators .................................................................................... 24 6.3 Network of Quantum Oscillators ................................................................................... 26 7 Random graphs ..................................................................................................................... 28 8 Introducing complex networks ............................................................................................. 30 9 Small-World networks .......................................................................................................... 32 10 Degree distributions ............................................................................................................ 34 10.1 ‘Scale-free’ networks................................................................................................... 36 11 Network motifs ................................................................................................................... 37 12 Centrality measures ............................................................................................................. 38 1
13 Statistical mechanics of networks ....................................................................................... 41 13.1 Communicability in networks ...................................................................................... 42 14 Communities in networks ................................................................................................... 43 15 Dynamical processes on networks ...................................................................................... 45 15.1 Consensus ..................................................................................................................... 45 15.2 Synchronization in networks ........................................................................................ 47 15.3 Epidemics on networks ................................................................................................ 48 Glossary ................................................................................................................................... 50 List of works cited ................................................................................................................... 51 Further reading ......................................................................................................................... 53 Introduction Graph Theory was born in 1736 when Leonhard Euler published “Solutio problematic as geometriam situs pertinentis” (The solution of a problem relating to the theory of position) (Euler, 1736). This history is well documented (Biggs et al., 1976) and widely available in any textbook of graph or network theory. However, the word graph appeared for the first time in the context of natural sciences in 1878, when the English mathematician James J. Sylvester wrote a paper entitled “Chemistry and Algebra” which was published in Nature (Sylvester, 1877-78), where he wrote that “Every invariant and covariant thus becomes expressible by a graph precisely identical with a Kekulean diagram or chemicograph”. The use of graph theory in condensed matter physics, pioneered by many chemical and physical graph theorists (Harary, 1968; Trinajstić, 1992), is today well established; it has become even more popular after the recent discovery of graphene. There are few, if any, areas of physics in the XXIst century in which graphs and network are not involved directly or indirectly. Hence it is impossible to cover all of them in this Chapter. Thus I owe the reader an apology for the incompleteness of this Chapter and a promise to write a more complete treatise. For instance, quantum graphs are not considered in this Chapter and the reader is referred to a recent introductory monograph on this topic for details (Berkolaiko, Kuchment, 2013). In this chapter we will cover some of the most important areas of applications of graph theory in physics. These include condensed matter physics, statistical physics, quantum electrodynamics, electrical networks and vibrational problems. In the second part we summarise some of the most important aspects of the study of complex networks. This is an interdisciplinary area which has emerged with tremendous impetus in the XXIst century which studies networks appearing in complex systems. These systems range from molecular and biological to ecological, social and technological systems. Thus graph theory and network theory have helped to broaden the horizons of physics to embrace the study of new complex systems. We hope this chapter motivates the reader to find more about the connections between graph/network theory and physics, consolidating this discipline as an important part of the curriculum for the physicists of the XXIst century. 2
1 The language of graphs and networks The first thing that needs to be clarified is that the terms graphs and networks are used indistingtly in the literature. In this Chapter we will reserve the term graph for the abstract mathematical concept, in general referred to small, artificial formations of nodes and edges. The term network is then reserved for the graphs representing real-world objects in which the nodes represent entities of the system and the edges represent the relationships among them. Therefore, it is clear that we will refer to the system of individuals and their interactions as a ‘social network’ and not as a ‘social graph’. However, they should mean exactly the same. For the basic concepts of graph theory the reader is recommended to consult the introductory book by Harary (1967). We start by defining a graph formally. Let us consider a finite set V  v1 , v2 ,, vn  of unspecified elements and let V V be the set of all ordered pairs vi , v j  of the elements of V . A relation on the set V is any subset E  V  V . The relation E is symmetric if vi , v j   E implies v j , vi   E and it is reflexive if v V , v, v E . The relation E is antireflexive if vi , v j   E implies vi  v j . Now we can define a simple graph as the pair G  V , E  , where V is a finite set of nodes, vertices or points and E is a symmetric and antireflexive relation on V , whose elements are known as the edges or links of the graph. In a directed graph the relation E is non-symmetric. In many physical applications the edges of the graphs are required to support weights, i.e., real numbers indicating a specific property of the edge. In this case the following more general definition is convenient. A weighted graph is the quadruple G  V , E,W , f  where V is a finite set of nodes, E  V  V  e1, e2 ,, em  is a set of edges, W  w1 , w2 , , wr  is a set of weights such that wi  and f : E  W is a surjective mapping that assigns a weight to each edge. If the weights are natural numbers then the resulting graph is a multigraph in which there could be multiple edges between pairs of vertices. That is, if the weight between nodes p and q is k  N it means that there are k links between the two nodes. In an undirected graph we say that wo nodes p and q are adjacent if they are joined by an edge e   p, q . In this case we say that the nodes p and q are incident to the link e , and the link e is incident to the nodes p and q . The two nodes are called the end nodes of the edge. Two edges e1   p, q and e2  r , s are adjacent if they are both incident to at least one node. A simple but important characteristic of a node is its degree, which is defined as the number of edges which are incident to it or similarly the number of nodes adjacent to it. Slightly different definitions apply to directed graphs. The node p is adjacent to node q if there is a directed link from p to q , e   p, q  . We also say that a link from p to q is incident from p and incident to q ; p is incident to e and q is incident from e . Consequently, we have two different kinds of degrees in directed graphs. The in-degree of a node is the number of links incident to it and its out-degree is the number of links incident from it. 1.1 Graph operators The incidence and adjacency relations in graphs allow us to define the following graph operators. We consider an undirected graph for which we construct its incidence matrix with 3
an arbitrary orientation of its entries. This is necessary to consider that the incidence matrix is a discrete analogous of the gradient. That is, for every edge  p, q , p is the positive (head) and q the negative (tail) end of the oriented link. Let the links of the graph be labeled as e1 , e2 ,, em . Hence the oriented incidence matrix   G  :  1 node vi is the head of link e j   ij G    1 node vi is the tail of link e j  0 otherwise  We remark that the results obtained below are independent of the orientation of the links but assume that once the links are oriented, this orientation is not changed. Let the vertex LV and edge LE spaces be the vector spaces of all real-valued functions defined on V and E , respectively. The incidence operator of the graph is then defined as   G  : LV  LE , (1.1) such that for an arbitrary function f : V  , G  f : E   is given by (1.2) G f e  f p f q , where p are the starting (head) and q the ending (tail) points of the oriented link e . Here we consider that f is a real or vector-valued function on the graph with f being  - measurable for certain measure  on the graph. On the other hand, let H be a Hilbert space with scalar product , and norm  . Let G  V , E  be a simple graph. The adjacency operator is an operator acting on the Hilbert space H : l 2 V  defined as (1.3)  Af  p  :  f  q , f  H , i V . u ,vE The adjacency operator of an undirected network is a self-adjoint operator, which is bounded on l 2 V  . We recall that l 2 is the Hilbert space of square summable sequences with inner product, and that an operator is self-adjoint if its matrix is equal to its own conjugate transpose, i.e., it is Hermitian. It is worth pointing out here that the adjacency operator of a directed network might not be self-adjoint. The matrix representation of this operator is the adjacency matrix A , which for a simple graph is defined as Aij  1 if i, j  E (1.4) 0 otherwise. A third operator which is related to the previous two and which plays a fundamental role in the applications of graph theory in physics is the Laplacian operator. This operator is defined by L G f    f , (1.5) and it is the graph version of the Laplacian operator 2 f 2 f 2 f . (1.6) f     2 2 2 x1 x2 xn The negative sign in (1.5) is used by convention. Then the Laplacian operator acting on the function f previously defined is given by      LG f u     f u  f v,   (1.7) u ,v E which in matrix form is given by 4

Lecture Notes