Graph Theory - Random Graphs

---

---

Random Graphs

A random graph is a graph made by connecting vertices (nodes) randomly. This means the connections between the nodes are determined randomly, based on certain rules or probabilities. Random graphs are used to study networks where the exact connections are not known beforehand, but follow some random pattern.

Some of the important characteristics of random graphs are −

• Probabilistic Connections: Edges between vertices are created based on a probability. For example, each pair of vertices might have an edge with probability "p".
• Variety of Models: There are different models for generating random graphs, such as the ErdsRnyi model, where each edge is included with a fixed probability, or the BarabsiAlbert model, which generates scale-free networks.
• Applications: Random graphs are used in network theory, biology, computer science, and social sciences to study properties of real-world networks, like robustness, connectivity, and the spread of information or diseases.

Generating Random Graphs

Random graphs are generated using various algorithms. The major challenge is to model randomness while preserving certain properties of real-world networks. Following are methods for generating random graphs −

Random Graph Generation with G(n, p)

The most basic way to generate a random graph is by creating an ErdsRnyi random graph (G(n, p)) where n is the number of vertices, and p is the probability of each edge being present.

Algorithm

Following are the basic steps to generate random graph by creating an ErdsRnyi random graph −

• Initialize a graph with n vertices and no edges.
• For each pair of vertices, generate a random number.
• If the number is less than p, add an edge between the two vertices.
• Repeat the process until all edges are evaluated.

Random Regular Graph Generation

To generate a random regular graph, you need to ensure that each vertex has the same degree. This is done by randomly pairing vertices and connecting them while ensuring that the degree constraint is satisfied.

Algorithm

Following are the basic steps to generate random regular graph −

• Start with n vertices.
• Assign a degree d to each vertex.
• Randomly pair the vertices and add edges.
• If any vertex exceeds its degree constraint, reassign the edges.

Types of Random Graphs

Random graphs can be categorized based on the random processes used to generate them. Following are the primary types of random graphs −

Erdoan-Rnyi Random Graphs

The Erdoan-Rnyi random graph, denoted by G(n, p), is one of the most well-known types of random graphs. It is a graph with n vertices, where each pair of vertices is connected by an edge with probability p, independently of the other edges.

This means that for any pair of vertices, there is a probability p that an edge exists between them, and a probability 1-p that no edge exists.

Following are the properties of G(n, p) Graphs −

• Degree Distribution: The degree of each vertex in G(n, p) follows a binomial distribution, and for large n, it can be approximated by a Poisson distribution.
• Connectivity: As the edge probability p increases, the graph becomes more connected. When p reaches a critical threshold, the graph almost surely becomes connected.
• Phase Transition: A significant property of G(n, p) graphs is the phase transition where the graph abruptly changes from being disconnected to being connected as p crosses a critical threshold.

Gilbert's Random Graphs

Gilbert's Random Graphs, also known as the ErdsRnyi model, is a type of random graph where each possible edge between a pair of vertices is included in the graph with a fixed probability. This model is named after mathematicians Paul Erds and Alfrd Rnyi, who introduced it, and Edgar Gilbert, who independently studied a similar model.

The major characteristics of Gilbert's random graphs are −

• Fixed Probability: Each edge is added independently with a fixed probability "p". This means for each pair of vertices, there is a chance "p" that an edge will connect them.
• Randomness: The structure of the graph is highly random, with no predetermined pattern for which edges will be included.

Random Regular Graphs

A random regular graph is a graph where each vertex has the same degree, and the edges are placed randomly, subject to the condition that all vertices must have the same degree.

The major properties of random regular graphs are −

• Degree Sequence: The degree sequence of a random regular graph is fixed for all vertices.
• Clustering Coefficient: Random regular graphs often have clustering coefficients that are different from those of random graphs.

Small-World Graphs

Small-world networks are random graphs where most vertices are not neighbors, but there are small paths connecting most pairs of vertices. These graphs are designed to mimic real-world networks such as social networks, the internet, and neural networks.

Following are the properties of small-world networks −

• Short Path Lengths: Despite a low average degree, small-world networks have relatively short paths between most vertices, making them efficient in terms of communication or spread of information.
• High Clustering Coefficient: Small-world graphs displays a high degree of clustering, meaning that vertices tend to form tightly connected groups.

Properties of Random Graphs

Random graphs have various properties that can be analyzed statistically. The most important properties are as follows −

Degree Distribution

In random graphs like G(n, p), the degree distribution is binomial for small n, but it approaches a Poisson distribution as n becomes large. The degree distribution of random regular graphs is uniform, as every vertex has the same degree.

Giant Component

In large random graphs, there is a giant connected component that contains a significant fraction of the vertices. The size of this giant component depends on the edge probability p or degree of vertices in the graph.

Clustering Coefficient

The clustering coefficient measures how connected the neighbors of a given vertex are. In many random graphs, the clustering coefficient is low, especially in G(n, p) graphs. In contrast, small-world networks have a high clustering coefficient.

Diameter

The diameter of a graph is the length of the longest shortest path between any two vertices. In large random graphs, the diameter tends to be small (often logarithmic in size), especially for ErdsRnyi graphs.

Average Path Length

The average path length is the average number of edges in the shortest path between any two vertices in the graph. Random graphs typically have small average path lengths, even when the graph is large.

Challenges in Random Graph Theory

Random graphs offer valuable understandings, but they also come with several challenges, they are −

Scaling Up

As the number of vertices grows, the complexity of analyzing random graphs increases. This requires new algorithms and techniques to handle large-scale random graph data.

Realistic Network Models

Random graphs are useful, but they don't always match real-world networks, which often have patterns like groups and clusters. To fix this, more advanced models like small-world and scale-free networks are used.

Algorithmic Complexity

Generating and analyzing random graphs, especially large ones, can use a lot of resources. Efficient algorithms are needed to handle large graphs in real situations.