# Zero Knowledge Proofs — A Primer

In this post we’ll get a strong taste for zero knowledge proofs by exploring the graph isomorphism problem in detail. In the next post, we’ll see how this relates to cryptography and the bigger picture. The goal of this post is to get a strong understanding of the terms “prover,” “verifier,” and “simulator,” and “zero knowledge” in the context of a specific zero-knowledge proof. Then next time we’ll see how the same concepts (though not the same proof) generalizes to a cryptographically interesting setting.

## Graph isomorphism

Let’s start with an extended example. We are given two graphs $G_1, G_2$, and we’d like to know whether they’re isomorphic, meaning they’re the same graph, but “drawn” different ways.

The problem of telling if two graphs are isomorphic seems hard. The pictures above, which are all different drawings of the same graph (or are they?), should give you pause if you thought it was easy.

To add a tiny bit of formalism, a graph $G$ is a list of edges, and each edge $(u,v)$ is a pair of integers between 1 and the total number of vertices of the graph, say $n$. Using this representation, an isomorphism between $G_1$ and $G_2$ is a permutation $\pi$ of the numbers $\{1, 2, \dots, n \}$ with the property that $(i,j)$ is an edge in $G_1$ if and only if $(\pi(i), \pi(j))$ is an edge of $G_2$. You swap around the labels on the vertices, and that’s how you get from one graph to another isomorphic one.

Given two arbitrary graphs as input on a large number of vertices $n$, nobody knows of an efficient—i.e., polynomial time in $n$—algorithm that can always decide whether the input graphs are isomorphic. Even if you promise me that the inputs are isomorphic, nobody knows of an algorithm that could construct an isomorphism. (If you think about it, such an algorithm could be used to solve the decision problem!)

## A game

Now let’s play a game. In this game, we’re given two enormous graphs on a billion nodes. I claim they’re isomorphic, and I want to prove it to you. However, my life’s fortune is locked behind these particular graphs (somehow), and if you actually had an isomorphism between these two graphs you could use it to steal all my money. But I still want to convince you that I do, in fact, own all of this money, because we’re about to start a business and you need to know I’m not broke.

Is there a way for me to convince you beyond a reasonable doubt that these two graphs are indeed isomorphic? And moreover, could I do so without you gaining access to my secret isomorphism? It would be even better if I could guarantee you learn nothing about my isomorphism or any isomorphism, because even the slightest chance that you can steal my money is out of the question.

Zero knowledge proofs have exactly those properties, and here’s a zero knowledge proof for graph isomorphism. For the record, $G_1$ and $G_2$ are public knowledge, (common inputs to our protocol for the sake of tracking runtime), and the protocol itself is common knowledge. However, I have an isomorphism $f: G_1 \to G_2$ that you don’t know.

Step 1: I will start by picking one of my two graphs, say $G_1$, mixing up the vertices, and sending you the resulting graph. In other words, I send you a graph $H$ which is chosen uniformly at random from all isomorphic copies of $G_1$. I will save the permutation $\pi$ that I used to generate $H$ for later use.

Step 2: You receive a graph $H$ which you save for later, and then you randomly pick an integer $t$ which is either 1 or 2, with equal probability on each. The number $t$ corresponds to your challenge for me to prove $H$ is isomorphic to $G_1$ or $G_2$. You send me back $t$, with the expectation that I will provide you with an isomorphism between $H$ and $G_t$.

Step 3: Indeed, I faithfully provide you such an isomorphism. If I you send me $t=1$, I’ll give you back $\pi^{-1} : H \to G_1$, and otherwise I’ll give you back $f \circ \pi^{-1}: H \to G_2$. Because composing a fixed permutation with a uniformly random permutation is again a uniformly random permutation, in either case I’m sending you a uniformly random permutation.

Step 4: You receive a permutation $g$, and you can use it to verify that $H$ is isomorphic to $G_t$. If the permutation I sent you doesn’t work, you’ll reject my claim, and if it does, you’ll accept my claim.

Before we analyze, here’s some Python code that implements the above scheme. You can find the full, working example in a repository on this blog’s Github page.

First, a few helper functions for generating random permutations (and turning their list-of-zero-based-indices form into a function-of-positive-integers form)

import random

def randomPermutation(n):
L = list(range(n))
random.shuffle(L)
return L

def makePermutationFunction(L):
return lambda i: L[i - 1] + 1

def makeInversePermutationFunction(L):
return lambda i: 1 + L.index(i - 1)

def applyIsomorphism(G, f):
return [(f(i), f(j)) for (i, j) in G]


Here’s a class for the Prover, the one who knows the isomorphism and wants to prove it while keeping the isomorphism secret:

class Prover(object):
def __init__(self, G1, G2, isomorphism):
'''
isomomorphism is a list of integers representing
an isomoprhism from G1 to G2.
'''
self.G1 = G1
self.G2 = G2
self.n = numVertices(G1)
assert self.n == numVertices(G2)

self.isomorphism = isomorphism
self.state = None

def sendIsomorphicCopy(self):
isomorphism = randomPermutation(self.n)
pi = makePermutationFunction(isomorphism)

H = applyIsomorphism(self.G1, pi)

self.state = isomorphism
return H

def proveIsomorphicTo(self, graphChoice):
randomIsomorphism = self.state
piInverse = makeInversePermutationFunction(randomIsomorphism)

if graphChoice == 1:
return piInverse
else:
f = makePermutationFunction(self.isomorphism)
return lambda i: f(piInverse(i))


The prover has two methods, one for each round of the protocol. The first creates an isomorphic copy of $G_1$, and the second receives the challenge and produces the requested isomorphism.

And here’s the corresponding class for the verifier

class Verifier(object):
def __init__(self, G1, G2):
self.G1 = G1
self.G2 = G2
self.n = numVertices(G1)
assert self.n == numVertices(G2)

def chooseGraph(self, H):
choice = random.choice([1, 2])
self.state = H, choice
return choice

def accepts(self, isomorphism):
'''
Return True if and only if the given isomorphism
is a valid isomorphism between the randomly
chosen graph in the first step, and the H presented
by the Prover.
'''
H, choice = self.state
graphToCheck = [self.G1, self.G2][choice - 1]
f = isomorphism

isValidIsomorphism = (graphToCheck == applyIsomorphism(H, f))
return isValidIsomorphism


Then the protocol is as follows:

def runProtocol(G1, G2, isomorphism):
p = Prover(G1, G2, isomorphism)
v = Verifier(G1, G2)

H = p.sendIsomorphicCopy()
choice = v.chooseGraph(H)
witnessIsomorphism = p.proveIsomorphicTo(choice)

return v.accepts(witnessIsomorphism)


Analysis: Let’s suppose for a moment that everyone is honestly following the rules, and that $G_1, G_2$ are truly isomorphic. Then you’ll always accept my claim, because I can always provide you with an isomorphism. Now let’s suppose that, actually I’m lying, the two graphs aren’t isomorphic, and I’m trying to fool you into thinking they are. What’s the probability that you’ll rightfully reject my claim?

Well, regardless of what I do, I’m sending you a graph $H$ and you get to make a random choice of $t = 1, 2$ that I can’t control. If $H$ is only actually isomorphic to either $G_1$ or $G_2$ but not both, then so long as you make your choice uniformly at random, half of the time I won’t be able to produce a valid isomorphism and you’ll reject. And unless you can actually tell which graph $H$ is isomorphic to—an open problem, but let’s say you can’t—then probability 1/2 is the best you can do.

Maybe the probability 1/2 is a bit unsatisfying, but remember that we can amplify this probability by repeating the protocol over and over again. So if you want to be sure I didn’t cheat and get lucky to within a probability of one-in-one-trillion, you only need to repeat the protocol 30 times. To be surer than the chance of picking a specific atom at random from all atoms in the universe, only about 400 times.

If you want to feel small, think of the number of atoms in the universe. If you want to feel big, think of its logarithm.

Here’s the code that repeats the protocol for assurance.

def convinceBeyondDoubt(G1, G2, isomorphism, errorTolerance=1e-20):
probabilityFooled = 1

while probabilityFooled > errorTolerance:
result = runProtocol(G1, G2, isomorphism)
assert result
probabilityFooled *= 0.5
print(probabilityFooled)


Running it, we see it succeeds

python graph-isomorphism.py 0.5 0.25 0.125 0.0625 0.03125 ... &lt;SNIP&gt; ... 1.3552527156068805e-20 6.776263578034403e-21  So it’s clear that this protocol is convincing. But how can we be sure that there’s no leakage of knowledge in the protocol? What does “leakage” even mean? That’s where this topic is the most difficult to nail down rigorously, in part because there are at least three a priori different definitions! The idea we want to capture is that anything that you can efficiently compute after the protocol finishes (i.e., you have the content of the messages sent to you by the prover) you could have computed efficiently given only the two graphs $G_1, G_2$, and the claim that they are isomorphic. Another way to say it is that you may go through the verification process and feel happy and confident that the two graphs are isomorphic. But because it’s a zero-knowledge proof, you can’t do anything with that information more than you could have done if you just took the assertion on blind faith. I’m confident there’s a joke about religion lurking here somewhere, but I’ll just trust it’s funny and move on. In the next post we’ll expand on this “leakage” notion, but before we get there it should be clear that the graph isomorphism protocol will have the strongest possible “no-leakage” property we can come up with. Indeed, in the first round the prover sends a uniform random isomorphic copy of $G_1$ to the verifier, but the verifier can compute such an isomorphism already without the help of the prover. The verifier can’t necessarily find the isomorphism that the prover used in retrospect, because the verifier can’t solve graph isomorphism. Instead, the point is that the probability space of “$G_1$ paired with an $H$ made by the prover” and the probability space of “$G_1$ paired with $H$ as made by the verifier” are equal. No information was leaked by the prover. For the second round, again the permutation $\pi$ used by the prover to generate $H$ is uniformly random. Since composing a fixed permutation with a uniform random permutation also results in a uniform random permutation, the second message sent by the prover is uniformly random, and so again the verifier could have constructed a similarly random permutation alone. Let’s make this explicit with a small program. We have the honest protocol from before, but now I’m returning the set of messages sent by the prover, which the verifier can use for additional computation. def messagesFromProtocol(G1, G2, isomorphism): p = Prover(G1, G2, isomorphism) v = Verifier(G1, G2) H = p.sendIsomorphicCopy() choice = v.chooseGraph(H) witnessIsomorphism = p.proveIsomorphicTo(choice) return [H, choice, witnessIsomorphism]  To say that the protocol is zero-knowledge (again, this is still colloquial) is to say that anything that the verifier could compute, given as input the return value of this function along with $G_1, G_2$ and the claim that they’re isomorphic, the verifier could also compute given only $G_1, G_2$ and the claim that $G_1, G_2$ are isomorphic. It’s easy to prove this, and we’ll do so with a python function called simulateProtocol. def simulateProtocol(G1, G2): # Construct data drawn from the same distribution as what is # returned by messagesFromProtocol choice = random.choice([1, 2]) G = [G1, G2][choice - 1] n = numVertices(G) isomorphism = randomPermutation(n) pi = makePermutationFunction(isomorphism) H = applyIsomorphism(G, pi) return H, choice, pi  The claim is that the distribution of outputs to messagesFromProtocol and simulateProtocol are equal. But simulateProtocol will work regardless of whether $G_1, G_2$ are isomorphic. Of course, it’s not convincing to the verifier because the simulating function made the choices in the wrong order, choosing the graph index before making $H$. But the distribution that results is the same either way. So if you were to use the actual Prover/Verifier protocol outputs as input to another algorithm (say, one which tries to compute an isomorphism of $G_1 \to G_2$), you might as well use the output of your simulator instead. You’d have no information beyond hard-coding the assumption that $G_1, G_2$ are isomorphic into your program. Which, as I mentioned earlier, is no help at all. In this post we covered one detailed example of a zero-knowledge proof. Next time we’ll broaden our view and see the more general power of zero-knowledge (that it captures all of NP), and see some specific cryptographic applications. Keep in mind the preceding discussion, because we’re going to re-use the terms “prover,” “verifier,” and “simulator” to mean roughly the same things as the classes Prover, Verifier and the function simulateProtocol. Until then! # Singular Value Decomposition Part 2: Theorem, Proof, Algorithm I’m just going to jump right into the definitions and rigor, so if you haven’t read the previous post motivating the singular value decomposition, go back and do that first. This post will be theorem, proof, algorithm, data. The data set we test on is a thousand-story CNN news data set. All of the data, code, and examples used in this post is in a github repository, as usual. We start with the best-approximating $k$-dimensional linear subspace. Definition: Let $X = \{ x_1, \dots, x_m \}$ be a set of $m$ points in $\mathbb{R}^n$. The best approximating $k$-dimensional linear subspace of $X$ is the $k$-dimensional linear subspace $V \subset \mathbb{R}^n$ which minimizes the sum of the squared distances from the points in $X$ to $V$. Let me clarify what I mean by minimizing the sum of squared distances. First we’ll start with the simple case: we have a vector $x \in X$, and a candidate line $L$ (a 1-dimensional subspace) that is the span of a unit vector $v$. The squared distance from $x$ to the line spanned by $v$ is the squared length of $x$ minus the squared length of the projection of $x$ onto $v$. Here’s a picture. I’m saying that the pink vector $z$ in the picture is the difference of the black and green vectors $x-y$, and that the “distance” from $x$ to $v$ is the length of the pink vector. The reason is just the Pythagorean theorem: the vector $x$ is the hypotenuse of a right triangle whose other two sides are the projected vector $y$ and the difference vector $z$. Let’s throw down some notation. I’ll call $\textup{proj}_v: \mathbb{R}^n \to \mathbb{R}^n$ the linear map that takes as input a vector $x$ and produces as output the projection of $x$ onto $v$. In fact we have a brief formula for this when $v$ is a unit vector. If we call $x \cdot v$ the usual dot product, then $\textup{proj}_v(x) = (x \cdot v)v$. That’s $v$ scaled by the inner product of $x$ and $v$. In the picture above, since the line $L$ is the span of the vector $v$, that means that $y = \textup{proj}_v(x)$ and $z = x -\textup{proj}_v(x) = x-y$. The dot-product formula is useful for us because it allows us to compute the squared length of the projection by taking a dot product $|x \cdot v|^2$. So then a formula for the distance of $x$ from the line spanned by the unit vector $v$ is $\displaystyle (\textup{dist}_v(x))^2 = \left ( \sum_{i=1}^n x_i^2 \right ) - |x \cdot v|^2$ This formula is just a restatement of the Pythagorean theorem for perpendicular vectors. $\displaystyle \sum_{i} x_i^2 = (\textup{proj}_v(x))^2 + (\textup{dist}_v(x))^2$ In particular, the difference vector we originally called $z$ has squared length $\textup{dist}_v(x)^2$. The vector $y$, which is perpendicular to $z$ and is also the projection of $x$ onto $L$, it’s squared length is $(\textup{proj}_v(x))^2$. And the Pythagorean theorem tells us that summing those two squared lengths gives you the squared length of the hypotenuse $x$. If we were trying to find the best approximating 1-dimensional subspace for a set of data points $X$, then we’d want to minimize the sum of the squared distances for every point $x \in X$. Namely, we want the $v$ that solves $\min_{|v|=1} \sum_{x \in X} (\textup{dist}_v(x))^2$. With some slight algebra we can make our life easier. The short version: minimizing the sum of squared distances is the same thing as maximizing the sum of squared lengths of the projections. The longer version: let’s go back to a single point $x$ and the line spanned by $v$. The Pythagorean theorem told us that $\displaystyle \sum_{i} x_i^2 = (\textup{proj}_v(x))^2 + (\textup{dist}_v(x))^2$ The squared length of $x$ is constant. It’s an input to the algorithm and it doesn’t change through a run of the algorithm. So we get the squared distance by subtracting $(\textup{proj}_v(x))^2$ from a constant number, $\displaystyle \sum_{i} x_i^2 - (\textup{proj}_v(x))^2 = (\textup{dist}_v(x))^2$ which means if we want to minimize the squared distance, we can instead maximize the squared projection. Maximizing the subtracted thing minimizes the whole expression. It works the same way if you’re summing over all the data points in $X$. In fact, we can say it much more compactly this way. If the rows of $A$ are your data points, then $Av$ contains as each entry the (signed) dot products $x_i \cdot v$. And the squared norm of this vector, $|Av|^2$, is exactly the sum of the squared lengths of the projections of the data onto the line spanned by $v$. The last thing is that maximizing a square is the same as maximizing its square root, so we can switch freely between saying our objective is to find the unit vector $v$ that maximizes $|Av|$ and that which maximizes $|Av|^2$. At this point you should be thinking, Great, we have written down an optimization problem: $\max_{v : |v|=1} |Av|$. If we could solve this, we’d have the best 1-dimensional linear approximation to the data contained in the rows of $A$. But (1) how do we solve that problem? And (2) you promised a $k$-dimensional approximating subspace. I feel betrayed! Swindled! Bamboozled! Here’s the fantastic thing. We can solve the 1-dimensional optimization problem efficiently (we’ll do it later in this post), and (2) is answered by the following theorem. The SVD Theorem: Computing the best $k$-dimensional subspace reduces to $k$ applications of the one-dimensional problem. We will prove this after we introduce the terms “singular value” and “singular vector.” ## Singular values and vectors As I just said, we can get the best $k$-dimensional approximating linear subspace by solving the one-dimensional maximization problem $k$ times. The singular vectors of $A$ are defined recursively as the solutions to these sub-problems. That is, I’ll call $v_1$ the first singular vector of $A$, and it is: $\displaystyle v_1 = \arg \max_{v, |v|=1} |Av|$ And the corresponding first singular value, denoted $\sigma_1(A)$, is the maximal value of the optimization objective, i.e. $|Av_1|$. (I will use this term frequently, that $|Av|$ is the “objective” of the optimization problem.) Informally speaking, $(\sigma_1(A))^2$ represents how much of the data was captured by the first singular vector. Meaning, how close the vectors are to lying on the line spanned by $v_1$. Larger values imply the approximation is better. In fact, if all the data points lie on a line, then $(\sigma_1(A))^2$ is the sum of the squared norms of the rows of $A$. Now here is where we see the reduction from the $k$-dimensional case to the 1-dimensional case. To find the best 2-dimensional subspace, you first find the best one-dimensional subspace (spanned by $v_1$), and then find the best 1-dimensional subspace, but only considering those subspaces that are the spans of unit vectors perpendicular to $v_1$. The notation for “vectors $v$ perpendicular to $v_1$” is $v \perp v_1$. Restating, the second singular vector $v _2$ is defined as $\displaystyle v_2 = \arg \max_{v \perp v_1, |v| = 1} |Av|$ And the SVD theorem implies the subspace spanned by $\{ v_1, v_2 \}$ is the best 2-dimensional linear approximation to the data. Likewise $\sigma_2(A) = |Av_2|$ is the second singular value. Its squared magnitude tells us how much of the data that was not “captured” by $v_1$ is captured by $v_2$. Again, if the data lies in a 2-dimensional subspace, then the span of $\{ v_1, v_2 \}$ will be that subspace. We can continue this process. Recursively define $v_k$, the $k$-th singular vector, to be the vector which maximizes $|Av|$, when $v$ is considered only among the unit vectors which are perpendicular to $\textup{span} \{ v_1, \dots, v_{k-1} \}$. The corresponding singular value $\sigma_k(A)$ is the value of the optimization problem. As a side note, because of the way we defined the singular values as the objective values of “nested” optimization problems, the singular values are decreasing, $\sigma_1(A) \geq \sigma_2(A) \geq \dots \geq \sigma_n(A) \geq 0$. This is obvious: you only pick $v_2$ in the second optimization problem because you already picked $v_1$ which gave a bigger singular value, so $v_2$‘s objective can’t be bigger. If you keep doing this, one of two things happen. Either you reach $v_n$ and since the domain is $n$-dimensional there are no remaining vectors to choose from, the $v_i$ are an orthonormal basis of $\mathbb{R}^n$. This means that the data in $A$ contains a full-rank submatrix. The data does not lie in any smaller-dimensional subspace. This is what you’d expect from real data. Alternatively, you could get to a stage $v_k$ with $k < n$ and when you try to solve the optimization problem you find that every perpendicular $v$ has $Av = 0$. In this case, the data actually does lie in a $k$-dimensional subspace, and the first-through-$k$-th singular vectors you computed span this subspace. Let’s do a quick sanity check: how do we know that the singular vectors $v_i$ form a basis? Well formally they only span a basis of the column space of $A$, i.e. a basis of the subspace spanned by the data contained in the columns of $A$. But either way the point is that each $v_{i+1}$ spans a new dimension from the previous $v_1, \dots, v_i$ because we’re choosing $v_{i+1}$ to be orthogonal to all the previous $v_i$. So the answer to our sanity check is “by construction.” Back to the singular vectors, the discussion from the last post tells us intuitively that the data is probably never in a small subspace. You never expect the process of finding singular vectors to stop before step $n$, and if it does you take a step back and ask if something deeper is going on. Instead, in real life you specify how much of the data you want to capture, and you keep computing singular vectors until you’ve passed the threshold. Alternatively, you specify the amount of computing resources you’d like to spend by fixing the number of singular vectors you’ll compute ahead of time, and settle for however good the $k$-dimensional approximation is. Before we get into any code or solve the 1-dimensional optimization problem, let’s prove the SVD theorem. Proof of SVD theorem. Recall we’re trying to prove that the first $k$ singular vectors provide a linear subspace $W$ which maximizes the squared-sum of the projections of the data onto $W$. For $k=1$ this is trivial, because we defined $v_1$ to be the solution to that optimization problem. The case of $k=2$ contains all the important features of the general inductive step. Let $W$ be any best-approximating 2-dimensional linear subspace for the rows of $A$. We’ll show that the subspace spanned by the two singular vectors $v_1, v_2$ is at least as good (and hence equally good). Let $w_1, w_2$ be any orthonormal basis for $W$ and let $|Aw_1|^2 + |Aw_2|^2$ be the quantity that we’re trying to maximize (and which $W$ maximizes by assumption). Moreover, we can pick the basis vector $w_2$ to be perpendicular to $v_1$. To prove this we consider two cases: either $v_1$ is already perpendicular to $W$ in which case it’s trivial, or else $v_1$ isn’t perpendicular to $W$ and you can choose $w_1$ to be $\textup{proj}_W(v_1)$ and choose $w_2$ to be any unit vector perpendicular to $w_1$. Now since $v_1$ maximizes $|Av|$, we have $|Av_1|^2 \geq |Aw_1|^2$. Moreover, since $w_2$ is perpendicular to $v_1$, the way we chose $v_2$ also makes $|Av_2|^2 \geq |Aw_2|^2$. Hence the objective $|Av_1|^2 + |Av_2|^2 \geq |Aw_1|^2 + |Aw_2|^2$, as desired. For the general case of $k$, the inductive hypothesis tells us that the first $k$ terms of the objective for $k+1$ singular vectors is maximized, and we just have to pick any vector $w_{k+1}$ that is perpendicular to all $v_1, v_2, \dots, v_k$, and the rest of the proof is just like the 2-dimensional case. $\square$ Now remember that in the last post we started with the definition of the SVD as a decomposition of a matrix $A = U\Sigma V^T$? And then we said that this is a certain kind of change of basis? Well the singular vectors $v_i$ together form the columns of the matrix $V$ (the rows of $V^T$), and the corresponding singular values $\sigma_i(A)$ are the diagonal entries of $\Sigma$. When $A$ is understood we’ll abbreviate the singular value as $\sigma_i$. To reiterate with the thoughts from last post, the process of applying $A$ is exactly recovered by the process of first projecting onto the (full-rank space of) singular vectors $v_1, \dots, v_k$, scaling each coordinate of that projection according to the corresponding singular values, and then applying this $U$ thing we haven’t talked about yet. So let’s determine what $U$ has to be. The way we picked $v_i$ to make $A$ diagonal gives us an immediate suggestion: use the $Av_i$ as the columns of $U$. Indeed, define $u_i = Av_i$, the images of the singular vectors under $A$. We can swiftly show the $u_i$ form a basis of the image of $A$. The reason is because if $v = \sum_i c_i v_i$ (using all $n$ of the singular vectors $v_i$), then by linearity $Av = \sum_{i} c_i Av_i = \sum_i c_i u_i$. It is also easy to see why the $u_i$ are orthogonal (prove it as an exercise). Let’s further make sure the $u_i$ are unit vectors and redefine them as $u_i = \frac{1}{\sigma_i}Av_i$ If you put these thoughts together, you can say exactly what $A$ does to any given vector $x$. Since the $v_i$ form an orthonormal basis, $x = \sum_i (x \cdot v_i) v_i$, and then applying $A$ gives \displaystyle \begin{aligned}Ax &= A \left ( \sum_i (x \cdot v_i) v_i \right ) \\ &= \sum_i (x \cdot v_i) A_i v_i \\ &= \sum_i (x \cdot v_i) \sigma_i u_i \end{aligned} If you’ve been closely reading this blog in the last few months, you’ll recognize a very nice way to write the last line of the above equation. It’s an outer product. So depending on your favorite symbols, you’d write this as either $A = \sum_{i} \sigma_i u_i \otimes v_i$ or $A = \sum_i \sigma_i u_i v_i^T$. Or, if you like expressing things as matrix factorizations, as $A = U\Sigma V^T$. All three are describing the same object. Let’s move on to some code. ## A black box example Before we implement SVD from scratch (an urge that commands me from the depths of my soul!), let’s see a black-box example that uses existing tools. For this we’ll use the numpy library. Recall our movie-rating matrix from the last post: The code to compute the svd of this matrix is as simple as it gets: from numpy.linalg import svd movieRatings = [ [2, 5, 3], [1, 2, 1], [4, 1, 1], [3, 5, 2], [5, 3, 1], [4, 5, 5], [2, 4, 2], [2, 2, 5], ] U, singularValues, V = svd(movieRatings)  Printing these values out gives [[-0.39458526 0.23923575 -0.35445911 -0.38062172 -0.29836818 -0.49464816 -0.30703202 -0.29763321] [-0.15830232 0.03054913 -0.15299759 -0.45334816 0.31122898 0.23892035 -0.37313346 0.67223457] [-0.22155201 -0.52086121 0.39334917 -0.14974792 -0.65963979 0.00488292 -0.00783684 0.25934607] [-0.39692635 -0.08649009 -0.41052882 0.74387448 -0.10629499 0.01372565 -0.17959298 0.26333462] [-0.34630257 -0.64128825 0.07382859 -0.04494155 0.58000668 -0.25806239 0.00211823 -0.24154726] [-0.53347449 0.19168874 0.19949342 -0.03942604 0.00424495 0.68715732 -0.06957561 -0.40033035] [-0.31660464 0.06109826 -0.30599517 -0.19611823 -0.01334272 0.01446975 0.85185852 0.19463493] [-0.32840223 0.45970413 0.62354764 0.1783041 0.17631186 -0.39879476 0.06065902 0.25771578]] [ 15.09626916 4.30056855 3.40701739] [[-0.54184808 -0.67070995 -0.50650649] [-0.75152295 0.11680911 0.64928336] [ 0.37631623 -0.73246419 0.56734672]]  Now this is a bit weird, because the matrices $U, V$ are the wrong shape! Remember, there are only supposed to be three vectors since the input matrix has rank three. So what gives? This is a distinction that goes by the name “full” versus “reduced” SVD. The idea goes back to our original statement that $U \Sigma V^T$ is a decomposition with $U, V^T$ both orthogonal and square matrices. But in the derivation we did in the last section, the $U$ and $V$ were not square. The singular vectors $v_i$ could potentially stop before even becoming full rank. In order to get to square matrices, what people sometimes do is take the two bases $v_1, \dots, v_k$ and $u_1, \dots, u_k$ and arbitrarily choose ways to complete them to a full orthonormal basis of their respective vector spaces. In other words, they just make the matrix square by filling it with data for no reason other than that it’s sometimes nice to have a complete basis. We don’t care about this. To be honest, I think the only place this comes in useful is in the desire to be particularly tidy in a mathematical formulation of something. We can still work with it programmatically. By fudging around a bit with numpy’s shapes to get a diagonal matrix, we can reconstruct the input rating matrix from the factors. Sigma = np.vstack([ np.diag(singularValues), np.zeros((5, 3)), ]) print(np.round(movieRatings - np.dot(U, np.dot(Sigma, V)), decimals=10))  And the output is, as one expects, a matrix of all zeros. Meaning that we decomposed the movie rating matrix, and built it back up from the factors. We can actually get the SVD as we defined it (with rectangular matrices) by passing a special flag to numpy’s svd. U, singularValues, V = svd(movieRatings, full_matrices=False) print(U) print(singularValues) print(V) Sigma = np.diag(singularValues) print(np.round(movieRatings - np.dot(U, np.dot(Sigma, V)), decimals=10))  And the result [[-0.39458526 0.23923575 -0.35445911] [-0.15830232 0.03054913 -0.15299759] [-0.22155201 -0.52086121 0.39334917] [-0.39692635 -0.08649009 -0.41052882] [-0.34630257 -0.64128825 0.07382859] [-0.53347449 0.19168874 0.19949342] [-0.31660464 0.06109826 -0.30599517] [-0.32840223 0.45970413 0.62354764]] [ 15.09626916 4.30056855 3.40701739] [[-0.54184808 -0.67070995 -0.50650649] [-0.75152295 0.11680911 0.64928336] [ 0.37631623 -0.73246419 0.56734672]] [[-0. -0. -0.] [-0. -0. 0.] [ 0. -0. 0.] [-0. -0. -0.] [-0. -0. -0.] [-0. -0. -0.] [-0. -0. -0.] [ 0. -0. -0.]]  This makes the reconstruction less messy, since we can just multiply everything without having to add extra rows of zeros to $\Sigma$. What do the singular vectors and values tell us about the movie rating matrix? (Besides nothing, since it’s a contrived example) You’ll notice that the first singular vector $\sigma_1 > 15$ while the other two singular values are around $4$. This tells us that the first singular vector covers a large part of the structure of the matrix. I.e., a rank-1 matrix would be a pretty good approximation to the whole thing. As an exercise to the reader, write a program that evaluates this claim (how good is “good”?). ## The greedy optimization routine Now we’re going to write SVD from scratch. We’ll first implement the greedy algorithm for the 1-d optimization problem, and then we’ll perform the inductive step to get a full algorithm. Then we’ll run it on the CNN data set. The method we’ll use to solve the 1-dimensional problem isn’t necessarily industry strength (see this document for a hint of what industry strength looks like), but it is simple conceptually. It’s called the power method. Now that we have our decomposition of theorem, understanding how the power method works is quite easy. Let’s work in the language of a matrix decomposition $A = U \Sigma V^T$, more for practice with that language than anything else (using outer products would give us the same result with slightly different computations). Then let’s observe $A^T A$, wherein we’ll use the fact that $U$ is orthonormal and so $U^TU$ is the identity matrix: $\displaystyle A^TA = (U \Sigma V^T)^T(U \Sigma V^T) = V \Sigma U^TU \Sigma V^T = V \Sigma^2 V^T$ So we can completely eliminate $U$ from the discussion, and look at just $V \Sigma^2 V^T$. And what’s nice about this matrix is that we can compute its eigenvectors, and eigenvectors turn out to be exactly the singular vectors. The corresponding eigenvalues are the squared singular values. This should be clear from the above derivation. If you apply $(V \Sigma^2 V^T)$ to any $v_i$, the only parts of the product that aren’t zero are the ones involving $v_i$ with itself, and the scalar $\sigma_i^2$ factors in smoothly. It’s dead simple to check. Theorem: Let $x$ be a random unit vector and let $B = A^TA = V \Sigma^2 V^T$. Then with high probability, $\lim_{s \to \infty} B^s x$ is in the span of the first singular vector $v_1$. If we normalize $B^s x$ to a unit vector at each $s$, then furthermore the limit is $v_1$. Proof. Start with a random unit vector $x$, and write it in terms of the singular vectors $x = \sum_i c_i v_i$. That means $Bx = \sum_i c_i \sigma_i^2 v_i$. If you recursively apply this logic, you get $B^s x = \sum_i c_i \sigma_i^{2s} v_i$. In particular, the dot product of $(B^s x)$ with any $v_j$ is $c_i \sigma_j^{2s}$. What this means is that so long as the first singular value $\sigma_1$ is sufficiently larger than the second one $\sigma_2$, and in turn all the other singular values, the part of $B^s x$ corresponding to $v_1$ will be much larger than the rest. Recall that if you expand a vector in terms of an orthonormal basis, in this case $B^s x$ expanded in the $v_i$, the coefficient of $B^s x$ on $v_j$ is exactly the dot product. So to say that $B^sx$ converges to being in the span of $v_1$ is the same as saying that the ratio of these coefficients, $|(B^s x \cdot v_1)| / |(B^s x \cdot v_j)| \to \infty$ for any $j$. In other words, the coefficient corresponding to the first singular vector dominates all of the others. And so if we normalize, the coefficient of $B^s x$ corresponding to $v_1$ tends to 1, while the rest tend to zero. Indeed, this ratio is just $(\sigma_1 / \sigma_j)^{2s}$ and the base of this exponential is bigger than 1. $\square$ If you want to be a little more precise and find bounds on the number of iterations required to converge, you can. The worry is that your random starting vector is “too close” to one of the smaller singular vectors $v_j$, so that if the ratio of $\sigma_1 / \sigma_j$ is small, then the “pull” of $v_1$ won’t outweigh the pull of $v_j$ fast enough. Choosing a random unit vector allows you to ensure with high probability that this doesn’t happen. And conditioned on it not happening (or measuring “how far the event is from happening” precisely), you can compute a precise number of iterations required to converge. The last two pages of these lecture notes have all the details. We won’t compute a precise number of iterations. Instead we’ll just compute until the angle between $B^{s+1}x$ and $B^s x$ is very small. Here’s the algorithm import numpy as np from numpy.linalg import norm from random import normalvariate from math import sqrt def randomUnitVector(n): unnormalized = [normalvariate(0, 1) for _ in range(n)] theNorm = sqrt(sum(x * x for x in unnormalized)) return [x / theNorm for x in unnormalized] def svd_1d(A, epsilon=1e-10): ''' The one-dimensional SVD ''' n, m = A.shape x = randomUnitVector(m) lastV = None currentV = x B = np.dot(A.T, A) iterations = 0 while True: iterations += 1 lastV = currentV currentV = np.dot(B, lastV) currentV = currentV / norm(currentV) if abs(np.dot(currentV, lastV)) > 1 - epsilon: print("converged in {} iterations!".format(iterations)) return currentV  We start with a random unit vector $x$, and then loop computing $x_{t+1} = Bx_t$, renormalizing at each step. The condition for stopping is that the magnitude of the dot product between $x_t$ and $x_{t+1}$ (since they’re unit vectors, this is the cosine of the angle between them) is very close to 1. And using it on our movie ratings example: if __name__ == "__main__": movieRatings = np.array([ [2, 5, 3], [1, 2, 1], [4, 1, 1], [3, 5, 2], [5, 3, 1], [4, 5, 5], [2, 4, 2], [2, 2, 5], ], dtype='float64') print(svd_1d(movieRatings))  With the result converged in 6 iterations! [-0.54184805 -0.67070993 -0.50650655]  Note that the sign of the vector may be different from numpy’s output because we start with a random vector to begin with. The recursive step, getting from $v_1$ to the entire SVD, is equally straightforward. Say you start with the matrix $A$ and you compute $v_1$. You can use $v_1$ to compute $u_1$ and $\sigma_1(A)$. Then you want to ensure you’re ignoring all vectors in the span of $v_1$ for your next greedy optimization, and to do this you can simply subtract the rank 1 component of $A$ corresponding to $v_1$. I.e., set $A' = A - \sigma_1(A) u_1 v_1^T$. Then it’s easy to see that $\sigma_1(A') = \sigma_2(A)$ and basically all the singular vectors shift indices by 1 when going from $A$ to $A'$. Then you repeat. If that’s not clear enough, here’s the code. def svd(A, epsilon=1e-10): n, m = A.shape svdSoFar = [] for i in range(m): matrixFor1D = A.copy() for singularValue, u, v in svdSoFar[:i]: matrixFor1D -= singularValue * np.outer(u, v) v = svd_1d(matrixFor1D, epsilon=epsilon) # next singular vector u_unnormalized = np.dot(A, v) sigma = norm(u_unnormalized) # next singular value u = u_unnormalized / sigma svdSoFar.append((sigma, u, v)) # transform it into matrices of the right shape singularValues, us, vs = [np.array(x) for x in zip(*svdSoFar)] return singularValues, us.T, vs  And we can run this on our movie rating matrix to get the following >>> theSVD = svd(movieRatings) >>> theSVD[0] array([ 15.09626916, 4.30056855, 3.40701739]) >>> theSVD[1] array([[ 0.39458528, -0.23923093, 0.35446407], [ 0.15830233, -0.03054705, 0.15299815], [ 0.221552 , 0.52085578, -0.39336072], [ 0.39692636, 0.08649568, 0.41052666], [ 0.34630257, 0.64128719, -0.07384286], [ 0.53347448, -0.19169154, -0.19948959], [ 0.31660465, -0.0610941 , 0.30599629], [ 0.32840221, -0.45971273, -0.62353781]]) >>> theSVD[2] array([[ 0.54184805, 0.67071006, 0.50650638], [ 0.75151641, -0.11679644, -0.64929321], [-0.37632934, 0.73246611, -0.56733554]])  Checking this against our numpy output shows it’s within a reasonable level of precision (considering the power method took on the order of ten iterations!) >>> np.round(np.abs(npSVD[0]) - np.abs(theSVD[1]), decimals=5) array([[ -0.00000000e+00, -0.00000000e+00, 0.00000000e+00], [ 0.00000000e+00, -0.00000000e+00, 0.00000000e+00], [ 0.00000000e+00, -1.00000000e-05, 1.00000000e-05], [ 0.00000000e+00, 0.00000000e+00, -0.00000000e+00], [ 0.00000000e+00, -0.00000000e+00, 1.00000000e-05], [ -0.00000000e+00, 0.00000000e+00, -0.00000000e+00], [ 0.00000000e+00, -0.00000000e+00, 0.00000000e+00], [ -0.00000000e+00, 1.00000000e-05, -1.00000000e-05]]) >>> np.round(np.abs(npSVD[2]) - np.abs(theSVD[2]), decimals=5) array([[ 0.00000000e+00, 0.00000000e+00, -0.00000000e+00], [ -1.00000000e-05, -1.00000000e-05, 1.00000000e-05], [ 1.00000000e-05, 0.00000000e+00, -1.00000000e-05]]) >>> np.round(np.abs(npSVD[1]) - np.abs(theSVD[0]), decimals=5) array([ 0., 0., -0.])  So there we have it. We added an extra little bit to the svd function, an argument $k$ which stops computing the svd after it reaches rank $k$. ## CNN stories One interesting use of the SVD is in topic modeling. Topic modeling is the process of taking a bunch of documents (news stories, or emails, or movie scripts, whatever) and grouping them by topic, where the algorithm gets to choose what counts as a “topic.” Topic modeling is just the name that natural language processing folks use instead of clustering. The SVD can help one model topics as follows. First you construct a matrix $A$ called a document-term matrix whose rows correspond to words in some fixed dictionary and whose columns correspond to documents. The $(i,j)$ entry of $A$ contains the number of times word $i$ shows up in document $j$. Or, more precisely, some quantity derived from that count, like a normalized count. See this table on wikipedia for a list of options related to that. We’ll just pick one arbitrarily for use in this post. The point isn’t how we normalize the data, but what the SVD of $A = U \Sigma V^T$ means in this context. Recall that the domain of $A$, as a linear map, is a vector space whose dimension is the number of stories. We think of the vectors in this space as documents, or rather as an “embedding” of the abstract concept of a document using the counts of how often each word shows up in a document as a proxy for the semantic meaning of the document. Likewise, the codomain is the space of all words, and each word is embedded by which documents it occurs in. If we compare this to the movie rating example, it’s the same thing: a movie is the vector of ratings it receives from people, and a person is the vector of ratings of various movies. Say you take a rank 3 approximation to $A$. Then you get three singular vectors $v_1, v_2, v_3$ which form a basis for a subspace of words, i.e., the “idealized” words. These idealized words are your topics, and you can compute where a “new word” falls by looking at which documents it appears in (writing it as a vector in the domain) and saying its “topic” is the closest of the $v_1, v_2, v_3$. The same process applies to new documents. You can use this to cluster existing documents as well. The dataset we’ll use for this post is a relatively small corpus of a thousand CNN stories picked from 2012. Here’s an excerpt from one of them  cat data/cnn-stories/story479.txt
3 things to watch on Super Tuesday
Here are three things to watch for: Romney's big day. He's been the off-and-on frontrunner throughout the race, but a big Super Tuesday could begin an end game toward a sometimes hesitant base coalescing behind former Massachusetts Gov. Mitt Romney. Romney should win his home state of Massachusetts, neighboring Vermont and Virginia, ...


So let’s first build this document-term matrix, with the normalized values, and then we’ll compute it’s SVD and see what the topics look like.

Step 1 is cleaning the data. We used a bunch of routines from the nltk library that boils down to this loop:

    for filename, documentText in documentDict.items():
tokens = tokenize(documentText)
tagged_tokens = pos_tag(tokens)
wnl = WordNetLemmatizer()
stemmedTokens = [wnl.lemmatize(word, wordnetPos(tag)).lower()
for word, tag in tagged_tokens]


This turns the Super Tuesday story into a list of words (with repetition):

["thing", "watch", "three", "thing", "watch", "big", ... ]


If you’ll notice the name Romney doesn’t show up in the list of words. I’m only keeping the words that show up in the top 100,000 most common English words, and then lemmatizing all of the words to their roots. It’s not a perfect data cleaning job, but it’s simple and good enough for our purposes.

Now we can create the document term matrix.

def makeDocumentTermMatrix(data):
words = allWords(data)  # get the set of all unique words

wordToIndex = dict((word, i) for i, word in enumerate(words))
indexToWord = dict(enumerate(words))
indexToDocument = dict(enumerate(data))

matrix = np.zeros((len(words), len(data)))
for docID, document in enumerate(data):
docWords = Counter(document['words'])
for word, count in docWords.items():
matrix[wordToIndex[word], docID] = count

return matrix, (indexToWord, indexToDocument)


This creates a matrix with the raw integer counts. But what we need is a normalized count. The idea is that a common word like “thing” shows up disproportionately more often than “election,” and we don’t want raw magnitude of a word count to outweigh its semantic contribution to the classification. This is the applied math part of the algorithm design. So what we’ll do (and this technique together with SVD is called latent semantic indexing) is normalize each entry so that it measures both the frequency of a term in a document and the relative frequency of a term compared to the global frequency of that term. There are many ways to do this, and we’ll just pick one. See the github repository if you’re interested.

So now lets compute a rank 10 decomposition and see how to cluster the results.

    data = load()
matrix, (indexToWord, indexToDocument) = makeDocumentTermMatrix(data)
matrix = normalize(matrix)
sigma, U, V = svd(matrix, k=10)


This uses our svd, not numpy’s. Though numpy’s routine is much faster, it’s fun to see things work with code written from scratch. The result is too large to display here, but I can report the singular values.

>>> sigma
array([ 42.85249098,  21.85641975,  19.15989197,  16.2403354 ,
15.40456779,  14.3172779 ,  13.47860033,  13.23795002,
12.98866537,  12.51307445])


Now we take our original inputs and project them onto the subspace spanned by the singular vectors. This is the part that represents each word (resp., document) in terms of the idealized words (resp., documents), the singular vectors. Then we can apply a simple k-means clustering algorithm to the result, and observe the resulting clusters as documents.

    projectedDocuments = np.dot(matrix.T, U)
projectedWords = np.dot(matrix, V.T)

documentCenters, documentClustering = cluster(projectedDocuments)
wordCenters, wordClustering = cluster(projectedWords)

wordClusters = [
[indexToWord[i] for (i, x) in enumerate(wordClustering) if x == j]
for j in range(len(set(wordClustering)))
]

documentClusters = [
[indexToDocument[i]['text']
for (i, x) in enumerate(documentClustering) if x == j]
for j in range(len(set(documentClustering)))
]


And now we can inspect individual clusters. Right off the bat we can tell the clusters aren’t quite right simply by looking at the sizes of each cluster.

>>> Counter(wordClustering)
Counter({1: 9689, 2: 1051, 8: 680, 5: 557, 3: 321, 7: 225, 4: 174, 6: 124, 9: 123})
>>> Counter(documentClustering)
Counter({7: 407, 6: 109, 0: 102, 5: 87, 9: 85, 2: 65, 8: 55, 4: 47, 3: 23, 1: 15})


What looks wrong to me is the size of the largest word cluster. If we could group words by topic, then this is saying there’s a topic with over nine thousand words associated with it! Inspecting it even closer, it includes words like “vegan,” “skunk,” and “pope.” On the other hand, some word clusters are spot on. Examine, for example, the fifth cluster which includes words very clearly associated with crime stories.

>>> wordClusters[4]
['account', 'accuse', 'act', 'affiliate', 'allegation', 'allege', 'altercation', 'anything', 'apartment', 'arrest', 'arrive', 'assault', 'attorney', 'authority', 'bag', 'black', 'blood', 'boy', 'brother', 'bullet', 'candy', 'car', 'carry', 'case', 'charge', 'chief', 'child', 'claim', 'client', 'commit', 'community', 'contact', 'convenience', 'court', 'crime', 'criminal', 'cry', 'dead', 'deadly', 'death', 'defense', 'department', 'describe', 'detail', 'determine', 'dispatcher', 'district', 'document', 'enforcement', 'evidence', 'extremely', 'family', 'father', 'fear', 'fiancee', 'file', 'five', 'foot', 'friend', 'front', 'gate', 'girl', 'girlfriend', 'grand', 'ground', 'guilty', 'gun', 'gunman', 'gunshot', 'hand', 'happen', 'harm', 'head', 'hear', 'heard', 'hoodie', 'hour', 'house', 'identify', 'immediately', 'incident', 'information', 'injury', 'investigate', 'investigation', 'investigator', 'involve', 'judge', 'jury', 'justice', 'kid', 'killing', 'lawyer', 'legal', 'letter', 'life', 'local', 'man', 'men', 'mile', 'morning', 'mother', 'murder', 'near', 'nearby', 'neighbor', 'newspaper', 'night', 'nothing', 'office', 'officer', 'online', 'outside', 'parent', 'person', 'phone', 'police', 'post', 'prison', 'profile', 'prosecute', 'prosecution', 'prosecutor', 'pull', 'racial', 'racist', 'release', 'responsible', 'return', 'review', 'role', 'saw', 'scene', 'school', 'scream', 'search', 'sentence', 'serve', 'several', 'shoot', 'shooter', 'shooting', 'shot', 'slur', 'someone', 'son', 'sound', 'spark', 'speak', 'staff', 'stand', 'store', 'story', 'student', 'surveillance', 'suspect', 'suspicious', 'tape', 'teacher', 'teen', 'teenager', 'told', 'tragedy', 'trial', 'vehicle', 'victim', 'video', 'walk', 'watch', 'wear', 'whether', 'white', 'witness', 'young']


As sad as it makes me to see that ‘black’ and ‘slur’ and ‘racial’ appear in this category, it’s a reminder that naively using the output of a machine learning algorithm can perpetuate racism.

Here’s another interesting cluster corresponding to economic words:

>>> wordClusters[6]
['agreement', 'aide', 'analyst', 'approval', 'approve', 'austerity', 'average', 'bailout', 'beneficiary', 'benefit', 'bill', 'billion', 'break', 'broadband', 'budget', 'class', 'combine', 'committee', 'compromise', 'conference', 'congressional', 'contribution', 'core', 'cost', 'currently', 'cut', 'deal', 'debt', 'defender', 'deficit', 'doc', 'drop', 'economic', 'economy', 'employee', 'employer', 'erode', 'eurozone', 'expire', 'extend', 'extension', 'fee', 'finance', 'fiscal', 'fix', 'fully', 'fund', 'funding', 'game', 'generally', 'gleefully', 'growth', 'hamper', 'highlight', 'hike', 'hire', 'holiday', 'increase', 'indifferent', 'insistence', 'insurance', 'job', 'juncture', 'latter', 'legislation', 'loser', 'low', 'lower', 'majority', 'maximum', 'measure', 'middle', 'negotiation', 'offset', 'oppose', 'package', 'pass', 'patient', 'pay', 'payment', 'payroll', 'pension', 'plight', 'portray', 'priority', 'proposal', 'provision', 'rate', 'recession', 'recovery', 'reduce', 'reduction', 'reluctance', 'repercussion', 'rest', 'revenue', 'rich', 'roughly', 'sale', 'saving', 'scientist', 'separate', 'sharp', 'showdown', 'sign', 'specialist', 'spectrum', 'spending', 'strength', 'tax', 'tea', 'tentative', 'term', 'test', 'top', 'trillion', 'turnaround', 'unemployed', 'unemployment', 'union', 'wage', 'welfare', 'worker', 'worth']


One can also inspect the stories, though the clusters are harder to print out here. Interestingly the first cluster of documents are stories exclusively about Trayvon Martin. The second cluster is mostly international military conflicts. The third cluster also appears to be about international conflict, but what distinguishes it from the first cluster is that every story in the second cluster discusses Syria.

>>> len([x for x in documentClusters[1] if 'Syria' in x]) / len(documentClusters[1])
0.05555555555555555
>>> len([x for x in documentClusters[2] if 'Syria' in x]) / len(documentClusters[2])
1.0


Anyway, you can explore the data more at your leisure (and tinker with the parameters to improve it!).

## Issues with the power method

Though I mentioned that the power method isn’t an industry strength algorithm I didn’t say why. Let’s revisit that before we finish. The problem is that the convergence rate of even the 1-dimensional problem depends on the ratio of the first and second singular values, $\sigma_1 / \sigma_2$. If that ratio is very close to 1, then the convergence will take a long time and need many many matrix-vector multiplications.

One way to alleviate that is to do the trick where, to compute a large power of a matrix, you iteratively square $B$. But that requires computing a matrix square (instead of a bunch of matrix-vector products), and that requires a lot of time and memory if the matrix isn’t sparse. When the matrix is sparse, you can actually do the power method quite quickly, from what I’ve heard and read.

But nevertheless, the industry standard methods involve computing a particular matrix decomposition that is not only faster than the power method, but also numerically stable. That means that the algorithm’s runtime and accuracy doesn’t depend on slight changes in the entries of the input matrix. Indeed, you can have two matrices where $\sigma_1 / \sigma_2$ is very close to 1, but changing a single entry will make that ratio much larger. The power method depends on this, so it’s not numerically stable. But the industry standard technique is not. This technique involves something called Householder reflections. So while the power method was great for a proof of concept, there’s much more work to do if you want true SVD power.

Until next time!

# Hashing to Estimate the Size of a Stream

Problem: Estimate the number of distinct items in a data stream that is too large to fit in memory.

Solution: (in python)

import random

def randomHash(modulus):
a, b = random.randint(0,modulus-1), random.randint(0,modulus-1)
def f(x):
return (a*x + b) % modulus
return f

def average(L):
return sum(L) / len(L)

def numDistinctElements(stream, numParallelHashes=10):
modulus = 2**20
hashes = [randomHash(modulus) for _ in range(numParallelHashes)]
minima = [modulus] * numParallelHashes
currentEstimate = 0

for i in stream:
hashValues = [h(i) for h in hashes]
for i, newValue in enumerate(hashValues):
if newValue < minima[i]:
minima[i] = newValue

currentEstimate = modulus / average(minima)

yield currentEstimate


Discussion: The technique used here is to use random hash functions. The central idea is the same as the general principle presented in our recent post on hashing for load balancing. In particular, if you have an algorithm that works under the assumption that the data is uniformly random, then the same algorithm will work (up to a good approximation) if you process the data through a randomly chosen hash function.

So if we assume the data in the stream consists of $N$ uniformly random real numbers between zero and one, what we would do is the following. Maintain a single number $x_{\textup{min}}$ representing the minimum element in the list, and update it every time we encounter a smaller number in the stream. A simple probability calculation or an argument by symmetry shows that the expected value of the minimum is $1/(N+1)$. So your estimate would be $1/(x_{\textup{min}}+1)$. (The extra +1 does not change much as we’ll see.) One can spend some time thinking about the variance of this estimate (indeed, our earlier post is great guidance for how such a calculation would work), but since the data is not random we need to do more work. If the elements are actually integers between zero and $k$, then this estimate can be scaled by $k$ and everything basically works out the same.

Processing the data through a hash function $h$ chosen randomly from a 2-universal family (and we proved in the aforementioned post that this modulus thing is 2-universal) makes the outputs “essentially random” enough to have the above technique work with some small loss in accuracy. And to reduce variance, you can process the stream in parallel with many random hash functions. This rough sketch results in the code above. Indeed, before I state a formal theorem, let’s see the above code in action. First on truly random data:

S = [random.randint(1,2**20) for _ in range(10000)]

for k in range(10,301,10):
for est in numDistinctElements(S, k):
pass
print(abs(est))

# output
18299.75567190227
7940.7497160166595
12034.154552410098
12387.19432959244
15205.56844547564
8409.913113220158
8057.99978043693
9987.627098464103
10313.862295081966
9084.872639057356
10952.745228373375
10360.569781803211
11022.469475216301
9741.250165892501
11474.896038520465
10538.452261306533
10068.793492995934
10100.266495424627
9780.532155130093
8806.382800033594
10354.11482578643
10001.59202254498
10623.87031408308
9400.404915767062
10710.246772348424
10210.087633885101
9943.64709187974
10459.610972568578
10159.60175069326
9213.120899718839


As you can see the output is never off by more than a factor of 2. Now with “adversarial data.”

S = range(10000) #[random.randint(1,2**20) for _ in range(10000)]

for k in range(10,301,10):
for est in numDistinctElements(S, k):
pass
print(abs(est))

# output

12192.744186046511
15935.80547112462
10167.188106011634
12977.425742574258
6454.364151175674
7405.197740112994
11247.367453263867
4261.854392115023
8453.228233608026
7706.717624577393
7582.891328643745
5152.918628936483
1996.9365093316926
8319.20208545846
3259.0787592465967
6812.252720480753
4975.796789951151
8456.258064516129
8851.10133724288
7317.348220516398
10527.871485943775
3999.76974425661
3696.2999065091117
8308.843106180666
6740.999794281012
8468.603733730935
5728.532232608959
5822.072220349402
6382.349459544548
8734.008940222673


The estimates here are off by a factor of up to 5, and this estimate seems to get better as the number of hash functions used increases. The formal theorem is this:

Theorem: If $S$ is the set of distinct items in the stream and $n = |S|$ and $m > 100 n$, then with probability at least 2/3 the estimate $m / x_{\textup{min}}$ is between $n/6$ and $6n$.

We omit the proof (see below for references and better methods). As a quick analysis, since we’re only storing a constant number of integers at any given step, the algorithm has space requirement $O(\log m) = O(\log n)$, and each step takes time polynomial in $\log(m)$ to update in each step (since we have to compute multiplication and modulus of $m$).

This method is just the first ripple in a lake of research on this topic. The general area is called “streaming algorithms,” or “sublinear algorithms.” This particular problem, called cardinality estimation, is related to a family of problems called estimating frequency moments. The literature gets pretty involved in the various tradeoffs between space requirements and processing time per stream element.

As far as estimating cardinality goes, the first major results were due to Flajolet and Martin in 1983, where they provided a slightly more involved version of the above algorithm, which uses logarithmic space.

Later revisions to the algorithm (2003) got the space requirement down to $O(\log \log n)$, which is exponentially better than our solution. And further tweaks and analysis improved the variance bounds to something like a multiplicative factor of $\sqrt{m}$. This is called the HyperLogLog algorithm, and it has been tested in practice at Google.

Finally, a theoretically optimal algorithm (achieving an arbitrarily good estimate with logarithmic space) was presented and analyzed by Kane et al in 2010.

# Load Balancing and the Power of Hashing

Here’s a bit of folklore I often hear (and retell) that’s somewhere between a joke and deep wisdom: if you’re doing a software interview that involves some algorithms problem that seems hard, your best bet is to use hash tables.

More succinctly put: Google loves hash tables.

As someone with a passion for math and theoretical CS, it’s kind of silly and reductionist. But if you actually work with terabytes of data that can’t fit on a single machine, it also makes sense.

But to understand why hash tables are so applicable, you should have at least a fuzzy understanding of the math that goes into it, which is surprisingly unrelated to the actual act of hashing. Instead it’s the guarantees that a “random enough” hash provides that makes it so useful. The basic intuition is that if you have an algorithm that works well assuming the input data is completely random, then you can probably get a good guarantee by preprocessing the input by hashing.

In this post I’ll explain the details, and show the application to an important problem that one often faces in dealing with huge amounts of data: how to allocate resources efficiently (load balancing). As usual, all of the code used in the making of this post is available on Github.

Next week, I’ll follow this post up with another application of hashing to estimating the number of distinct items in a set that’s too large to store in memory.

## Families of Hash Functions

To emphasize which specific properties of hash functions are important for a given application, we start by introducing an abstraction: a hash function is just some computable function that accepts strings as input and produces numbers between 1 and $n$ as output. We call the set of allowed inputs $U$ (for “Universe”). A family of hash functions is just a set of possible hash functions to choose from. We’ll use a scripty $\mathscr{H}$ for our family, and so every hash function $h$ in $\mathscr{H}$ is a function $h : U \to \{ 1, \dots, n \}$.

You can use a single hash function $h$ to maintain an unordered set of objects in a computer. The reason this is a problem that needs solving is because if you were to store items sequentially in a list, and if you want to determine if a specific item is already in the list, you need to potentially check every item in the list (or do something fancier). In any event, without hashing you have to spend some non-negligible amount of time searching. With hashing, you can choose the location of an element $x \in U$ based on the value of its hash $h(x)$. If you pick your hash function well, then you’ll have very few collisions and can deal with them efficiently. The relevant section on Wikipedia has more about the various techniques to deal with collisions in hash tables specifically, but we want to move beyond that in this post.

Here we have a family of random hash functions. So what’s the use of having many hash functions? You can pick a hash randomly from a “good” family of hash functions. While this doesn’t seem so magical, it has the informal property that it makes arbitrary data “random enough,” so that an algorithm which you designed to work with truly random data will also work with the hashes of arbitrary data. Moreover, even if an adversary knows $\mathscr{H}$ and knows that you’re picking a hash function at random, there’s no way for the adversary to manufacture problems by feeding bad data. With overwhelming probability the worst-case scenario will not occur. Our first example of this is in load-balancing.

You can imagine load balancing in two ways, concretely and mathematically. In the concrete version you have a public-facing server that accepts requests from users, and forwards them to a back-end server which processes them and sends a response to the user. When you have a billion users and a million servers, you want to forward the requests in such a way that no server gets too many requests, or else the users will experience delays. Moreover, you’re worried that the League of Tanzanian Hackers is trying to take down your website by sending you requests in a carefully chosen order so as to screw up your load balancing algorithm.

The mathematical version of this problem usually goes with the metaphor of balls and bins. You have some collection of $m$ balls and $n$ bins in which to put the balls, and you want to put the balls into the bins. But there’s a twist: an adversary is throwing balls at you, and you have to put them into the bins before the next ball comes, so you don’t have time to remember (or count) how many balls are in each bin already. You only have time to do a small bit of mental arithmetic, sending ball $i$ to bin $f(i)$ where $f$ is some simple function. Moreover, whatever rule you pick for distributing the balls in the bins, the adversary knows it and will throw balls at you in the worst order possible.

A young man applying his knowledge of balls and bins. That’s totally what he’s doing.

There is one obvious approach: why not just pick a uniformly random bin for each ball? The problem here is that we need the choice to be persistent. That is, if the adversary throws the same ball at us a second time, we need to put it in the same bin as the first time, and it doesn’t count toward the overall load. This is where the ball/bin metaphor breaks down. In the request/server picture, there is data specific to each user stored on the back-end server between requests (a session), and you need to make sure that data is not lost for some reasonable period of time. And if we were to save a uniform random choice after each request, we’d need to store a number for every request, which is too much. In short, we need the mapping to be persistent, but we also want it to be “like random” in effect.

So what do you do? The idea is to take a “good” family of hash functions $\mathscr{H}$, pick one $h \in \mathscr{H}$ uniformly at random for the whole game, and when you get a request/ball $x \in U$ send it to server/bin $h(x)$. Note that in this case, the adversary knows your universal family $\mathscr{H}$ ahead of time, and it knows your algorithm of committing to some single randomly chosen $h \in \mathscr{H}$, but the adversary does not know which particular $h$ you chose.

The property of a family of hash functions that makes this strategy work is called 2-universality.

Definition: A family of functions $\mathscr{H}$ from some universe $U \to \{ 1, \dots, n \}$. is called 2-universal if, for every two distinct $x, y \in U$, the probability over the random choice of a hash function $h$ from $\mathscr{H}$ that $h(x) = h(y)$ is at most $1/n$. In notation,

$\displaystyle \Pr_{h \in \mathscr{H}}[h(x) = h(y)] \leq \frac{1}{n}$

I’ll give an example of such a family shortly, but let’s apply this to our load balancing problem. Our load-balancing algorithm would fail if, with even some modest probability, there is some server that receives many more than its fair share ($m/n$) of the $m$ requests. If $\mathscr{H}$ is 2-universal, then we can compute an upper bound on the expected load of a given server, say server 1. Specifically, pick any element $x$ which hashes to 1 under our randomly chosen $h$. Then we can compute an upper bound on the expected number of other elements that hash to 1. In this computation we’ll only use the fact that expectation splits over sums, and the definition of 2-universal. Call $\mathbf{1}_{h(y) = 1}$ the random variable which is zero when $h(y) \neq 1$ and one when $h(y) = 1$, and call $X = \sum_{y \in U} \mathbf{1}_{h(y) = 1}$. In words, $X$ simply represents the number of inputs that hash to 1. Then

So in expectation we can expect server 1 gets its fair share of requests. And clearly this doesn’t depend on the output hash being 1; it works for any server. There are two obvious questions.

1. How do we measure the risk that, despite the expectation we computed above, some server is overloaded?
2. If it seems like (1) is on track to happen, what can you do?

For 1 we’re asking to compute, for a given deviation $t$, the probability that $X - \mathbb{E}[X] > t$. This makes more sense if we jump to multiplicative factors, since it’s usually okay for a server to bear twice or three times its usual load, but not like $\sqrt{n}$ times more than it’s usual load. (Industry experts, please correct me if I’m wrong! I’m far from an expert on the practical details of load balancing.)

So we want to know what is the probability that $X - \mathbb{E}[X] > t \cdot \mathbb{E}[X]$ for some small number $t$, and we want this to get small quickly as $t$ grows. This is where the Chebyshev inequality becomes useful. For those who don’t want to click the link, for our sitauation Chebyshev’s inequality is the statement that, for any random variable $X$

$\displaystyle \Pr[|X - \mathbb{E}[X]| > t\mathbb{E}[X]] \leq \frac{\textup{Var}[X]}{t^2 \mathbb{E}^2[X]}.$

So all we need to do is compute the variance of the load of a server. It’s a bit of a hairy calculation to write down, but rest assured it doesn’t use anything fancier than the linearity of expectation and 2-universality. Let’s dive in. We start by writing the definition of variance as an expectation, and then we split $X$ up into its parts, expand the product and group the parts.

$\displaystyle \textup{Var}[X] = \mathbb{E}[(X - \mathbb{E}[X])^2] = \mathbb{E}[X^2] - (\mathbb{E}[X])^2$

The easy part is $(\mathbb{E}[X])^2$, it’s just $(1 + (m-1)/n)^2$, and the hard part is $\mathbb{E}[X^2]$. So let’s compute that

In order to continue (and get a reasonable bound) we need an additional property of our hash family which is not immediately spelled out by 2-universality. Specifically, we need that for every $h$ and $i$, $\Pr_x[h(x) = i] = O(\frac{1}{n})$. In other words, each hash function should evenly split the inputs across servers.

The reason this helps is because we can split $\Pr[h(x) = h(y) = 1]$  into $\Pr[h(x) = h(y) \mid h(x) = 1] \cdot \Pr[h(x) = 1]$. Using 2-universality to bound the left term, this quantity is at most $1/n^2$, and since there are $\binom{m}{2}$ total terms in the double sum above, the whole thing is at most $O(m/n + m^2 / n^2) = O(m^2 / n^2)$. Note that in our big-O analysis we’re assuming $m$ is much bigger than $n$.

Sweeping some of the details inside the big-O, this means that our variance is $O(m^2/n^2)$, and so our bound on the deviation of $X$ from its expectation by a multiplicative factor of $t$ is at most $O(1/t^2)$.

Now we computed a bound on the probability that a single server is not overloaded, but if we want to extend that to the worst-case server, the typical probability technique is to take the union bound over all servers. This means we just add up all the individual bounds and ignore how they relate. So the probability that none of the servers has a load more than a multiplicative factor of $t$ is at most $O(n/t^2)$. This is only less than one when $t = \Omega(\sqrt{n})$, so all we can say with this analysis is that (with some small constant probability) no server will have a load worse than $\sqrt{n}$ times more than the expected load.

So we have this analysis that seems not so good. If we have a million servers then the worst load on one server could potentially be a thousand times higher than the expected load. This doesn’t scale, and the problem could be in any (or all) of three places:

1. Our analysis is weak, and we should use tighter bounds because the true max load is actually much smaller.
2. Our hash families don’t have strong enough properties, and we should beef those up to get tighter bounds.
3. The whole algorithm sucks and needs to be improved.

It turns out all three are true. One heuristic solution is easy and avoids all math. Have some second server (which does not process requests) count hash collisions. When some server exceeds a factor of $t$ more than the expected load, send a message to the load balancer to randomly pick a new hash function from $\mathscr{H}$ and for any requests that don’t have existing sessions (this is included in the request data), use the new hash function. Once the old sessions expire, switch any new incoming requests from those IPs over to the new hash function.

But there are much better solutions out there. Unfortunately their analyses are too long for a blog post (they fill multiple research papers). Fortunately their descriptions and guarantees are easy to describe, and they’re easy to program. The basic idea goes by the name “the power of two choices,” which we explored on this blog in a completely different context of random graphs.

In more detail, the idea is that you start by picking two random hash functions $h_1, h_2 \in \mathscr{H}$, and when you get a new request, you compute both hashes, inspect the load of the two servers indexed by those hashes, and send the request to the server with the smaller load.

This has the disadvantage of requiring bidirectional talk between the load balancer and the server, rather than obliviously forwarding requests. But the advantage is an exponential decrease in the worst-case maximum load. In particular, the following theorem holds for the case where the hashes are fully random.

Theorem: Suppose one places $m$ balls into $n$ bins in order according to the following procedure: for each ball pick two uniformly random and independent integers $1 \leq i,j \leq n$, and place the ball into the bin with the smallest current size. If there are ties pick the bin with the smaller index. Then with high probability the largest bin has no more than $\Theta(m/n) + O(\log \log (n))$ balls.

This theorem appears to have been proved in a few different forms, with the best analysis being by Berenbrink et al. You can improve the constant on the $\log \log n$ by computing more than 2 hashes. How does this relate to a good family of hash functions, which is not quite fully random? Let’s explore the answer by implementing the algorithm in python.

## An example of universal hash functions, and the load balancing algorithm

In order to implement the load balancer, we need to have some good hash functions under our belt. We’ll go with the simplest example of a hash function that’s easy to prove nice properties for. Specifically each hash in our family just performs some arithmetic modulo a random prime.

Definition: Pick any prime $p > m$, and for any $1 \leq a < p$ and $0 \leq b \leq n$ define $h_{a,b}(x) = (ax + b \mod p) \mod m$. Let $\mathscr{H} = \{ h_{a,b} \mid 0 \leq b < p, 1 \leq a < p \}$.

This family of hash functions is 2-universal.

Theorem: For every $x \neq y \in \{0, \dots, p\}$,

$\Pr_{h \in \mathscr{H}}[h(x) = h(y)] \leq 1/p$

Proof. To say that $h(x) = h(y)$ is to say that $ax+b = ay+b + i \cdot m \mod p$ for some integer $i$. I.e., the two remainders of $ax+b$ and $ay+b$ are equivalent mod $m$. The $b$‘s cancel and we can solve for $a$

$a = im (x-y)^{-1} \mod p$

Since $a \neq 0$, there are $p-1$ possible choices for $a$. Moreover, there is no point to pick $i$ bigger than $p/m$ since we’re working modulo $p$. So there are $(p-1)/m$ possible values for the right hand side of the above equation. So if we chose them uniformly at random, (remember, $x-y$ is fixed ahead of time, so the only choice is $a, i$), then there is a $(p-1)/m$ out of $p-1$ chance that the equality holds, which is at most $1/m$. (To be exact you should account for taking a floor of $(p-1)/m$ when $m$ does not evenly divide $p-1$, but it only decreases the overall probability.)

$\square$

If $m$ and $p$ were equal then this would be even more trivial: it’s just the fact that there is a unique line passing through any two distinct points. While that’s obviously true from standard geometry, it is also true when you work with arithmetic modulo a prime. In fact, it works using arithmetic over any field.

Implementing these hash functions is easier than shooting fish in a barrel.

import random

def draw(p, m):
a = random.randint(1, p-1)
b = random.randint(0, p-1)

return lambda x: ((a*x + b) % p) % m


To encapsulate the process a little bit we implemented a UniversalHashFamily class which computes a random probable prime to use as the modulus and stores $m$. The interested reader can see the Github repository for more.

If we try to run this and feed in a large range of inputs, we can see how the outputs are distributed. In this example $m$ is a hundred thousand and $n$ is a hundred (it’s not two terabytes, but give me some slack it’s a demo and I’ve only got my desktop!). So the expected bin size for any 2-universal family is just about 1,000.

>>> m = 100000
>>> n = 100
>>> H = UniversalHashFamily(numBins=n, primeBounds=[n, 2*n])
>>> results = []
>>> for simulation in range(100):
...    bins = [0] * n
...    h = H.draw()
...    for i in range(m):
...       bins[h(i)] += 1
...    results.append(max(bins))
...
>>> max(bins) # a single run
1228
>>> min(bins)
613
>>> max(results) # the max bin size over all runs
1228
>>> min(results)
1227


Indeed, the max is very close to the expected value.

But this example is misleading, because the point of this was that some adversary would try to screw us over by picking a worst-case input. If the adversary knew exactly which $h$ was chosen (which it doesn’t) then the worst case input would be the set of all inputs that have the given hash output value. Let’s see it happen live.

>>> h = H.draw()
>>> badInputs = [i for i in range(m) if h(i) == 9]
1227
[0, 0, 0, 0, 0, 0, 0, 0, 0, 1227, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]


The expected size of a bin is 12, but as expected this is 100 times worse (linearly worse in $n$). But if we instead pick a random $h$ after the bad inputs are chosen, the result is much better.

>>> testInputs(n,m,badInputs) # randomly picks a hash
[19, 20, 20, 19, 18, 18, 17, 16, 16, 16, 16, 17, 18, 18, 19, 20, 20, 19, 18, 17, 17, 16, 16, 16, 16, 17, 18, 18, 19, 20, 20, 19, 18, 17, 17, 16, 16, 16, 16, 8, 8, 9, 9, 10, 10, 10, 10, 9, 9, 8, 8, 8, 8, 8, 8, 9, 9, 10, 10, 10, 10, 9, 9, 8, 8, 8, 8, 8, 8, 9, 9, 10, 10, 10, 10, 9, 8, 8, 8, 8, 8, 8, 8, 9, 9, 10, 10, 10, 10, 9, 8, 8, 8, 8, 8, 8, 8, 9, 9, 10]


However, if you re-ran this test many times, you’d eventually get unlucky and draw the hash function for which this actually is the worst input, and get a single huge bin. Other times you can get a bad hash in which two or three bins have all the inputs.

An interesting question is, what is really the worst-case input for this algorithm? I suspect it’s characterized by some choice of hash output values, taking all inputs for the chosen outputs. If this is the case, then there’s a tradeoff between the number of inputs you pick and how egregious the worst bin is. As an exercise to the reader, empirically estimate this tradeoff and find the best worst-case input for the adversary. Also, for your choice of parameters, estimate by simulation the probability that the max bin is three times larger than the expected value.

Now that we’ve played around with the basic hashing algorithm and made a family of 2-universal hashes, let’s see the power of two choices. Recall, this algorithm picks two random hash functions and sends an input to the bin with the smallest size. This obviously generalizes to $k$ choices, although the theoretical guarantee only improves by a constant factor, so let’s implement the more generic version.

class ChoiceHashFamily(object):
def __init__(self, hashFamily, queryBinSize, numChoices=2):
self.queryBinSize = queryBinSize
self.hashFamily = hashFamily
self.numChoices = numChoices

def draw(self):
hashes = [self.hashFamily.draw()
for _ in range(self.numChoices)]

def h(x):
indices = [h(x) for h in hashes]
counts = [self.queryBinSize(i) for i in indices]
count, index = min([(c,i) for (c,i) in zip(counts,indices)])
return index

return h


And if we test this with the bad inputs (as used previously, all the inputs that hash to 9), as a typical output we get

>>> bins
[15, 16, 15, 15, 16, 14, 16, 14, 16, 15, 16, 15, 15, 15, 17, 14, 16, 14, 16, 16, 15, 16, 15, 16, 15, 15, 17, 15, 16, 15, 15, 15, 15, 16, 15, 14, 16, 14, 16, 15, 15, 15, 14, 16, 15, 15, 15, 14, 17, 14, 15, 15, 14, 16, 13, 15, 14, 15, 15, 15, 14, 15, 13, 16, 14, 16, 15, 15, 15, 16, 15, 15, 13, 16, 14, 15, 15, 16, 14, 15, 15, 15, 11, 13, 11, 12, 13, 14, 13, 11, 11, 12, 14, 14, 13, 10, 16, 12, 14, 10]


And a typical list of bin maxima is

>>> results
[16, 16, 16, 18, 17, 365, 18, 16, 16, 365, 18, 17, 17, 17, 17, 16, 16, 17, 18, 16, 17, 18, 17, 16, 17, 17, 18, 16, 18, 17, 17, 17, 17, 18, 18, 17, 17, 16, 17, 365, 17, 18, 16, 16, 18, 17, 16, 18, 365, 16, 17, 17, 16, 16, 18, 17, 17, 17, 17, 17, 18, 16, 18, 16, 16, 18, 17, 17, 365, 16, 17, 17, 17, 17, 16, 17, 16, 17, 16, 16, 17, 17, 16, 365, 18, 16, 17, 17, 17, 17, 17, 18, 17, 17, 16, 18, 18, 17, 17, 17]


Those big bumps are the times when we picked an unlucky hash function, which is scarily large, although this bad event would be proportionally less likely as you scale up. But in the good case the load is clearly more even than the previous example, and the max load would get linearly smaller as you pick between a larger set of randomly chosen hashes (obviously).

Coupling this with the technique of switching hash functions when you start to observe a large deviation, and you have yourself an elegant solution.

In addition to load balancing, hashing has a ton of applications. Remember, the main key that you may want to use hashing is when you have an algorithm that works well when the input data is random. This comes up in streaming and sublinear algorithms, in data structure design and analysis, and many other places. We’ll be covering those applications in future posts on this blog.

Until then!