Zero-One Laws for Random Graphs

Last time we saw a number of properties of graphs, such as connectivity, where the probability that an Erdős–Rényi random graph G(n,p) satisfies the property is asymptotically either zero or one. And this zero or one depends on whether the parameter p is above or below a universal threshold (that depends only on n and the property in question).

To remind the reader, the Erdős–Rényi random “graph” G(n,p) is a distribution over graphs that you draw from by including each edge independently with probability p. Last time we saw that the existence of an isolated vertex has a sharp threshold at (\log n) / n, meaning if p is asymptotically smaller than the threshold there will certainly be isolated vertices, and if p is larger there will certainly be no isolated vertices. We also gave a laundry list of other properties with such thresholds.

One might want to study this phenomenon in general. Even if we might not be able to find all the thresholds we want for a given property, can we classify which properties have thresholds and which do not?

The answer turns out to be mostly yes! For large classes of properties, there are proofs that say things like, “either this property holds with probability tending to one, or it holds with probability tending to zero.” These are called “zero-one laws,” and they’re sort of meta theorems. We’ll see one such theorem in this post relating to constant edge-probabilities in random graphs, and we’ll remark on another at the end.

Sentences about graphs in first order logic

A zero-one law generally works by defining a class of properties, and then applying a generic first/second moment-type argument to every property in the class.

So first we define what kinds of properties we’ll discuss. We’ll pick a large class: anything that can be expressed in first-order logic in the language of graphs. That is, any finite logical statement that uses existential and universal quantifiers over variables, and whose only relation (test) is whether an edge exists between two vertices. We’ll call this test e(x,y). So you write some sentence P in this language, and you take a graph G, and you can ask P(G) = 1, whether the graph satisfies the sentence.

This seems like a really large class of properties, and it is, but let’s think carefully about what kinds of properties can be expressed this way. Clearly the existence of a triangle can be written this way, it’s just the sentence

\exists x,y,z : e(x,y) \wedge e(y,z) \wedge e(x,z)

I’m using \wedge for AND, and \vee for OR, and \neg for NOT. Similarly, one can express the existence of a clique of size k, or the existence of an independent set of size k, or a path of a fixed length, or whether there is a vertex of maximal degree n-1.

Here’s a question: can we write a formula which will be true for a graph if and only if it’s connected? Well such a formula seems like it would have to know about how many vertices there are in the graph, so it could say something like “for all x,y there is a path from x to y.” It seems like you’d need a family of such formulas that grows with n to make anything work. But this isn’t a proof; the question remains whether there is some other tricky way to encode connectivity.

But as it turns out, connectivity is not a formula you can express in propositional logic. We won’t prove it here, but we will note at the end of the article that connectivity is in a different class of properties that you can prove has a similar zero-one law.

The zero-one law for first order logic

So the theorem about first-order expressible sentences is as follows.

Theorem: Let P be a property of graphs that can be expressed in the first order language of graphs (with the e(x,y) relation). Then for any constant p, the probability that P holds in G(n,p) has a limit of zero or one as n \to \infty.

Proof. We’ll prove the simpler case of p=1/2, but the general case is analogous. Given such a graph G drawn from G(n,p), what we’ll do is define a countably infinite family of propositional formulas \varphi_{k,l}, and argue that they form a sort of “basis” for all first-order sentences about graphs.

First let’s describe the \varphi_{k,l}. For any k,l \in \mathbb{N}, the sentence will assert that for every set of k vertices and every set of l vertices, there is some other vertex connected to the first k but not the last l.

\displaystyle \varphi_{k,l} : \forall x_1, \dots, x_k, y_1, \dots, y_l \exists z : \\ e(z,x_1) \wedge \dots \wedge e(z,x_k) \wedge \neg e(z,y_1) \wedge \dots \wedge \neg e(z,y_l).

In other words, these formulas encapsulate every possible incidence pattern for a single vertex. It is a strange set of formulas, but they have a very nice property we’re about to get to. So for a fixed \varphi_{k,l}, what is the probability that it’s false on n vertices? We want to give an upper bound and hence show that the formula is true with probability approaching 1. That is, we want to show that all the \varphi_{k,l} are true with probability tending to 1.

Computing the probability: we have \binom{n}{k} \binom{n-k}{l} possibilities to choose these sets, and the probability that some other fixed vertex z has the good connections is 2^{-(k+l)} so the probability z is not good is 1 - 2^{-(k+l)}, and taking a product over all choices of z gives the probability that there is some bad vertex z with an exponent of (n - (k + l)). Combining all this together gives an upper bound of \varphi_{k,l} being false of:

\displaystyle \binom{n}{k}\binom{n-k}{l} (1-2^{-k-1})^{n-k-l}

And k, l are constant, so the left two terms are polynomials while the rightmost term is an exponentially small function, and this implies that the whole expression tends to zero, as desired.

Break from proof.

A bit of model theory

So what we’ve proved so far is that the probability of every formula of the form \varphi_{k,l} being satisfied in G(n,1/2) tends to 1.

Now look at the set of all such formulas

\displaystyle \Phi = \{ \varphi_{k,l} : k,l \in \mathbb{N} \}

We ask: is there any graph which satisfies all of these formulas? Certainly it cannot be finite, because a finite graph would not be able to satisfy formulas with sufficiently large values of l, k > n. But indeed, there is a countably infinite graph that works. It’s called the Rado graph, pictured below.

rado

The Rado graph has some really interesting properties, such as that it contains every finite and countably infinite graph as induced subgraphs. Basically this means, as far as countably infinite graphs go, it’s the big momma of all graphs. It’s the graph in a very concrete sense of the word. It satisfies all of the formulas in \Phi, and in fact it’s uniquely determined by this, meaning that if any other countably infinite graph satisfies all the formulas in \Phi, then that graph is isomorphic to the Rado graph.

But for our purposes (proving a zero-one law), there’s a better perspective than graph theory on this object. In the logic perspective, the set \Phi is called a theory, meaning a set of statements that you consider “axioms” in some logical system. And we’re asking whether there any model realizing the theory. That is, is there some logical system with a semantic interpretation (some mathematical object based on numbers, or sets, or whatever) that satisfies all the axioms?

A good analogy comes from the rational numbers, because they satisfy a similar property among all ordered sets. In fact, the rational numbers are the unique countable, ordered set with the property that it has no biggest/smallest element and is dense. That is, in the ordering there is always another element between any two elements you want. So the theorem says if you have two countable sets with these properties, then they are actually isomorphic as ordered sets, and they are isomorphic to the rational numbers.

So, while we won’t prove that the Rado graph is a model for our theory \Phi, we will use that fact to great benefit. One consequence of having a theory with a model is that the theory is consistent, meaning it can’t imply any contradictions. Another fact is that this theory \Phi is complete. Completeness means that any formula or it’s negation is logically implied by the theory. Note these are syntactical implications (using standard rules of propositional logic), and have nothing to do with the model interpreting the theory.

The proof that \Phi is complete actually follows from the uniqueness of the Rado graph as the only countable model of \Phi. Suppose the contrary, that \Phi is not consistent, then there has to be some formula \psi that is not provable, and it’s negation is also not provable, by starting from \Phi. Now extend \Phi in two ways: by adding \psi and by adding \neg \psi. Both of the new theories are still countable, and by a theorem from logic this means they both still have countable models. But both of these new models are also countable models of \Phi, so they have to both be the Rado graph. But this is very embarrassing for them, because we assumed they disagree on the truth of \psi.

So now we can go ahead and prove the zero-one law theorem.

Return to proof.

Given an arbitrary property \varphi \not \in \Psi. Now either \varphi or it’s negation can be derived from \Phi. Without loss of generality suppose it’s \varphi. Take all the formulas from the theory you need to derive \varphi, and note that since it is a proof in propositional logic you will only finitely many such \varphi_{k,l}. Now look at the probabilities of the \varphi_{k,l}: they are all true with probability tending to 1, so the implied statement of the proof of \varphi (i.e., \varphi itself) must also hold with probability tending to 1. And we’re done!

\square

If you don’t like model theory, there is another “purely combinatorial” proof of the zero-one law using something called Ehrenfeucht–Fraïssé games. It is a bit longer, though.

Other zero-one laws

One might naturally ask two questions: what if your probability is not constant, and what other kinds of properties have zero-one laws? Both great questions.

For the first, there are some extra theorems. I’ll just describe one that has always seemed very strange to me. If your probability is of the form p = n^{-\alpha} but \alpha is irrational, then the zero-one law still holds! This is a theorem of Baldwin-Shelah-Spencer, and it really makes you wonder why irrational numbers would be so well behaved while rational numbers are not 🙂

For the second question, there is another theorem about monotone properties of graphs. Monotone properties come in two flavors, so called “increasing” and “decreasing.” I’ll describe increasing monotone properties and the decreasing counterpart should be obvious. A property is called monotone increasing if adding edges can never destroy the property. That is, with an empty graph you don’t have the property (or maybe you do), and as you start adding edges eventually you suddenly get the property, but then adding more edges can’t cause you to lose the property again. Good examples of this include connectivity, or the existence of a triangle.

So the theorem is that there is an identical zero-one law for monotone properties. Great!

It’s not so often that you get to see these neat applications of logic and model theory to graph theory and (by extension) computer science. But when you do get to apply them they seem very powerful and mysterious. I think it’s a good thing.

Until next time!

Community Detection in Graphs — a Casual Tour

Graphs are among the most interesting and useful objects in mathematics. Any situation or idea that can be described by objects with connections is a graph, and one of the most prominent examples of a real-world graph that one can come up with is a social network.

Recall, if you aren’t already familiar with this blog’s gentle introduction to graphs, that a graph G is defined by a set of vertices V, and a set of edges E, each of which connects two vertices. For this post the edges will be undirected, meaning connections between vertices are symmetric.

One of the most common topics to talk about for graphs is the notion of a community. But what does one actually mean by that word? It’s easy to give an informal definition: a subset of vertices C such that there are many more edges between vertices in C than from vertices in C to vertices in V - C (the complement of C). Try to make this notion precise, however, and you open a door to a world of difficult problems and open research questions. Indeed, nobody has yet come to a conclusive and useful definition of what it means to be a community. In this post we’ll see why this is such a hard problem, and we’ll see that it mostly has to do with the word “useful.” In future posts we plan to cover some techniques that have found widespread success in practice, but this post is intended to impress upon the reader how difficult the problem is.

The simplest idea

The simplest thing to do is to say a community is a subset of vertices which are completely connected to each other. In the technical parlance, a community is a subgraph which forms a clique. Sometimes an n-clique is also called a complete graph on n vertices, denoted K_n. Here’s an example of a 5-clique in a larger graph:

"Where's Waldo" for graph theorists: a clique hidden in a larger graph.

“Where’s Waldo” for graph theorists: a clique hidden in a larger graph.

Indeed, it seems reasonable that if we can reliably find communities at all, then we should be able to find cliques. But as fate should have it, this problem is known to be computationally intractable. In more detail, the problem of finding the largest clique in a graph is NP-hard. That essentially means we don’t have any better algorithms to find cliques in general graphs than to try all possible subsets of the vertices and check to see which, if any, form cliques. In fact it’s much worse, this problem is known to be hard to approximate to any reasonable factor in the worst case (the error of the approximation grows polynomially with the size of the graph!). So we can’t even hope to find a clique half the size of the biggest, or a thousandth the size!

But we have to take these impossibility results with a grain of salt: they only say things about the worst case graphs. And when we’re looking for communities in the real world, the worst case will never show up. Really, it won’t! In these proofs, “worst case” means that they encode some arbitrarily convoluted logic problem into a graph, so that finding the clique means solving the logic problem. To think that someone could engineer their social network to encode difficult logic problems is ridiculous.

So what about an “average case” graph? To formulate this typically means we need to consider graphs randomly drawn from a distribution.

Random graphs

The simplest kind of “randomized” graph you could have is the following. You fix some set of vertices, and then run an experiment: for each pair of vertices you flip a coin, and if the coin is heads you place an edge and otherwise you don’t. This defines a distribution on graphs called G(n, 1/2), which we can generalize to G(n, p) for a coin with bias p. With a slight abuse of notation, we call G(n, p) the Erdős–Rényi random graph (it’s not a graph but a distribution on graphs). We explored this topic form a more mathematical perspective earlier on this blog.

So we can sample from this distribution and ask questions like: what’s the probability of the largest clique being size at least 20? Indeed, cliques in Erdős–Rényi random graphs are so well understood that we know exactly how they work. For example, if p=1/2 then the size of the largest clique is guaranteed (with overwhelming probability as n grows) to have size k(n) or k(n)+1, where k(n) is about 2 \log n. Just as much is known about other values of p as well as other properties of G(n,p), see Wikipedia for a short list.

In other words, if we wanted to find the largest clique in an Erdős–Rényi random graph, we could check all subsets of size roughly 2\log(n), which would take about (n / \log(n))^{\log(n)} time. This is pretty terrible, and I’ve never heard of an algorithm that does better (contrary to the original statement in this paragraph that showed I can’t count). In any case, it turns out that the Erdős–Rényi random graph, and using cliques to represent communities, is far from realistic. There are many reasons why this is the case, but here’s one example that fits with the topic at hand. If I thought the world’s social network was distributed according to G(n, 1/2) and communities were cliques, then I would be claiming that the largest community is of size 65 or 66. Estimated world population: 7 billion, 2 \log(7 \cdot 10^9) \sim 65. Clearly this is ridiculous: there are groups of larger than 66 people that we would want to call “communities,” and there are plenty of communities that don’t form bona-fide cliques.

Another avenue shows that things are still not as easy as they seem in Erdős–Rényi land. This is the so-called planted clique problem. That is, you draw a graph G from G(n, 1/2). You give G to me and I pick a random but secret subset of r vertices and I add enough edges to make those vertices form an r-clique. Then I ask you to find the r-clique. Clearly it doesn’t make sense when r < 2 \log (n) because you won’t be able to tell it apart from the guaranteed cliques in G. But even worse, nobody knows how to find the planted clique when r is even a little bit smaller than \sqrt{n} (like, r = n^{9/20} even). Just to solidify this with some numbers, we don’t know how to reliably find a planted clique of size 60 in a random graph on ten thousand vertices, but we do when the size of the clique goes up to 100. The best algorithms we know rely on some sophisticated tools in spectral graph theory, and their details are beyond the scope of this post.

So Erdős–Rényi graphs seem to have no hope. What’s next? There are a couple of routes we can take from here. We can try to change our random graph model to be more realistic. We can relax our notion of communities from cliques to something else. We can do both, or we can do something completely different.

Other kinds of random graphs

There is an interesting model of Barabási and Albert, often called the “preferential attachment” model, that has been described as a good model of large, quickly growing networks like the internet. Here’s the idea: you start off with a two-clique G = K_2, and at each time step t you add a new vertex v to G, and new edges so that the probability that the edge (v,w) is added to G is proportional to the degree of w (as a fraction of the total number of edges in G). Here’s an animation of this process:

Image source: Wikipedia

The significance of this random model is that it creates graphs with a small number of hubs, and a large number of low-degree vertices. In other words, the preferential attachment model tends to “make the rich richer.” Another perspective is that the degree distribution of such a graph is guaranteed to fit a so-called power-law distribution. Informally, this means that the overall fraction of small-degree vertices gives a significant contribution to the total number of edges. This is sometimes called a “fat-tailed” distribution. Since power-law distributions are observed in a wide variety of natural settings, some have used this as justification for working in the preferential attachment setting. On the other hand, this model is known to have no significant community structure (by any reasonable definition, certainly not having cliques of nontrivial size), and this has been used as evidence against the model. I am not aware of any work done on planting dense subgraphs in graphs drawn from a preferential attachment model, but I think it’s likely to be trivial and uninteresting. On the other hand, Bubeck et al. have looked at changing the initial graph (the “seed”) from a 2-clique to something else, and seeing how that affects the overall limiting distribution.

Another model that often shows up is a model that allows one to make a random graph starting with any fixed degree distribution, not just a power law. There are a number of models that do this to some fashion, and you’ll hear a lot of hyphenated names thrown around like Chung-Lu and Molloy-Reed and Newman-Strogatz-Watts. The one we’ll describe is quite simple. Say you start with a set of vertices V, and a number d_v for each vertex v, such that the sum of all the d_v is even. This condition is required because in any graph the sum of the degrees of a vertex is twice the number of edges. Then you imagine each vertex v having d_v “edge-stubs.” The name suggests a picture like the one below:

Each node has a prescribed number of "edge stubs," which are randomly connected to form a graph.

Each node has a prescribed number of “edge stubs,” which are randomly connected to form a graph.

Now you pick two edge stubs at random and connect them. One usually allows self-loops and multiple edges between vertices, so that it’s okay to pick two edge stubs from the same vertex. You keep doing this until all the edge stubs are accounted for, and this is your random graph. The degrees were fixed at the beginning, so the only randomization is in which vertices are adjacent. The same obvious biases apply, that any given vertex is more likely to be adjacent to high-degree vertices, but now we get to control the biases with much more precision.

The reason such a model is useful is that when you’re working with graphs in the real world, you usually have statistical information available. It’s simple to compute the degree of each vertex, and so you can use this random graph as a sort of “prior” distribution and look for anomalies. In particular, this is precisely how one of the leading measures of community structure works: the measure of modularity. We’ll talk about this in the next section.

Other kinds of communities

Here’s one easy way to relax our notion of communities. Rather than finding complete subgraphs, we could ask about finding very dense subgraphs (ignoring what happens outside the subgraph). We compute density as the average degree of vertices in the subgraph.

If we impose no bound on the size of the subgraph an algorithm is allowed to output, then there is an efficient algorithm for finding the densest subgraph in a given graph. The general exact solution involves solving a linear programming problem and a little extra work, but luckily there is a greedy algorithm that can get within half of the optimal density. You start with all the vertices S_n = V, and remove any vertex of minimal degree to get S_{n-1}. Continue until S_0, and then compute the density of all the S_i. The best one is guaranteed to be at least half of the optimal density. See this paper of Moses Charikar for a more formal analysis.

One problem with this is that the size of the densest subgraph might be too big. Unfortunately, if you fix the size of the dense subgraph you’re looking for (say, you want to find the densest subgraph of size at most k where k is an input), then the problem once again becomes NP-hard and suffers from the same sort of inapproximability theorems as finding the largest clique.

A more important issue with this is that a dense subgraph isn’t necessarily a community. In particular, we want communities to be dense on the inside and sparse on the outside. The densest subgraph analysis, however, might rate the following graph as one big dense subgraph instead of two separately dense communities with some modest (but not too modest) amount of connections between them.

What should be identified as communities?

What are the correct communities here?

Indeed, we want a quantifiable a notion of “dense on the inside and sparse on the outside.” One such formalization is called modularity. Modularity works as follows. If you give me some partition of the vertices of G into two sets, modularity measures how well this partition reflects two separate communities. It’s the definition of “community” here that makes it interesting. Rather than ask about densities exactly, you can compare the densities to the expected densities in a given random graph model.

In particular, we can use the fixed-degree distribution model from the last section. If we know the degrees of all the vertices ahead of time, we can compute the probability that we see some number of edges going between the two pieces of the partition relative to what we would see at random. If the difference is large (and largely biased toward fewer edges across the partition and more edges within the two subsets), then we say it has high modularity. This involves a lot of computations  — the whole measure can be written as a quadratic form via one big matrix — but the idea is simple enough. We intend to write more about modularity and implement the algorithm on this blog, but the excited reader can see the original paper of M.E.J. Newman.

Now modularity is very popular but it too has shortcomings. First, even though you can compute the modularity of a given partition, there’s still the problem of finding the partition that globally maximizes modularity. Sadly, this is known to be NP-hard. Mover, it’s known to be NP-hard even if you’re just trying to find a partition into two pieces that maximizes modularity, and even still when the graph is regular (every vertex has the same degree).

Still worse, while there are some readily accepted heuristics that often “do well enough” in practice, we don’t even know how to approximate modularity very well. Bhaskar DasGupta has a line of work studying approximations of maximum modularity, and he has proved that for dense graphs you can’t even approximate modularity to within any constant factor. That is, the best you can do is have an approximation that gets worse as the size of the graph grows. It’s similar to the bad news we had for finding the largest clique, but not as bad. For example, when the graph is sparse it’s known that one can approximate modularity to within a \log(n) factor of the optimum, where n is the number of vertices of the graph (for cliques the factor was like n^c for some c, and this is drastically worse).

Another empirical issue is that modularity seems to fail to find small communities. That is, if your graph has some large communities and some small communities, strictly maximizing the modularity is not the right thing to do. So we’ve seen that even the leading method in the field has some issues.

Something completely different

The last method I want to sketch is in the realm of “something completely different.” The notion is that if we’re given a graph, we can run some experiment on the graph, and the results of that experiment can give us insight into where the communities are.

The experiment I’m going to talk about is the random walk. That is, say you have a vertex v in a graph G and you want to find some vertices that are “closest” to v. That is, those that are most likely to be in the same community as v. What you can do is run a random walk starting at v. By a “random walk” I mean you start at v, you pick a neighbor at random and move to it, then repeat. You can compute statistics about the vertices you visit in a sample of such walks, and the vertices that you visit most often are those you say are “in the same community as v. One important parameter is how long the walk is, but it’s generally believed to be best if you keep it between 3-6 steps.

Of course, this is not a partition of the vertices, so it’s not a community detection algorithm, but you can turn it into one. Run this process for each vertex, and use it to compute a “distance” between all the pairs of vertices. Then you compute a tree of partitions by lumping the closest pairs of vertices into the same community, one at a time, until you’ve got every vertex. At each step of the way, you compute the modularity of the partition, and when you’re done you choose the partition that maximizes modularity. This algorithm as a whole is called the walktrap clustering algorithm, and was introduced by Pons and Latapy in 2005.

This sounds like a really great idea, because it’s intuitive: there’s a relatively high chance that the friends of your friends are also your friends. It’s also really great because there is an easily measurable tradeoff between runtime and quality: you can tune down the length of the random walk, and the number of samples you take for each vertex, to speed up the runtime but lower the quality of your statistical estimates. So if you’re working on huge graphs, you get a lot of control and a clear idea of exactly what’s going on inside the algorithm (something which is not immediately clear in a lot of these papers).

Unfortunately, I’m not aware of any concrete theoretical guarantees for walktrap clustering. The one bit of theoretical justification I’ve read over the last year is that you can relate the expected distances you get to certain spectral properties of the graph that are known to be related to community structure, but the lower bounds on maximizing modularity already suggest (though they do not imply) that walktrap won’t do that well in the worst case.

So many algorithms, so little time!

I have only brushed the surface of the literature on community detection, and the things I have discussed are heavily biased toward what I’ve read about and used in my own research. There are methods based on information theory, label propagation, and obscure physics processes like “spin glass” (whatever that is, it sounds frustrating).

And we have only been talking about perfect community structure. What if you want to allow people to be in multiple communities, or have communities at varying levels of granularity (e.g. a sports club within a school versus the whole student body of that school)? What if we want to allow people to be “members” of a community at varying degrees of intensity? How do we deal with noisy signals in our graphs? For example, if we get our data from observing people talk, are two people who have heated arguments considered to be in the same community? Since a lot social network data comes from sources like Twitter and Facebook where arguments are rampant, how do we distinguish between useful and useless data? More subtly, how do we determine useful information if a group within the social network are trying to mask their discovery? That is, how do we deal with adversarial noise in a graph?

And all of this is just on static graphs! What about graphs that change over time? You can keep making the problem more and more complicated as it gets more realistic.

With the huge wealth of research that has already been done just on the simplest case, and the difficult problems and known barriers to success even for the simple problems, it seems almost intimidating to even begin to try to answer these questions. But maybe that’s what makes them fascinating, not to mention that governments and big businesses pour many millions of dollars into this kind of research.

In the future of this blog we plan to derive and implement some of the basic methods of community detection. This includes, as a first outline, the modularity measure and the walktrap clustering algorithm. Considering that I’m also going to spend a large part of the summer thinking about these problems (indeed, with some of the leading researchers and upcoming stars under the sponsorship of the American Mathematical Society), it’s unlikely to end there.

Until next time!

The Erdős-Rényi Random Graph

During the 1950’s the famous mathematician Paul Erdős and Alfred Rényi put forth the concept of a random graph and in the subsequent years of study transformed the world of combinatorics. The random graph is the perfect example of a good mathematical definition: it’s simple, has surprisingly intricate structure, and yields many applications.

In this post we’ll explore basic facts about random graphs, slowly detail a proof on their applications to graph theory, and explore their more interesting properties computationally (a prelude to proofs about their structure). We assume the reader is familiar with the definition of a graph, which we’ve written about at length for non-mathematical audiences, and has some familiarity with undergraduate-level probability and combinatorics for the more mathematical sections of the post. We’ll do our best to remind the reader of these prerequisites as we go, and we welcome any clarification questions in the comment section.

The Erdős-Rényi Model

The definition of a random graph is simple enough that we need not defer it to the technical section of the article.

Definition: Given a positive integer n and a probability value 0 \leq p \leq 1, define the graph G(n,p) to be the undirected graph on n vertices whose edges are chosen as follows. For all pairs of vertices v,w there is an edge (v,w) with probability p.

Of course, there is no single random graph. What we’ve described here is a process for constructing a graph. We create a set of n vertices, and for each possible pair of vertices we flip a coin (often a biased coin) to determine if we should add an edge connecting them. Indeed, every graph can be made by this process if one is sufficiently lucky (or unlucky), but it’s very unlikely that we will have no edges at all if p is large enough. So G(n,p) is really a probability distribution over the set of all possible graphs on n vertices. We commit a horrendous abuse of notation by saying G or G(n,p) is a random graph instead of saying that G is sampled from the distribution. The reader will get used to it in time.

Why Do We Care?

Random graphs of all sorts (not just Erdős’s model) find applications in two very different worlds. The first is pure combinatorics, and the second is in the analysis of networks of all kinds.

In combinatorics we often wonder if graphs exist with certain properties. For instance, in graph theory we have the notion of graph colorability: can we color the vertices of a graph with k colors so that none of its edges are monochromatic? (See this blog’s primer on graph coloring for more) Indeed, coloring is known to be a very difficult problem on general graphs. The problem of determining whether a graph can be colored with a fixed number of colors has no known efficient algorithm; it is NP-complete. Even worse, much of our intuition about graphs fails for graph coloring. We would expect that sparse-looking graphs can be colored with fewer colors than dense graphs. One naive way to measure sparsity of a graph is to measure the length of its shortest cycle (recall that a cycle is a path which starts and ends at the same vertex). This measurement is called the girth of a graph. But Paul Erdős proved using random graphs, as we will momentarily, that for any choice of integers g,k there are graphs of girth \geq g which cannot be colored with fewer than k colors. Preventing short cycles in graphs doesn’t make coloring easier.

The role that random graphs play in this picture is to give us ways to ensure the existence of graphs with certain properties, even if we don’t know how to construct an example of such a graph. Indeed, for every theorem proved using random graphs, there is a theorem (or open problem) concerning how to algorithmically construct those graphs which are known to exist.

Pure combinatorics may not seem very useful to the real world, but models of random graphs (even those beyond the Erdős-Rényi model) are quite relevant. Here is a simple example. One can take a Facebook user u and form a graph of that users network of immediate friends N(u) (excluding u itself), where vertices are people and two people are connected by an edge if they are mutual friends; call this the user’s friendship neighborhood. It turns out that the characteristics of the average Facebook user’s friendship neighborhood resembles a random graph. So understanding random graphs helps us understand the structure of small networks of friends. If we’re particularly insightful, we can do quite useful things like identify anomalies, such as duplicitous accounts, which deviate quite far from the expected model. They can also help discover trends or identify characteristics that can allow for more accurate ad targeting. For more details on how such an idea is translated into mathematics and code, see Modularity (we plan to talk about modularity on this blog in the near future; lots of great linear algebra there!).

Random graphs, when they exhibit observed phenomena, have important philosophical consequences. From a bird’s-eye view, there are two camps of scientists. The first are those who care about leveraging empirically observed phenomena to solve problems. Many statisticians fit into this realm: they do wonderful magic with data fit to certain distributions, but they often don’t know and don’t care whether the data they use truly has their assumed properties. The other camp is those who want to discover generative models for the data with theoretical principles. This is more like theoretical physics, where we invent an arguably computational notion of gravity whose consequences explain our observations.

For applied purposes, the Erdős-Rényi random graph model is in the second camp. In particular, if something fits in the Erdős-Rényi model, then it’s both highly structured (as we will make clear in the sequel) and “essentially random.” Bringing back the example of Facebook, this says that most people in the average user’s immediate friendship neighborhood are essentially the same and essentially random in their friendships among the friends of u. This is not quite correct, but it’s close enough to motivate our investigation of random graph models. Indeed, even Paul Erdős in his landmark paper mentioned that equiprobability among all vertices in a graph is unrealistic. See this survey for a more thorough (and advanced!) overview, and we promise to cover models which better represent social phenomena in the future.

So lets go ahead and look at a technical proof using random graphs from combinatorics, and then write some programs to generate random graphs.

Girth and Chromatic Number, and Counting Triangles

As a fair warning, this proof has a lot of moving parts. Skip to the next section if you’re eager to see some programs.

Say we have a k and a g, and we wonder whether a graph can exist which simultaneously has no cycles of length less than g (the girth) and needs at least k colors to color. The following theorem settles this question affirmatively.  A bit of terminology: the chromatic number of a graph G, denoted \chi(G), is the smallest number of colors needed to properly color G.

Theorem: For any natural numbers k,g, there exist graphs of chromatic number at least k and girth at least g.

Proof. Taking our cue from random graphs, let’s see what the probability is that a random graph G(n,p) on n vertices will have our desired properties. Or easier, what’s the chance that it will not have the right properties? This is essentially a fancy counting argument, but it’s nicer if we phrase it in the language of probability theory.

The proof has a few twists and turns for those uninitiated to the probabilistic method of proof. First, we will look at an arbitrary G(n,p) (where n,p are variable) and ask two questions about it: what is the expected number of short cycles, and what is the expected “independence number” (which we will see is related to coloring). We’ll then pick a value of p, depending crucially on n, which makes both of these expectations small. Next, we’ll use the fact that if the probability that G(n,p) doesn’t have our properties is strictly less than 1, then there has to be some instance in our probability space which has those properties (if no instance had the property, then the probability would be one!). Though we will not know what the graphs look like, their existence is enough to prove the theorem.

So let’s start with cycles. If we’re given a desired girth of g, the expected number of cycles of length \leq g in G(n,p) can be bounded by (np)^{g+1}/(np-1). To see this, the two main points are how to count the number of ways to pick k vertices in order to form a cycle, and a typical fact about sums of powers. Indeed, we can think of a cycle of length k as a way to seat a choice of k people around a circular table. There are \binom{n}{k} possible groups of people, and (k-1)! ways to seat one group. If we fix k and let n grow, then the product \binom{n}{k}(k-1)! will eventually be smaller than n^k (actually, this happens almost immediately in almost all cases). For each such choice of an ordering, the probability of the needed edges occurring to form a cycle is p^j, since all edges must occur independently of each other with probability p.

So the probability that we get a cycle of j vertices is

\displaystyle \binom{n}{j}(j-1)!p^j

And by the reasoning above we can bound this by n^jp^j. Summing over all numbers j = 3, \dots, g (we are secretly using the union bound), we bound the expected number of cycles of length \leq g from above:

\displaystyle \sum_{j=3}^g n^j p^j < \sum_{j=0}^g n^j p^j = \frac{(np)^{g+1}}{np - 1}

Since we want relatively few cycles to occur, we want it to be the case that the last quantity, (np)^{j+1}/(np-1), goes to zero as n goes to infinity. One trick is to pick p depending on n. If p = n^l, our upper bound becomes n^{(l+1)(g+1)} / (n^{1+l} - 1), and if we want the quantity to tend to zero it must be that (l+1)(g+1) < 1. Solving this we get that -1 < l < \frac{1}{g+1} - 1 < 0. Pick such an l (it doesn’t matter which), and keep this in mind: for our choice of p, the expected number of cycles goes to zero as n tends to infinity.

On the other hand, we want to make sure that such a graph has high chromatic number. To do this we’ll look at a related property: the size of the largest independent set. An independent set of a graph G = (V,E) is a set of vertices S \subset V so that there are no edges in E between vertices of S. We call \alpha(G) the size of the largest independent set. The values \alpha(G) and \chi(G) are related, because any time you have an independent set you can color all the vertices with a single color. In particular, this proves the inequality \chi(G) \alpha(G) \geq n, the number of vertices of G, or equivalently \chi(G) \geq n / \alpha(G). So if we want to ensure \chi(G) is large, it suffices to show \alpha(G) is small (rigorously, \alpha(G) \leq n / k implies \chi(G) \geq k).

The expected number of independent sets (again using the union bound) is at most the product of the number of possible independent sets and the probability of one of these having no edges. We let r be arbitrary and look at independent sets of size r Since there are \binom{n}{r} sets and each has a probability (1-p)^{\binom{r}{2}} of being independent, we get the probability that there is an independent set of size r is bounded by

\displaystyle \textup{P}(\alpha(G) \geq r) \leq \binom{n}{r}(1-p)^{\binom{r}{2}}

We use the fact that 1-x < e^{-x} for all x to translate the (1-p) part. Combining this with the usual \binom{n}{r} \leq n^r we get the probability of having an independent set of size r at most

\displaystyle \textup{P}(\alpha(G) \geq r) \leq \displaystyle n^re^{-pr(r-1)/2}

Now again we want to pick r so that this quantity goes to zero asymptotically, and it’s not hard to see that r = \frac{3}{p}\log(n) is good enough. With a little arithmetic we get the probability is at most n^{(1-a)/2}, where a > 1.

So now we have two statements: the expected number of short cycles goes to zero, and the probability that there is an independent set of size at least r goes to zero. If we pick a large enough n, then the expected number of short cycles is less than n/5, and using Markov’s inequality we see that the probability that there are more than n/2 cycles of length at most g is strictly less than 1/2. At the same time, if we pick a large enough n then \alpha(G) \geq r with probability strictly less than 1/2. Combining these two (once more with the union bound), we get

\textup{P}(\textup{at least } n/2 \textup{ cycles of length } \leq g \textup{ and } \alpha(G) \geq r) < 1

Now we can conclude that for all sufficiently large n there has to be a graph on at least n vertices which has neither of these two properties! Pick one and call it G. Now G still has cycles of length \leq g, but we can fix that by removing a vertex from each short cycle (it doesn’t matter which). Call this new graph G'. How does this operation affect independent sets, i.e. what is \alpha(G')? Well removing vertices can only decrease the size of the largest independent set. So by our earlier inequality, and calling n' the number of vertices of G', we can make a statement about the chromatic number:

\displaystyle \chi(G') \geq \frac{n'}{\alpha(G')} \geq \frac{n/2}{\log(n) 3/p} = \frac{n/2}{3n^l \log(n)} = \frac{n^{1-l}}{6 \log(n)}

Since -1 < l < 0 the numerator grows asymptotically faster than the denominator, and so for sufficiently large n the chromatic number will be larger than any k we wish. Hence we have found a graph with girth at least g and chromatic number at least k.

\square

Connected Components

The statistical properties of a random graph are often quite easy to reason about. For instance, the degree of each vertex in G(n,p) is np in expectation. Local properties like this are easy, but global properties are a priori very mysterious. One natural question we can ask in this vein is: when is G(n,p) connected? We would very much expect the answer to depend on how p changes in relation to n. For instance, p might look like p(n) = 1/n^2 or \log(n) / n or something similar. We could ask the following question:

As n tends to infinity, what limiting proportion of random graphs G(n,p) are connected?

Certainly for some p(n) which are egregiously small (for example, p(n) = 0), the answer will be that no graphs are connected. On the other extreme, if p(n) = 1 then all graphs will be connected (and complete graphs, at that!). So our goal is to study the transition phase between when the graphs are disconnected and when they are connected. A priori this boundary could be a gradual slope, where the proportion grows from zero to one, or it could be a sharp jump. Next time, we’ll formally state and prove the truth, but for now let’s see if we can figure out which answer to expect by writing an exploratory program.

We wrote the code for this post in Python, and as usual it is all available for download on this blog’s Github page.

We start with a very basic Node class to represent each vertex in a graph, and a function to generate random graphs

import random
class Node:
   def __init__(self, index):
      self.index = index
      self.neighbors = []

   def __repr__(self):
      return repr(self.index)

def randomGraph(n,p):
   vertices = [Node(i) for i in range(n)]
   edges = [(i,j) for i in xrange(n) for j in xrange(i) if random.random() < p]

   for (i,j) in edges:
      vertices[i].neighbors.append(vertices[j])
      vertices[j].neighbors.append(vertices[i])

   return vertices

The randomGraph function simply creates a list of edges chosen uniformly at random from all possible edges, and then constructs the corresponding graph. Next we have a familiar sight: the depth-first search. We use it to compute the graph components one by one (until all vertices have been found in some run of a DFS).

def dfsComponent(node, visited):
   for v in node.neighbors:
      if v not in visited:
         visited.add(v)
         dfsComponent(v, visited)

def connectedComponents(vertices):
   components = []
   cumulativeVisited = set()

   for v in vertices:
      if v not in cumulativeVisited:
        componentVisited = set([v])
        dfsComponent(v, componentVisited)

        components.append(componentVisited)
        cumulativeVisited |= componentVisited

   return components

The dfsComponent function simply searches in breadth-first fashion, adding every vertex it finds to the “visited” set.  The connectedComponents function keeps track of the list of components found so far, as well as the cumulative set of all vertices found in any run of bfsComponent. Hence, as we iterate through the vertices we can ignore vertices we’ve found in previous runs of bfsComponent. The “x |= y” notation is python shorthand for updating x via a union with y.

Finally, we can make a graph of the largest component of (independently generated) random graphs as the probability of an edge varies.

def sizeOfLargestComponent(vertices):
   return max(len(c) for c in connectedComponents(vertices))

def graphLargestComponentSize(n, theRange):
   return [(p, sizeOfLargestComponent(randomGraph(n, p))) for p in theRange]

Running this code and plotting it for p varying from zero to 0.5 gives the following graph.

zoomedout-50-1000

The blue line is the size of the largest component, and the red line gives a moving average estimate of the data.  As we can see, there is a very sharp jump peaking at p=0.1 at which point the whole graph is connected. It would appear there is a relatively quick “phase transition” happening here. Let’s zoom in closer on the interesting part.

zoomedin-50-1000

It looks like the transition begins around 0.02, and finishes at around 0.1. Interesting… Let’s change the parameters a bit, and increase the size of the graph. Here’s the same chart (in the same range of p values) for a graph with a hundred vertices.

zoomedin-100-1000

Now the phase transition appears to have shifted to about (0.01, 0.05), which is about multiplying the endpoints of the phase transition interval above by 1/2. The plot thickens… Once more, let’s move up to a graph on 500 vertices.

zoomedin-5000-1000

Again it’s too hard to see, so let’s zoom in.

zoomedin-500-1000

This one looks like the transition starts at 0.002 and ends at 0.01. This is a 5-fold decrease from the previous one, and we increased the number of vertices by 5. Could this be a pattern? Here’s a conjecture to formalize it:

Conjecture: The random graph G(n,p) enters a phase transition at p=1/n and becomes connected almost surely at p=5/n.

This is not quite rigorous enough to be a true conjecture, but it sums up our intuition that we’ve learned so far. Just to back this up even further, here’s an animation showing the progression of the phase transition as n = 20 \dots 500 in steps of twenty. Note that the p range is changing to maintain our conjectured window.

phase-transition-n-grows

Looks pretty good. Next time we’ll see some formal mathematics validating our intuition (albeit reformulated in a nicer way), and we’ll continue to investigate other random graph models.

Until then!