### Functional Programming

Abstract: The recent breakthroughs in deciphering the language and the literature left behind by the now extinct Twinklean civilization provides valuable insights into their history, science, and philosophy.

The oldest documents discovered on the third planet of the star Lambda Combinatoris (also known as the Twinkle star) talk about the prehistory of the Twinklean thought. The ancient Book of Application postulated that the Essence of Being is decomposition, expressed symbolically as

$A = B C$

meaning that $A$ can be decomposed into $B$ and $C$. The breakthrough came with the realization that, if $C$ itself can be decomposed

$C = F G$

then $A$ could be further decomposed into

$A = B (F G)$

Similarly, if $B$ can be decomposed

$B = D E$

then

$A = (D E) C$

In the latter case (but not the former), it became customary to drop the parentheses and simply write it as

$A = D E C$

Following these discoveries, the Twinklean civilization went through a period called The Great Decomposition that lasted almost three thousand years, during which essentially anything that could be decomposed was successfully decomposed.

At the end of The Great Decomposition, a new school of thought emerged, claiming that, if things can be decomposed into parts, they can be also recomposed from these parts.

Initially there was strong resistance to this idea. The argument was put forward that decomposition followed by recomposition doesn’t change anything. This was settled by the introduction of a special object called The Eye, denoted by $I,$ defined by the unique property of leaving things alone

$I A = A$

After the introduction of $I$, a long period of general stagnation accompanied by lack of change followed.

We also don’t have many records from the next period, as it was marked by attempts at forgetting things and promoting ignorance. It started by the introduction of $K$, which ignores one of its inputs

$K A B = A$

Notice that this definition is a shorthand for the parenthesized version

$(K A) B = A$

The argument for introducing $K$ was that ignorance is an important part of understanding. By rejecting $B$ we are saying that $A$ is important. We are abstracting away the inessential part $B$.

For instance—the argument went—if we decompose $C$

$C = A B$

and $D$ happens to have a similar decomposition

$D = A E$

then $K$ will abstract the $A$ part from both $C$ and $D$. From the perspective of $K$, there is no difference between $C$ and $D$.

The only positive outcome of the Era of Ignorance was the development of abstract mathematics. Twinklean thinkers argued that, if you disregard the particularities of the fruit in question, there is no difference between having three apples and three oranges. Number three was thus born, followed by many others (four and seven, to name just a few).

The final Industrial phase of the Twinklean civilization that ultimately led to their demise was marked by the introduction of $S$. The Twinklean industry was based on the principle of mass production; and mass production starts with duplication and reuse. Suppose you have a reusable part $C$. $S$ allows you to duplicate $C$ and combine it with both $A$ and $B$.

$S A B C = (A C) (B C)$

If you think of $A$ and $B$ as abstractions—that is the results of ignoring some parts of the whole—$S$ lets you substitute $C$ in place of those forgotten parts.

Or, conversely, it tells you that the object

$E = S A B C$

can be decomposed into two parts that have something in common. This common part is $C$.

Unfortunately, during the Industrial period, a lot of Twinkleans lost their identity. They discovered that

$I = S K K$

Indeed

$I A = S K K A = K A (K A) = A$

But ultimately, what precipitated their end was the existential crisis. They lost their will to live because they couldn’t figure out $Y$.

## Postscript

After submitting this paper to the journal of Compositionality, we have been informed by the reviewer that a similar theory of SKI combinators was independently developed on Earth by a Russian logician, Moses Schönfinkel. According to this reviewer, the answer to the meaning of life is the $Y$ combinator, which introduces recursion and can be expressed as

$Y = S(K(SII))(S(S(KS)K)(K(SII)))$

We were unable to verify this assertion, as it led us into a rabbit hole.

This post is based on the talk I gave in Moscow, Russia, in February 2015 to an audience of C++ programmers.

Let’s agree on some preliminaries.

C++ is a low level programming language. It’s very close to the machine. C++ is engineering at its grittiest.

Category theory is the most abstract branch of mathematics. It’s very very high in the layers of abstraction. Category theory is mathematics at its highest.

So why have I decided to speak about category theory to C++ programmers? There are many reasons.

The main reason is that category theory captures the essence of programming. We can program at many levels, and if I ask somebody “What is programming?” most C++ programmers will probably say that it’s about telling the computer what to do. How to move bytes from memory to the processor, how to manipulate them, and so on.

But there is another view of programming and it’s related to the human side of programming. We are humans writing programs. We decide what to tell the computer to do.

We are solving problems. We are finding solutions to problems and translating them in the language that is understandable to the computer.

But what is problem solving? How do we, humans, approach problem solving? It was only a recent development in our evolution that we have acquired these fantastic brains of ours. For hundreds of millions of years not much was happening under the hood, and suddenly we got this brain, and we used this brain to help us chase animals, shoot arrows, find mates, organize hunting parties, and so on. It’s been going on for a few hundred thousand years. And suddenly the same brain is supposed to solve problems in software engineering.

So how do we approach problem solving? There is one general approach that we humans have developed for problem solving. We had to develop it because of the limitations of our brain, not because of the limitations of computers or our tools. Our brains have this relatively small cache memory, so when we’re dealing with a huge problem, we have to split it into smaller parts. We have to decompose bigger problems into smaller problems. And this is very human. This is what we do. We decompose, and then we attack each problem separately, find the solution; and once we have solutions to all the smaller problems, we recompose them.

So the essence of programming is composition.

If we want to be good programmers, we have to understand composition. And who knows more about composing than musicians? They are the original composers!

So let me show you an example. This is a piece by Johann Sebastian Bach. I’ll show you two versions of this composition. One is low level, and one is high level.

The low level is just sampled sound. These are bytes that approximate the waveform of the sound.

And this is the same piece in typical music notation.

Which one is easier to manipulate? Which one is easier to reason about? Obviously, the high level one!

Notice that, in the high level language, we use a lot of different abstractions that can be processed separately. We split the problem into smaller parts. We know that there are things called notes, and they can be reproduced, in this particular case, using violins. There are also some letters like E, A, B7: these are chords. They describe harmony. There is melody, there is harmony, there is the bass line.

Musicians, when they compose music, use higher level abstractions. These higher level abstractions are easier to manipulate, reason about, and modify when necessary.

And this is probably what Bach was hearing in his head.

And he chose to represent it using the high level language of musical notation.

Now, if you’re a rap musician, you work with samples, and you learn how to manipulate the low level description of music. It’s a very different process. It’s much closer to low-level C++ programming. We often do copy and paste, just like rap musicians. There’s nothing wrong with that, but sometimes we would like to be more like Bach.

So how do we approach this problem as programmers and not as musicians. We cannot use musical notation to lift ourselves to higher levels of abstraction. We have to use mathematics. And there is one particular branch of mathematics, category theory, that is exactly about composition. If programming is about composition, then this is what we should be looking at.

Category theory, in general, is not easy to learn, but the basic concepts of category theory are embarrassingly simple. So I will talk about some of those embarrassingly simple concepts from category theory, and then explain how to use them in programming in some weird ways that would probably not have occurred to you when you’re programming.

## Categories

So what is this concept of a category? Two things: object and arrows between objects.

In category theory you don’t ask what these objects are. You call them objects, you give them names like A, B, C, D, etc., but you don’t ask what they are or what’s inside them. And then you have arrows that connect objects. Every arrow starts at some object and ends at some object. You can have many arrows going between two objects, or none whatsoever. Again, you don’t ask what these arrows are. You just give them names like f, g, h, etc.

And that’s it—that’s how you visualize a category: a bunch of objects and a bunch of arrows between them.

There are some operations on arrows and some laws that they have to obey, and they are also very simple.

Since composition is the essence of category theory (and of programming), we have to define composition in a category.

Whenever you have an arrow f going from object A to object B, here represented by two little piggies, and another arrow g going from object B to object C, there is an arrow called their composition, g ∘ f, that goes directly from object A to object C. We pronounce this “g after f.”

Composition is part of the definition of a category. Again, since we don’t know what these arrows are, we don’t ask what composition is. We just know that for any two composable arrows — such that the end of one coincides with the start of the other — there exists another arrow that’s their composition.

And this is exactly what we do when we solve problems. We find an arrow from A to B — that’s our subproblem. We find an arrow from B to C, that’s another subproblem. And then we compose them to get an arrow from A to C, and that’s a solution to our bigger problem. We can repeat this process, building larger and larger solutions by solving smaller problems and composing the solutions.

Notice that when we have three arrows to compose, there are two ways of doing that, depending on which pair we compose first. We don’t want composition to have history. We want to be able to say: This arrow is a composition of these three arrows: h after g after f, without having to use parentheses for grouping. That’s called associativity:

 (f ∘ g) ∘ h = f ∘ (g ∘ h)

Composition in a category must be associative.

And finally, every object has to have an identity arrow. It’s an arrow that goes from the object back to itself. You can have many arrows that loop back to the same object. But there is always one such loop for every object that is the identity with respect to composition.

It has the property that if you compose it with any other arrow that’s composable with it — meaning it either starts or ends at this object — you get that arrow back. It acts like multiplication by one. It’s an identity — it doesn’t change anything.

## Monoid

I can immediately give you an example of a very simple category that I’m sure you know very well and have used all your adult life. It’s called a monoid. It’s another embarrassingly simple concept. It’s a category that has only one object. It may have lots of arrows, but all these arrows have to start at this object and end at this object, so they are all composable. You can compose any two arrows in this category to get another arrow. And there is one arrow that’s the identity. When composed with any other arrow it will give you back the same arrow.

There are some very simple examples of monoids. We have natural numbers with addition and zero. An arrow corresponds to adding a number. For instance, you will have an arrow that corresponds to adding 5. You compose it with an arrow that corresponds to adding 3, and you get an arrow that corresponds to adding 8. Identity arrow corresponds to adding zero.

Multiplication forms a monoid too. The identity arrow corresponds to multiplying by 1. The composition rule for these arrows is just a multiplication table.

Strings form another interesting monoid. An arrow corresponds to appending a particular string. Unit arrow appends an empty string. What’s interesting about this monoid is that it has no additional structure. In particular, it doesn’t have an inverse for any of its arrows. There are no “negative” strings. There is no anti-“world” string that, when appended to “Hello world”, would result in the string “Hello“.

In each of these monoids, you can think of the one object as being a set: a set of all numbers, or a set of all strings. But that’s just an aid to imagination. All information about the monoid is in the composition rules — the multiplication table for arrows.

In programming we encounter monoids all over the place. We just normally don’t call them that. But every time you have something like logging, gathering data, or auditing, you are using a monoid structure. You’re basically adding some information to a log, appending, or concatenating, so that’s a monoidal operation. And there is an identity log entry that you may use when you have nothing interesting to add.

## Types and Functions

So monoid is one example, but there is something closer to our hearts as programmers, and that’s the category of types and functions. And the funny thing is that this category of types and functions is actually almost enough to do programming, and in functional languages that’s what people do. In C++ there is a little bit more noise, so it’s harder to abstract this part of programming, but we do have types — it’s a strongly typed language, modulo implicit conversions. And we do have functions. So let’s see why this is a category and how it’s used in programming.

This category is actually called Set — a category of sets — because, to the lowest approximation, types are just sets of values. The type bool is a set of two values, true and false. The type int is a set of integers from something like negative two billion to two billion (on a 32-bit machine). All types are sets: whether it’s numbers, enums, structs, or objects of a class. There could be an infinite set of possible values, but it’s okay — a set may be infinite. And functions are just mappings between these sets. I’m talking about the simplest functions, ones that take just one value of some type and return another value of another type. So these are arrows from one type to another.

Can we compose these functions? Of course we can. We do it all the time. We call one function, it returns some value, and with this value we call another function. That’s function composition. In fact this is the basis of procedural decomposition, the first serious approach to formalizing problem solving in programming.

Here’s a piece of C++ code that composes two functions f and g.

C g_after_f(A x) {
B y = f(x);
return g(y);
}

In modern C++ you can make this code generic — a higher order function that accepts two functions and returns a third function that’s the composition of the two.

Can you compose any two functions? Yes — if they are composable. The output type of one must match the input type of another. That’s the essence of strong typing in C++ (modulo implicit conversions).

Is there an identity function? Well, in C++ we don’t have an identity function in the library, which is too bad. That’s because there’s a complex issue of how you pass things: is it by value, by reference, by const reference, by move, and so on. But in functional languages there is just one function called identity. It takes an argument and returns it back. But even in C++, if you limit yourself to functions that take arguments by value and return values, then it’s very easy to define a generic identity function.

Notice that the functions I’m talking about are actually special kind of functions called pure functions. They can’t have any side effects. Mathematically, a function is just a mapping from one set to another set, so it can’t have side effects. Also, a pure function must return the same value when called with the same arguments. This is called referential transparency.

A pure function doesn’t have any memory or state. It doesn’t have static variables, and doesn’t use globals. A pure function is an ideal we strive towards in programming, especially when writing reusable components and libraries. We don’t like having global variables, and we don’t like state hidden in static variables.

Moreover, if a function is pure, you can memoize it. If a function takes a long time to evaluate, maybe you’ll want to cache the value, so it can be retrieved quickly next time you call it with the same arguments.

Another property of pure functions is that all dependencies in your code only come through composition. If the result of one function is used as an argument to another then obviously you can’t run them in parallel or reverse the order of execution. You have to call them in that particular order. You have to sequence their execution. The dependencies between functions are fully explicit. This is not true for functions that have side effects. They may look like independent functions, but they have to be executed in sequence, or their side effects will be different.

We know that compiler optimizers will try to rearrange our code, but it’s very hard to do it in C++ because of hidden dependencies. If you have two functions that are not composed, they just calculate different things, and you try to call them in a different order, you might get a completely different result. It’s because of the order of side effects, which are invisible to the compiler. You would have to go deep into the implementation of the functions; you would have to analyse everything they are doing, and the functions they are calling, and so on, in order to find out what these side effects are; and only then you could decide: Oh, I can swap these two functions.

In functional programming, where you only deal with pure functions, you can swap any two functions that are not explicitly composed, and composition is immediately visible.

At this point I would expect half of the audience to leave and say: “You can’t program with pure functions, Programming is all about side effects.” And it’s true. So in order to keep you here I will have to explain how to deal with side effects. But it’s important that you start with something that is easy to understand, something you can reason about, like pure functions, and then build side effects on top of these things, so you can build up abstractions on top of other abstractions.

You start with pure functions and then you talk about side effects, not the other way around.

## Auditing

Instead of explaining the general theory of side effects in category theory, I’ll give you an example from programming. Let’s solve this simple problem that, in all likelihood, most C++ programmers would solve using side effects. It’s about auditing.

You start with a sequence of functions that you want to compose. For instance, you have a function getKey. You give it a password and it returns a key. And you have another function, withdraw. You give it a key and gives you back money. You want to compose these two functions, so you start with a password and you get money. Excellent!

But now you have a new requirement: you want to have an audit trail. Every time one of these functions is called, you want to log something in the audit trail, so that you’ll know what things have happened and in what order. That’s a side effect, right?

How do we solve this problem? Well, how about creating a global variable to store the audit trail? That’s the simplest solution that comes to mind. And it’s exactly the same method that’s used for standard output in C++, with the global object std::cout. The functions that access a global variable are obviously not pure functions, we are talking about side effects.

string audit;

audit += passwd;
return 42;
}

double withdraw(int key){
audit += “withdrawing ”;
return 100.0;
}

So we have a string, audit, it’s a global variable, and in each of these functions we access this global variable and append something to it. For simplicity, I’m just returning some fake numbers, not to complicate things.

This is not a good solution, for many reasons. It doesn’t scale very well. It’s difficult to maintain. If you want to change the name of the variable, you’d have to go through all this code and modify it. And if, at some point, you decide you want to log more information, not just a string but maybe a timestamp as well, then you have to go through all this code again and modify everything. And I’m not even mentioning concurrency. So this is not the best solution.

But there is another solution that’s really pure. It’s based on the idea that whatever you’re accessing in a function, you should pass explicitly to it, and then return it, with modifications, from the function. That’s pure. So here’s the next solution.

pair<int, string>
return make_pair(42, audit + passwd);
}

pair<double, string>
withdraw(int key, string audit){
return make_pair(100.0
, audit + “withdrawing ”);
}

You modify all the functions so that they take an additional argument, the audit string. And the return type is also changed. When we had an int before, it’s now a pair of int and string. When we had a double before, it’s now a pair of double and string. These function now call make_pair before they return, and they put in whatever they were returning before, plus they do this concatenation of a new message at the end of the old audit string. This is a better solution because it uses pure functions. They only depend on their arguments. They don’t have any state, they don’t access any global variables. Every time you call them with the same arguments, they produce the same result.

The problem though is that they don’t memoize that well. Look at the function logIn: you normally get the same key for the same password. But if you want to memoize it when it takes two arguments, you suddenly have to memoize it for all possible histories. Even if you call it with the same password, but the audit string is different, you can’t just access the cache, you have to cache a new pair of values. Your cache explodes with all possible histories.

An even bigger problem is security. Each of these functions has access to the complete log, including the passwords.

Also, each of these functions has to care about things that maybe it shouldn’t be bothered with. It knows about how to concatenate strings. It knows the details of the implementation of the log: that the log is a string. It must know how to accumulate the log.

Now I want to show you a solution that maybe is not that obvious, maybe a little outside of what we would normally think of.

pair<int, string>
return make_pair(42, passwd);
}

pair<double, string>
withdraw(int key){
return make_pair(100.0
,“withdrawing ”);
}

We use modified functions, but they don’t take the audit string any more. They just return a pair of whatever they were returning before, plus a string. But each of them only creates a message about what it considers important. It doesn’t have access to any log and it doesn’t know how to work with an audit trail. It’s just doing its local thing. It’s only responsible for its local data. It’s not responsible for concatenation.

It still creates a pair and it has a modified return type.

We have one problem though: we don’t know how to compose these functions. We can’t pass a pair of key and string from logIn to withdraw, because withdraw expects an int. Of course we could extract the int and drop the string, but that would defeat the goal of auditing the code.

Let’s go back a little bit and see how we can abstract this thing. We have functions that used to return some types, and now they return pairs of the original type and a string. This should in principle work with any original type, not just an int or a double. In functional programming we call this “lifting.” Here we lift some type A to a new type, which is a pair of A and a string. Or we can say that we are “embellishing” the return type of a function by pairing it with a string.

I’ll create an alias for this new parameterised type and call it Writer.

template<class A>
using Writer = pair<A, string>;

My functions now return Writers: logIn returns a writer of int, and withdraw returns a writer of double. They return “embellished” types.

Writer<int> logIn(string passwd){
return make_pair(42, passwd);
}

Writer<double> withdraw(int key){
return make_pair(100.0, “withdrawing ”);
} 

So how do we compose these embellished functions?

In this case we want to compose logIn with withdraw to create a new function called transact. This new function transact will take a password, log the user in, withdraw money, and return the money plus the audit trail. But it will return the audit trail only from those two functions.

Writer<double> transact(string passwd){
auto p2 withdraw(p1.first);
return make_pair(p2.first
, p1.second + p2.second);
}

How is it done? It’s very simple. I call the first function, logIn, with the password. It returns a pair of key and string. Then I call the second function, passing it the first component of the pair — the key. I get a new pair with the money and a string. And then I perform the composition. I take the money, which is the first component of the second pair, and I pair it with the concatenation of the two string that were the second components of the pairs returned by logIn and withdraw.

So the accumulation of the log is done “in between” the calls (think of composition as happening between calls). I have these two functions, and I’m composing them in this funny way that involves the concatenation of strings. The accumulation of the log does not happen inside these two functions, as it happened before. It happens outside. And I can pull out this code and abstract the composition. It doesn’t really matter what functions I’m calling. I can do it for any two functions that return embellished results. I can write generic code that does it and I can call it “compose”.

template<class A, class B, class C>
function<Writer<C>(A)> compose(function<Writer<B>(A)> f
,function<Writer<C>(B)> g)
{
return [f, g](A x) {
auto p1 = f(x);
auto p2 = g(p1.first);
return make_pair(p2.first
, p1.second + p2.second);
};
}

What does compose do? It takes two functions. The first function takes A and returns a Writer of B. The second function takes a B and return a Writer of C. When I compose them, I get a function that takes an A and returns a Writer of C.

This higher order function just does the composition. It has no idea that there are functions like logIn or withdraw, or any other functions that I may come up with later. It takes two embellished functions and glues them together.

We’re lucky that in modern C++ we can work with higher order functions that take functions as arguments and return other functions.

This is how I would implement the transact function using compose.

Writer<double> transact(string passwd){
return compose<string, int, double>
}

The transact function is nothing but the composition of logIn and withdraw. It doesn’t contain any other logic. I’m using this special composition because I want to create an audit trail. And the audit trail is accumulated “between” the calls — it’s in the glue that glues these two functions together.

This particular implementation of compose requires explicit type annotations, which is kind of ugly. We would like the types to be inferred. And you can do it in C++14 using generalised lambdas with return type deduction. This code was contributed by Eric Niebler.

auto const compose = [](auto f, auto g) {
return [f, g](auto x) {
auto p1 = f(x);
auto p2 = g(p1.first);
return make_pair(p2.first
, p1.second + p2.second);
};
};
Writer<double> transact(string passwd){
}

## Back to Categories

Now that we’ve done this example, let’s go back to where we started. In category theory we have functions and we have composition of functions. Here we also have functions and composition, but it’s a funny composition. We have functions that take simple types, but they return embellished types. The types don’t match.

Let me remind you what we had before. We had a category of types and pure functions with the obvious composition.

• Objects: types,
• Arrows: pure functions,
• Composition: pass the result of one function as the argument to another.

What we have created just now is a different category. Slightly different. It’s a category of embellished functions. Objects are still types: Types A, B, C, like integers, doubles, strings, etc. But an arrow from A to B is not a function from type A to type B. It’s a function from type A to the embellishment of the type B. The embellished type depends on the type B — in our case it was a pair type that combined B and a string — the Writer of B.

Now we have to say how to compose these arrows. It’s not as trivial as it was before. We have one arrow that takes A into a pair of B and string, and we have another arrow that takes B into a pair of C and string, and the composition should take an A and return a pair of C and string. And I have just defined this composition. I wrote code that does this:

auto const compose = [](auto f, auto g) {
return [f, g](auto x) {
auto p1 = f(x);
auto p2 = g(p1.first);
return make_pair(p2.first
, p1.second + p2.second);
};
};

So do we have a category here? A category that’s different from the original category? Yes, we do! It has composition and it has identity.

What’s its identity? It has to be an arrow from the object to itself, from A to A. But an arrow from A to A is a function from A to a pair of A and string — to a Writer of A. Can we implement something like this? Yes, easily. We will return a pair that contains the original value and the empty string. The empty string will not contribute to our audit trail.

template<class A>
Writer<A> identity(A x) {
return make_pair(x, "");
}

Is this composition associative? Yes, it is, because the underlying composition is associative, and the concatenation of strings is associative.

We have a new category. We have incorporated side effects by modifying the original category. We are still only using pure functions and yet we are able to accumulate an audit trail as a side effect. And we moved the side effects to the definition of composition.

It’s a funny new way of looking at programming. We usually see the functions, and the data being passed between functions, and here suddenly we see a new dimension to programming that is orthogonal to this, and we can manipulate it. We change the way we compose functions. We have this new power to change composition. We have a new way of solving problems by moving to these embellished functions and defining a new way of composing them. We can define new combinators to compose functions, and we’ll let the combinators do some work that we don’t want these functions to do. We can factor these things out and make them orthogonal.

Does this approach generalize?

One easy generalisation is to observe that the Writer structure works for any monoid. It doesn’t have to be just strings. Look at how composition and identity are defined in our new cateogory. The only properties of the log we are using are concatenation and unit. Concatenation must be associative for the composition to be associative. And we need a unit of concatenation so that we can define identity in our category. We don’t need anything else. This construction will work with any monoid.

And that’s great because you have one more dimension in which you can modify your code without touching the rest. You can change the format of the log, and all you need to modify in your code is compose and identity. You don’t have to go through all your functions and modify the code. They will still work because all the concatenation of logs is done inside compose.

## Kleisli Categories

This was just a little taste of what is possible with category theory. The thing I called embellishment is called a functor in category theory. You can implement categorical functors in C++. There are all kinds of embellishments/functors that you can use here. And now I can tell you the secret: this funny composition of functions with the funny identity is really a monad in disguise. A monad is just a funny way of composing embellished functions so that they form a category. A category based on a monad is called a Kleisli category.

Are there any other interesting monads that I can use this construction with? Yes, lots! I’ll give you one example. Functions that return futures. That’s our new embellishment. Give me any type A and I will embellish it by making it into a future. This embellishment also produces a Kleisli category. The composition of functions that return futures is done through the combinator “then”. You call one function returning a future and compose it with another function returning a future by passing it to “then.” You can compose these function into chains without ever having to block for a thread to finish. And there is an identity, which is a function that returns a trivial future that’s always ready. It’s called make_ready_future. It’s an arrow that takes A and returns a future of A.

Now you understand what’s really happening. We are creating this new category based on future being a monad. We have new words to describe what we are doing. We are reusing an idea from category theory to solve a completely different problem.

## Resumable Functions

There is one little invonvenience with this approach. It requires writing a lot of so called “boilerplate” code. Repetitive code that obscures the simple logic. Here it’s the glue code, the “compose” and the “then.” What you’d like to do is to write your code directly in terms of embellished function, and the composition to be implicit. People noticed this and came up with solutions. In case of futures, the practical solution is called resumable functions.

Resumable functions are designed to hide the composition of functions that return futures. Here’s an example.

int cnt = 0;
do
{
if ( cnt == 0 ) break;
cnt = await streamW.write(cnt, buf);
} while (cnt > 0);

This code copies a file using a buffer, but it does it asynchronously. We call a function read that’s asynchronous. It doesn’t immediately fill the buffer, it returns a future instead. Then we call the function write that’s also asynchronous. We do it in a loop.

This code looks almost like sequential code, except that it has these await keywords. These are the points of insertion of our composition. These are the places where the code is chopped into pieces and composed using then.

I won’t go into details of the implementation. The point is that the composition of these embellished functions is almost entirely hidden. It doesn’t look like composition in a Kleisli category, but it really is.

This solution is usually described at a very low level, in terms of coroutines implemented as state machines with static variables and gotos. And what is being lost in all this engineering talk is how general this idea is — the idea of overloading composition to build a category of embellished functions.

Just to drive this home, here’s an example of different code that does completely different stuff. It calculates Fibonacci numbers on demand. It’s a generator of Fibonacci numbers.

generator<int> fib()
{
int a = 0;
int b = 1;
for (;;) {
__yield_value a;
auto next = a + b;
a = b;
b = next;
}
} 

Instead of await it has __yield_value. But it’s the same idea of resumable functions, only with a different monad. This monad is called a list monad. And this kind of code in combination with Eric Niebler’s proposed range library could lead to very powerful programming idioms.

## Conclusion

Why do we have to separate the two notions: that of resumable functions and that of generators, if they are based on the same abstraction? Why do we have to reinvent the wheel?

There’s this great opportunity for C++, and I’m afraid it will be missed like so many other opportunities for great generalisations that were missed in the past. It’s the opportunity to introduce one general solution based on monads, rather than keep creating ad-hoc solutions, one problem at a time. The same very general pattern can be used to control all kinds of side effects. It can be used for auditing, exceptions, ranges, futures, I/O, continuations, and all kinds of user-defined monads.

This amazing power could be ours if we start thinking in more abstract terms, if we reach into category theory.

Yes, it’s this time of the year again! I started a little tradition a year ago with Stalking a Hylomorphism in the Wild. This year I was reminded of the Advent of Code by a tweet with this succint C++ program:

This piece of code is probably unreadable to a regular C++ programmer, but makes perfect sense to a Haskell programmer.

Here’s the description of the problem: You are given a list of equal-length strings. Every string is different, but two of these strings differ only by one character. Find these two strings and return their matching part. For instance, if the two strings were “abcd” and “abxd”, you would return “abd”.

What makes this problem particularly interesting is its potential application to a much more practical task of matching strands of DNA while looking for mutations. I decided to explore the problem a little beyond the brute force approach. And, of course, I had a hunch that I might encounter my favorite wild beast–the hylomorphism.

## Brute force approach

First things first. Let’s do the boring stuff: read the file and split it into lines, which are the strings we are supposed to process. So here it is:

main = do
let cs = lines txt
print $findMatch cs The real work is done by the function findMatch, which takes a list of strings and produces the answer, which is a single string. findMatch :: [String] -> String First, let’s define a function that calculates the distance between any two strings. distance :: (String, String) -> Int We’ll define the distance as the count of mismatched characters. Here’s the idea: We have to compare strings (which, let me remind you, are of equal length) character by character. Strings are lists of characters. The first step is to take two strings and zip them together, producing a list of pairs of characters. In fact we can combine the zipping with the next operation–in this case, comparison for inequality, (/=)–using the library function zipWith. However, zipWith is defined to act on two lists, and we will want it to act on a pair of lists–a subtle distinction, which can be easily overcome by applying uncurry: uncurry :: (a -> b -> c) -> ((a, b) -> c) which turns a function of two arguments into a function that takes a pair. Here’s how we use it: uncurry (zipWith (/=)) The comparison operator (/=) produces a Boolean result, True or False. We want to count the number of differences, so we’ll covert True to one, and False to zero: fromBool :: Num a => Bool -> a fromBool False = 0 fromBool True = 1 (Notice that such subtleties as the difference between Bool and Int are blisfully ignored in C++.) Finally, we’ll sum all the ones using sum. Altogether we have: distance = sum . fmap fromBool . uncurry (zipWith (/=))  Now that we know how to find the distance between any two strings, we’ll just apply it to all possible pairs of strings. To generate all pairs, we’ll use list comprehension: let ps = [(s1, s2) | s1 <- ss, s2 <- ss] (In C++ code, this was done by cartesian_product.) Our goal is to find the pair whose distance is exactly one. To this end, we’ll apply the appropriate filter: filter ((== 1) . distance) ps For our purposes, we’ll assume that there is exactly one such pair (if there isn’t one, we are willing to let the program fail with a fatal exception). (s, s') = head$ filter ((== 1) . distance) ps

The final step is to remove the mismatched character:

filter (uncurry (==)) $zip s s' We use our friend uncurry again, because the equality operator (==) expects two arguments, and we are calling it with a pair of arguments. The result of filtering is a list of identical pairs. We’ll fmap fst to pick the first components. findMatch :: [String] -> String findMatch ss = let ps = [(s1, s2) | s1 <- ss, s2 <- ss] (s, s') = head$ filter ((== 1) . distance) ps
in fmap fst $filter (uncurry (==))$ zip s s'

This program produces the correct result and we could stop right here. But that wouldn’t be much fun, would it? Besides, it’s possible that other algorithms could perform better, or be more flexible when applied to a more general problem.

## Data-driven approach

The main problem with our brute-force approach is that we are comparing everything with everything. As we increase the number of input strings, the number of comparisons grows like a factorial. There is a standard way of cutting down on the number of comparison: organizing the input into a neat data structure.

We are comparing strings, which are lists of characters, and list comparison is done recursively. Assume that you know that two strings share a prefix. Compare the next character. If it’s equal in both strings, recurse. If it’s not, we have a single character fault. The rest of the two strings must now match perfectly to be considered a solution. So the best data structure for this kind of algorithm should batch together strings with equal prefixes. Such a data structure is called a prefix tree, or a trie (pronounced try).

At every level of our prefix tree we’ll branch based on the current character (so the maximum branching factor is, in our case, 26). We’ll record the character, the count of strings that share the prefix that led us there, and the child trie storing all the suffixes.

data Trie = Trie [(Char, Int, Trie)]
deriving (Show, Eq)

Here’s an example of a trie that stores just two strings, “abcd” and “abxd”. It branches after b.

   a 2
b 2
c 1    x 1
d 1    d 1

When inserting a string into a trie, we recurse both on the characters of the string and the list of branches. When we find a branch with the matching character, we increment its count and insert the rest of the string into its child trie. If we run out of branches, we create a new one based on the current character, give it the count one, and the child trie with the rest of the string:

insertS :: Trie -> String -> Trie
insertS t "" = t
insertS (Trie bs) s = Trie (inS bs s)
where
inS ((x, n, t) : bs) (c : cs) =
if c == x
then (c, n + 1, insertS t cs) : bs
else (x, n, t) : inS bs (c : cs)
inS [] (c : cs) = [(c, 1, insertS (Trie []) cs)]

We convert our input to a trie by inserting all the strings into an (initially empty) trie:

mkTrie :: [String] -> Trie
mkTrie = foldl insertS (Trie [])

Of course, there are many optimizations we could use, if we were to run this algorithm on big data. For instance, we could compress the branches as is done in radix trees, or we could sort the branches alphabetically. I won’t do it here.

I won’t pretend that this implementation is simple and elegant. And it will get even worse before it gets better. The problem is that we are dealing explicitly with recursion in multiple dimensions. We recurse over the input string, the list of branches at each node, as well as the child trie. That’s a lot of recursion to keep track of–all at once.

Now brace yourself: We have to traverse the trie starting from the root. At every branch we check the prefix count: if it’s greater than one, we have more than one string going down, so we recurse into the child trie. But there is also another possibility: we can allow to have a mismatch at the current level. The current characters may be different but, since we allow only one mismatch, the rest of the strings have to match exactly. That’s what the function exact does. Notice that exact t is used inside foldMap, which is a version of fold that works on monoids–here, on strings.

match1 :: Trie -> [String]
match1 (Trie bs) = go bs
where
go :: [(Char, Int, Trie)] -> [String]
go ((x, n, t) : bs) =
let a1s = if n > 1
then fmap (x:) $match1 t else [] a2s = foldMap (exact t) bs a3s = go bs -- recurse over list in a1s ++ a2s ++ a3s go [] = [] exact t (_, _, t') = matchAll t t' Here’s the function that finds all exact matches between two tries. It does it by generating all pairs of branches in which top characters match, and then recursing down. matchAll :: Trie -> Trie -> [String] matchAll (Trie bs) (Trie bs') = mAll bs bs' where mAll :: [(Char, Int, Trie)] -> [(Char, Int, Trie)] -> [String] mAll [] [] = [""] mAll bs bs' = let ps = [ (c, t, t') | (c, _, t) <- bs , (c', _', t') <- bs' , c == c'] in foldMap go ps go (c, t, t') = fmap (c:) (matchAll t t') When mAll reaches the leaves of the trie, it returns a singleton list containing an empty string. Subsequent actions of fmap (c:) will prepend characters to this string. Since we are expecting exactly one solution to the problem, we’ll extract it using head: findMatch1 :: [String] -> String findMatch1 cs = head$ match1 (mkTrie cs)

## Recursion schemes

As you hone your functional programming skills, you realize that explicit recursion is to be avoided at all cost. There is a small number of recursive patterns that have been codified, and they can be used to solve the majority of recursion problems (for some categorical background, see F-Algebras). Recursion itself can be expressed in Haskell as a data structure: a fixed point of a functor:

newtype Fix f = In { out :: f (Fix f) }

In particular, our trie can be generated from the following functor:

data TrieF a = TrieF [(Char, a)]
deriving (Show, Functor)

Notice how I have replaced the recursive call to the Trie type constructor with the free type variable a. The functor in question defines the structure of a single node, leaving holes marked by the occurrences of a for the recursion. When these holes are filled with full blown tries, as in the definition of the fixed point, we recover the complete trie.

I have also made one more simplification by getting rid of the Int in every node. This is because, in the recursion scheme I’m going to use, the folding of the trie proceeds bottom-up, rather than top-down, so the multiplicity information can be passed upwards.

The main advantage of recursion schemes is that they let us use simpler, non-recursive building blocks such as algebras and coalgebras. Let’s start with a simple coalgebra that lets us build a trie from a list of strings. A coalgebra is a fancy name for a particular type of function:

type Coalgebra f x = x -> f x

Think of x as a type for a seed from which one can grow a tree. A colagebra tells us how to use this seed to create a single node described by the functor f and populate it with (presumably smaller) seeds. We can then pass this coalgebra to a simple algorithm, which will recursively expand the seeds. This algorithm is called the anamorphism:

ana :: Functor f => Coalgebra f a -> a -> Fix f
ana coa = In . fmap (ana coa) . coa

Let’s see how we can apply it to the task of building a trie. The seed in our case is a list of strings (as per the definition of our problem, we’ll assume they are all equal length). We start by grouping these strings into bunches of strings that start with the same character. There is a library function called groupWith that does exactly that. We have to import the right library:

import GHC.Exts (groupWith)

This is the signature of the function:

groupWith :: Ord b => (a -> b) -> [a] -> [[a]]

It takes a function a -> b that converts each list element to a type that supports comparison (as per the typeclass Ord), and partitions the input into lists that compare equal under this particular ordering. In our case, we are going to extract the first character from a string using head and bunch together all strings that share that first character.

let sss = groupWith head ss

The tails of those strings will serve as seeds for the next tier of the trie. Eventually the strings will be shortened to nothing, triggering the end of recursion.

fromList :: Coalgebra TrieF [String]
fromList ss =
-- are strings empty? (checking one is enough)
then TrieF [] -- leaf
else
let sss = groupWith head ss
in TrieF $fmap mkBranch sss The function mkBranch takes a bunch of strings sharing the same first character and creates a branch seeded with the suffixes of those strings. mkBranch :: [String] -> (Char, [String]) mkBranch sss = let c = head (head sss) -- they're all the same in (c, fmap tail sss) Notice that we have completely avoided explicit recursion. The next step is a little harder. We have to fold the trie. Again, all we have to define is a step that folds a single node whose children have already been folded. This step is defined by an algebra: type Algebra f x = f x -> x Just as the type x described the seed in a coalgebra, here it describes the accumulator–the result of the folding of a recursive data structure. We pass this algebra to a special algorithm called a catamorphism that takes care of the recursion: cata :: Functor f => Algebra f a -> Fix f -> a cata alg = alg . fmap (cata alg) . out Notice that the folding proceeds from the bottom up: the algebra assumes that all the children have already been folded. The hardest part of designing an algebra is figuring out what information needs to be passed up in the accumulator. We obviously need to return the final result which, in our case, is the list of strings with one mismatched character. But when we are in the middle of a trie, we have to keep in mind that the mismatch may still happen above us. So we also need a list of strings that may serve as suffixes when the mismatch occurs. We have to keep them all, because they might be matched later with strings from other branches. In other words, we need to be accumulating two lists of strings. The first list accumulates all suffixes for future matching, the second accumulates the results: strings with one mismatch (after the mismatch has been removed). We therefore should implement the following algebra: Algebra TrieF ([String], [String]) To understand the implementation of this algebra, consider a single node in a trie. It’s a list of branches, or pairs, whose first component is the current character, and the second a pair of lists of strings–the result of folding a child trie. The first list contains all the suffixes gathered from lower levels of the trie. The second list contains partial results: strings that were matched modulo single-character defect. As an example, suppose that you have a node with two branches: [ ('a', (["bcd", "efg"], ["pq"])) , ('x', (["bcd"], []))] First we prepend the current character to strings in both lists using the function prep with the following signature: prep :: (Char, ([String], [String])) -> ([String], [String]) This way we convert each branch to a pair of lists. [ (["abcd", "aefg"], ["apq"]) , (["xbcd"], [])] We then merge all the lists of suffixes and, separately, all the lists of partial results, across all branches. In the example above, we concatenate the lists in the two columns. (["abcd", "aefg", "xbcd"], ["apq"])  Now we have to construct new partial results. To do this, we create another list of accumulated strings from all branches (this time without prefixing them): ss = concat$ fmap (fst . snd) bs

In our case, this would be the list:

["bcd", "efg", "bcd"]

To detect duplicate strings, we’ll insert them into a multiset, which we’ll implement as a map. We need to import the appropriate library:

import qualified Data.Map as M

and define a multiset Counts as:

type Counts a = M.Map a Int

Every time we add a new item, we increment the count:

add :: Ord a => Counts a -> a -> Counts a
add cs c = M.insertWith (+) c 1 cs

To insert all strings from a list, we use a fold:

mset = foldl add M.empty ss

We are only interested in items that have multiplicity greater than one. We can filter them and extract their keys:

dups = M.keys $M.filter (> 1) mset Here’s the complete algebra: accum :: Algebra TrieF ([String], [String]) accum (TrieF []) = ([""], []) accum (TrieF bs) = -- b :: (Char, ([String], [String])) let -- prepend chars to string in both lists pss = unzip$ fmap prep bs
(ss1, ss2) = both concat pss
-- find duplicates
ss = concat $fmap (fst . snd) bs mset = foldl add M.empty ss dups = M.keys$ M.filter (> 1) mset
in (ss1, dups ++ ss2)
where
prep :: (Char, ([String], [String])) -> ([String], [String])
prep (c, pss) = both (fmap (c:)) pss

I used a handy helper function that applies a function to both components of a pair:

both :: (a -> b) -> (a, a) -> (b, b)
both f (x, y) = (f x, f y)

And now for the grand finale: Since we create the trie using an anamorphism only to immediately fold it using a catamorphism, why don’t we cut the middle person? Indeed, there is an algorithm called the hylomorphism that does just that. It takes the algebra, the coalgebra, and the seed, and returns the fully charged accumulator.

hylo :: Functor f => Algebra f a -> Coalgebra f b -> b -> a
hylo alg coa = alg . fmap (hylo alg coa) . coa

And this is how we extract and print the final result:

print $head$ snd $hylo accum fromList cs ## Conclusion The advantage of using the hylomorphism is that, because of Haskell’s laziness, the trie is never wholly constructed, and therefore doesn’t require large amounts of memory. At every step enough of the data structure is created as is needed for immediate computation; then it is promptly released. In fact, the definition of the data structure is only there to guide the steps of the algorithm. We use a data structure as a control structure. Since data structures are much easier to visualize and debug than control structures, it’s almost always advantageous to use them to drive computation. In fact, you may notice that, in the very last step of the computation, our accumulator recreates the original list of strings (actually, because of laziness, they are never fully reconstructed, but that’s not the point). In reality, the characters in the strings are never copied–the whole algorithm is just a choreographed dance of internal pointers, or iterators. But that’s exactly what happens in the original C++ algorithm. We just use a higher level of abstraction to describe this dance. I haven’t looked at the performance of various implementations. Feel free to test it and report the results. The code is available on github. ## Acknowledgments I’m grateful to the participants of the Seattle Haskell Users’ Group for many helpful comments during my presentation. There is a lot of folklore about various data types that pop up in discussions about lenses. For instance, it’s known that FunList and Bazaar are equivalent, although I haven’t seen a proof of that. Since both data structures appear in the context of Traversable, which is of great interest to me, I decided to do some research. In particular, I was interested in translating these data structures into constructs in category theory. This is a continuation of my previous blog posts on free monoids and free applicatives. Here’s what I have found out: • FunList is a free applicative generated by the Store functor. This can be shown by expressing the free applicative construction using Day convolution. • Using Yoneda lemma in the category of applicative functors I can show that Bazaar is equivalent to FunList Let’s start with some definitions. FunList was first introduced by Twan van Laarhoven in his blog. Here’s a (slightly generalized) Haskell definition: data FunList a b t = Done t | More a (FunList a b (b -> t)) It’s a non-regular inductive data structure, in the sense that its data constructor is recursively called with a different type, here the function type b->t. FunList is a functor in t, which can be written categorically as: $L_{a b} t = t + a \times L_{a b} (b \to t)$ where $b \to t$ is a shorthand for the hom-set $Set(b, t)$. Strictly speaking, a recursive data structure is defined as an initial algebra for a higher-order functor. I will show that the higher order functor in question can be written as: $A_{a b} g = I + \sigma_{a b} \star g$ where $\sigma_{a b}$ is the (indexed) store comonad, which can be written as: $\sigma_{a b} s = \Delta_a s \times C(b, s)$ Here, $\Delta_a$ is the constant functor, and $C(b, -)$ is the hom-functor. In Haskell, this is equivalent to: newtype Store a b s = Store (a, b -> s) The standard (non-indexed) Store comonad is obtained by identifying a with b and it describes the objects of the slice category $C/s$ (morphisms are functions $f : a \to a'$ that make the obvious triangles commute). If you’ve read my previous blog posts, you may recognize in $A_{a b}$ the functor that generates a free applicative functor (or, equivalently, a free monoidal functor). Its fixed point can be written as: $L_{a b} = I + \sigma_{a b} \star L_{a b}$ The star stands for Day convolution–in Haskell expressed as an existential data type: data Day f g s where Day :: f a -> g b -> ((a, b) -> s) -> Day f g s Intuitively, $L_{a b}$ is a “list of” Store functors concatenated using Day convolution. An empty list is the identity functor, a one-element list is the Store functor, a two-element list is the Day convolution of two Store functors, and so on… In Haskell, we would express it as: data FunList a b t = Done t | More ((Day (Store a b) (FunList a b)) t) To show the equivalence of the two definitions of FunList, let’s expand the definition of Day convolution inside $A_{a b}$: $(A_{a b} g) t = t + \int^{c d} (\Delta_b c \times C(a, c)) \times g d \times C(c \times d, t)$ The coend $\int^{c d}$ corresponds, in Haskell, to the existential data type we used in the definition of Day. Since we have the hom-functor $C(a, c)$ under the coend, the first step is to use the co-Yoneda lemma to “perform the integration” over $c$, which replaces $c$ with $a$ everywhere. We get: $t + \int^d \Delta_b a \times g d \times C(a \times d, t)$ We can then evaluate the constant functor and use the currying adjunction: $C(a \times d, t) \cong C(d, a \to t)$ to get: $t + \int^d b \times g d \times C(d, a \to t)$ Applying the co-Yoneda lemma again, we replace $d$ with $a \to t$: $t + b \times g (a \to t)$ This is exactly the functor that generates FunList. So FunList is indeed the free applicative generated by Store. All transformations in this derivation were natural isomorphisms. Now let’s switch our attention to Bazaar, which can be defined as: type Bazaar a b t = forall f. Applicative f => (a -> f b) -> f t (The actual definition of Bazaar in the lens library is even more general–it’s parameterized by a profunctor in place of the arrow in a -> f b.) The universal quantification in the definition of Bazaar immediately suggests the application of my favorite double Yoneda trick in the functor category: The set of natural transformations (morphisms in the functor category) between two functors (objects in the functor category) is isomorphic, through Yoneda embedding, to the following end in the functor category: $Nat(h, g) \cong \int_{f \colon [C, Set]} Set(Nat(g, f), Nat(h, f))$ The end is equivalent (modulo parametricity) to Haskell forall. Here, the sets of natural transformations between pairs of functors are just hom-functors in the functor category and the end over $f$ is a set of higher-order natural transformations between them. In the double Yoneda trick we carefully select the two functors $g$ and $h$ to be either representable, or somehow related to representables. The universal quantification in Bazaar is limited to applicative functors, so we’ll pick our two functors to be free applicatives. We’ve seen previously that the higher-order functor that generates free applicatives has the form: $F g = Id + g \star F g$ Here’s the version of the Yoneda embedding in which $f$ varies over all applicative functors in the category $App$, and $g$ and $h$ are arbitrary functors in $[C, Set]$: $App(F h, F g) \cong \int_{f \colon App} Set(App(F g, f), App(F h, f))$ The free functor $F$ is the left adjoint to the forgetful functor $U$: $App(F g, f) \cong [C, Set](g, U f)$ Using this adjunction, we arrive at: $[C, Set](h, U (F g)) \cong \int_{f \colon App} Set([C, Set](g, U f), [C, Set](h, U f))$ We’re almost there–we just need to carefuly pick the functors $g$ and $h$. In order to arrive at the definition of Bazaar we want: $g = \sigma_{a b} = \Delta_a \times C(b, -)$ $h = C(t, -)$ The right hand side becomes: $\int_{f \colon App} Set\big(\int_c Set (\Delta_a c \times C(b, c), (U f) c)), \int_c Set (C(t, c), (U f) c)\big)$ where I represented natural transformations as ends. The first term can be curried: $Set \big(\Delta_a c \times C(b, c), (U f) c)\big) \cong Set\big(C(b, c), \Delta_a c \to (U f) c \big)$ and the end over $c$ can be evaluated using the Yoneda lemma. So can the second term. Altogether, the right hand side becomes: $\int_{f \colon App} Set\big(a \to (U f) b)), (U f) t)\big)$ In Haskell notation, this is just the definition of Bazaar: forall f. Applicative f => (a -> f b) -> f t The left hand side can be written as: $\int_c Set(h c, (U (F g)) c)$ Since we have chosen $h$ to be the hom-functor $C(t, -)$, we can use the Yoneda lemma to “perform the integration” and arrive at: $(U (F g)) t$ With our choice of $g = \sigma_{a b}$, this is exactly the free applicative generated by Store–in other words, FunList. This proves the equivalence of Bazaar and FunList. Notice that this proof is only valid for $Set$-valued functors, although a generalization to the enriched setting is relatively straightforward. There is another family of functors, Traversable, that uses universal quantification over applicatives: class (Functor t, Foldable t) => Traversable t where traverse :: forall f. Applicative f => (a -> f b) -> t a -> f (t b) The same double Yoneda trick can be applied to it to show that it’s related to Bazaar. There is, however, a much simpler derivation, suggested to me by Derek Elkins, by changing the order of arguments: traverse :: t a -> (forall f. Applicative f => (a -> f b) -> f (t b)) which is equivalent to: traverse :: t a -> Bazaar a b (t b) In view of the equivalence between Bazaar and FunList, we can also write it as: traverse :: t a -> FunList a b (t b) Note that this is somewhat similar to the definition of toList: toList :: Foldable t => t a -> [a] In a sense, FunList is able to freely accumulate the effects from traversable, so that they can be interpreted later. ## Acknowledgments I’m grateful to Edward Kmett and Derek Elkins for many discussions and valuable insights. # Abstract The use of free monads, free applicatives, and cofree comonads lets us separate the construction of (often effectful or context-dependent) computations from their interpretation. In this paper I show how the ad hoc process of writing interpreters for these free constructions can be systematized using the language of higher order algebras (coalgebras) and catamorphisms (anamorphisms). # Introduction Recursive schemes [meijer] are an example of successful application of concepts from category theory to programming. The idea is that recursive data structures can be defined as initial algebras of functors. This allows a separation of concerns: the functor describes the local shape of the data structure, and the fixed point combinator builds the recursion. Operations over data structures can be likewise separated into shallow, non-recursive computations described by algebras, and generic recursive procedures described by catamorphisms. In this way, data structures often replace control structures in driving computations. Since functors also form a category, it’s possible to define functors acting on functors. Such higher order functors show up in a number of free constructions, notably free monads, free applicatives, and cofree comonads. These free constructions have good composability properties and they provide means of separating the creation of effectful computations from their interpretation. This paper’s contribution is to systematize the construction of such interpreters. The idea is that free constructions arise as fixed points of higher order functors, and therefore can be approached with the same algebraic machinery as recursive data structures, only at a higher level. In particular, interpreters can be constructed as catamorphisms or anamorphisms of higher order algebras/coalgebras. # Initial Algebras and Catamorphisms The canonical example of a data structure that can be described as an initial algebra of a functor is a list. In Haskell, a list can be defined recursively: data List a = Nil | Cons a (List a)  There is an underlying non-recursive functor: data ListF a x = NilF | ConsF a x instance Functor (ListF a) where fmap f NilF = NilF fmap f (ConsF a x) = ConsF a (f x)  Once we have a functor, we can define its algebras. An algebra consist of a carrier c and a structure map (evaluator). An algebra can be defined for an arbitrary functor f: type Algebra f c = f c -> c  Here’s an example of a simple list algebra, with Int as its carrier: sum :: Algebra (ListF Int) Int sum NilF = 0 sum (ConsF a c) = a + c  Algebras for a given functor form a category. The initial object in this category (if it exists) is called the initial algebra. In Haskell, we call the carrier of the initial algebra Fix f. Its structure map is a function: f (Fix f) -> Fix f  By Lambek’s lemma, the structure map of the initial algebra is an isomorphism. In Haskell, this isomorphism is given by a pair of functions: the constructor In and the destructor out of the fixed point combinator: newtype Fix f = In { out :: f (Fix f) }  When applied to the list functor, the fixed point gives rise to an alternative definition of a list: type List a = Fix (ListF a)  The initiality of the algebra means that there is a unique algebra morphism from it to any other algebra. This morphism is called a catamorphism and, in Haskell, can be expressed as: cata :: Functor f => Algebra f a -> Fix f -> a cata alg = alg . fmap (cata alg) . out  A list catamorphism is known as a fold. Since the list functor is a sum type, its algebra consists of a value—the result of applying the algebra to NilF—and a function of two variables that corresponds to the ConsF constructor. You may recognize those two as the arguments to foldr: foldr :: (a -> c -> c) -> c -> [a] -> c  The list functor is interesting because its fixed point is a free monoid. In category theory, monoids are special objects in monoidal categories—that is categories that define a product of two objects. In Haskell, a pair type plays the role of such a product, with the unit type as its unit (up to isomorphism). As you can see, the list functor is the sum of a unit and a product. This formula can be generalized to an arbitrary monoidal category with a tensor product $\otimes$ and a unit $1$: $L\, a\, x = 1 + a \otimes x$ Its initial algebra is a free monoid . # Higher Algebras In category theory, once you performed a construction in one category, it’s easy to perform it in another category that shares similar properties. In Haskell, this might require reimplementing the construction. We are interested in the category of endofunctors, where objects are endofunctors and morphisms are natural transformations. Natural transformations are represented in Haskell as polymorphic functions: type f :~> g = forall a. f a -> g a infixr 0 :~>  In the category of endofunctors we can define (higher order) functors, which map functors to functors and natural transformations to natural transformations: class HFunctor hf where hfmap :: (g :~> h) -> (hf g :~> hf h) ffmap :: Functor g => (a -> b) -> hf g a -> hf g b  The first function lifts a natural transformation; and the second function, ffmap, witnesses the fact that the result of a higher order functor is again a functor. An algebra for a higher order functor hf consists of a functor f (the carrier object in the functor category) and a natural transformation (the structure map): type HAlgebra hf f = hf f :~> f  As with regular functors, we can define an initial algebra using the fixed point combinator for higher order functors: newtype FixH hf a = InH { outH :: hf (FixH hf) a }  Similarly, we can define a higher order catamorphism: hcata :: HFunctor h => HAlgebra h f -> FixH h :~> f hcata halg = halg . hfmap (hcata halg) . outH  The question is, are there any interesting examples of higher order functors and algebras that could be used to solve real-life programming problems? # Free Monad We’ve seen the usefulness of lists, or free monoids, for structuring computations. Let’s see if we can generalize this concept to higher order functors. The definition of a list relies on the cartesian structure of the underlying category. It turns out that there are multiple cartesian structures of interest that can be defined in the category of functors. The simplest one defines a product of two endofunctors as their composition. Any two endofunctors can be composed. The unit of functor composition is the identity functor. If you picture endofunctors as containers, you can easily imagine a tree of lists, or a list of Maybes. A monoid based on this particular monoidal structure in the endofunctor category is a monad. It’s an endofunctor m equipped with two natural transformations representing unit and multiplication: class Monad m where eta :: Identity :~> m mu :: Compose m m :~> m  In Haskell, the components of these natural transformations are known as return and join. A straightforward generalization of the list functor to the functor category can be written as: $L\, f\, g = 1 + f \circ g$ or, in Haskell, type FunctorList f g = Identity :+: Compose f g  where we used the operator :+: to define the coproduct of two functors: data (f :+: g) e = Inl (f e) | Inr (g e) infixr 7 :+:  Using more conventional notation, FunctorList can be written as: data MonadF f g a = DoneM a | MoreM (f (g a))  We’ll use it to generate a free monoid in the category of endofunctors. First of all, let’s show that it’s indeed a higher order functor in the second argument g: instance Functor f => HFunctor (MonadF f) where hfmap _ (DoneM a) = DoneM a hfmap nat (MoreM fg) = MoreM$ fmap nat fg
ffmap h (DoneM a)    = DoneM (h a)
ffmap h (MoreM fg)   = MoreM $fmap (fmap h) fg  In category theory, because of size issues, this functor doesn’t always have a fixed point. For most common choices of f (e.g., for algebraic data types), the initial higher order algebra for this functor exists, and it generates a free monad. In Haskell, this free monad can be defined as: type FreeMonad f = FixH (MonadF f)  We can show that FreeMonad is indeed a monad by implementing return and bind: instance Functor f => Monad (FreeMonad f) where return = InH . DoneM (InH (DoneM a)) >>= k = k a (InH (MoreM ffra)) >>= k = InH (MoreM (fmap (>>= k) ffra))  Free monads have many applications in programming. They can be used to write generic monadic code, which can then be interpreted in different monads. A very useful property of free monads is that they can be composed using coproducts. This follows from the theorem in category theory, which states that left adjoints preserve coproducts (or, more generally, colimits). Free constructions are, by definition, left adjoints to forgetful functors. This property of free monads was explored by Swierstra [swierstra] in his solution to the expression problem. I will use an example based on his paper to show how to construct monadic interpreters using higher order catamorphisms. ## Free Monad Example A stack-based calculator can be implemented directly using the state monad. Since this is a very simple example, it will be instructive to re-implement it using the free monad approach. We start by defining a functor, in which the free parameter k represents the continuation: data StackF k = Push Int k | Top (Int -> k) | Pop k | Add k deriving Functor  We use this functor to build a free monad: type FreeStack = FreeMonad StackF  You may think of the free monad as a tree with nodes that are defined by the functor StackF. The unary constructors, like Add or Pop, create linear list-like branches; but the Top constructor branches out with one child per integer. The level of indirection we get by separating recursion from the functor makes constructing free monad trees syntactically challenging, so it makes sense to define a helper function: liftF :: (Functor f) => f r -> FreeMonad f r liftF fr = InH$ MoreM $fmap (InH . DoneM) fr  With this function, we can define smart constructors that build leaves of the free monad tree: push :: Int -> FreeStack () push n = liftF (Push n ()) pop :: FreeStack () pop = liftF (Pop ()) top :: FreeStack Int top = liftF (Top id) add :: FreeStack () add = liftF (Add ())  All these preparations finally pay off when we are able to create small programs using do notation: calc :: FreeStack Int calc = do push 3 push 4 add x <- top pop return x  Of course, this program does nothing but build a tree. We need a separate interpreter to do the calculation. We’ll interpret our program in the state monad, with state implemented as a stack (list) of integers: type MemState = State [Int]  The trick is to define a higher order algebra for the functor that generates the free monad and then use a catamorphism to apply it to the program. Notice that implementing the algebra is a relatively simple procedure because we don’t have to deal with recursion. All we need is to case-analyze the shallow constructors for the free monad functor MonadF, and then case-analyze the shallow constructors for the functor StackF. runAlg :: HAlgebra (MonadF StackF) MemState runAlg (DoneM a) = return a runAlg (MoreM ex) = case ex of Top ik -> get >>= ik . head Pop k -> get >>= put . tail >> k Push n k -> get >>= put . (n : ) >> k Add k -> do (a: b: s) <- get put (a + b : s) k  The catamorphism converts the program calc into a state monad action, which can be run over an empty initial stack: runState (hcata runAlg calc) []  The real bonus is the freedom to define other interpreters by simply switching the algebras. Here’s an algebra whose carrier is the Const functor: showAlg :: HAlgebra (MonadF StackF) (Const String) showAlg (DoneM a) = Const "Done!" showAlg (MoreM ex) = Const$
case ex of
Push n k ->
"Push " ++ show n ++ ", " ++ getConst k
Top ik ->
"Top, " ++ getConst (ik 42)
Pop k ->
"Pop, " ++ getConst k


Runing the catamorphism over this algebra will produce a listing of our program:

getConst $hcata showAlg calc > "Push 3, Push 4, Add, Top, Pop, Done!" # Free Applicative There is another monoidal structure that exists in the category of functors. In general, this structure will work for functors from an arbitrary monoidal category $C$ to $Set$. Here, we’ll restrict ourselves to endofunctors on $Set$. The product of two functors is given by Day convolution, which can be implemented in Haskell using an existential type: data Day f g c where Day :: f a -> g b -> ((a, b) -> c) -> Day f g c  The intuition is that a Day convolution contains a container of some as, and another container of some bs, together with a function that can convert any pair (a, b) to c. Day convolution is a higher order functor: instance HFunctor (Day f) where hfmap nat (Day fx gy xyt) = Day fx (nat gy) xyt ffmap h (Day fx gy xyt) = Day fx gy (h . xyt)  In fact, because Day convolution is symmetric up to isomorphism, it is automatically functorial in both arguments. To complete the monoidal structure, we also need a functor that could serve as a unit with respect to Day convolution. In general, this would be the the hom-functor from the monoidal unit: $C(1, -)$ In our case, since $1$ is the singleton set, this functor reduces to the identity functor. We can now define monoids in the category of functors with the monoidal structure given by Day convolution. These monoids are equivalent to lax monoidal functors which, in Haskell, form the class: class Functor f => Monoidal f where unit :: f () (>*<) :: f x -> f y -> f (x, y)  Lax monoidal functors are equivalent to applicative functors [mcbride], as seen in this implementation of pure and <*>:  pure :: a -> f a pure a = fmap (const a) unit (<*>) :: f (a -> b) -> f a -> f b fs <*> as = fmap (uncurry ($)) (fs >*< as)


We can now use the same general formula, but with Day convolution as the product:

$L\, f\, g = 1 + f \star g$

to generate a free monoidal (applicative) functor:

data FreeF f g t =
DoneF t
| MoreF (Day f g t)


This is indeed a higher order functor:

instance HFunctor (FreeF f) where
hfmap _ (DoneF x)     = DoneF x
hfmap nat (MoreF day) = MoreF (hfmap nat day)
ffmap f (DoneF x)     = DoneF (f x)
ffmap f (MoreF day)   = MoreF (ffmap f day)


and it generates a free applicative functor as its initial algebra:

type FreeA f = FixH (FreeF f)


## Free Applicative Example

The following example is taken from the paper by Capriotti and Kaposi [capriotti]. It’s an option parser for a command line tool, whose result is a user record of the following form:

data User = User
, fullname :: String
, uid      :: Int
} deriving Show


A parser for an individual option is described by a functor that contains the name of the option, an optional default value for it, and a reader from string:

data Option a = Option
{ optName    :: String
, optDefault :: Maybe a
, optReader  :: String -> Maybe a
} deriving Functor


Since we don’t want to commit to a particular parser, we’ll create a parsing action using a free applicative functor:

userP :: FreeA Option User
userP  = pure User
<*> one (Option "username" (Just "John")  Just)
<*> one (Option "fullname" (Just "Doe")   Just)
<*> one (Option "uid"      (Just 0)       readInt)


where readInt is a reader of integers:

readInt :: String -> Maybe Int


and we used the following smart constructors:

one :: f a -> FreeA f a
one fa = InH $MoreF$ Day fa (done ()) fst

done :: a -> FreeA f a
done a = InH $DoneF a  We are now free to define different algebras to evaluate the free applicative expressions. Here’s one that collects all the defaults: alg :: HAlgebra (FreeF Option) Maybe alg (DoneF a) = Just a alg (MoreF (Day oa mb f)) = fmap f (optDefault oa >*< mb)  I used the monoidal instance for Maybe: instance Monoidal Maybe where unit = Just () Just x >*< Just y = Just (x, y) _ >*< _ = Nothing  This algebra can be run over our little program using a catamorphism: parserDef :: FreeA Option a -> Maybe a parserDef = hcata alg  And here’s an algebra that collects the names of all the options: alg2 :: HAlgebra (FreeF Option) (Const String) alg2 (DoneF a) = Const "." alg2 (MoreF (Day oa bs f)) = fmap f (Const (optName oa) >*< bs)  Again, this uses a monoidal instance for Const: instance Monoid m => Monoidal (Const m) where unit = Const mempty Const a >*< Const b = Const (a b)  We can also define the Monoidal instance for IO: instance Monoidal IO where unit = return () ax >*< ay = do a <- ax b <- ay return (a, b)  This allows us to interpret the parser in the IO monad: alg3 :: HAlgebra (FreeF Option) IO alg3 (DoneF a) = return a alg3 (MoreF (Day oa bs f)) = do putStrLn$ optName oa
s <- getLine
let ma = optReader oa s
a = fromMaybe (fromJust (optDefault oa)) ma
fmap f $return a >*< bs  # Cofree Comonad Every construction in category theory has its dual—the result of reversing all the arrows. The dual of a product is a coproduct, the dual of an algebra is a coalgebra, and the dual of a monad is a comonad. Let’s start by defining a higher order coalgebra consisting of a carrier f, which is a functor, and a natural transformation: type HCoalgebra hf f = f :~> hf f  An initial algebra is dualized to a terminal coalgebra. In Haskell, both are the results of applying the same fixed point combinator, reflecting the fact that the Lambek’s lemma is self-dual. The dual to a catamorphism is an anamorphism. Here is its higher order version: hana :: HFunctor hf => HCoalgebra hf f -> (f :~> FixH hf) hana hcoa = InH . hfmap (hana hcoa) . hcoa  The formula we used to generate free monoids: $1 + a \otimes x$ dualizes to: $1 \times a \otimes x$ and can be used to generate cofree comonoids . A cofree functor is the right adjoint to the forgetful functor. Just like the left adjoint preserved coproducts, the right adjoint preserves products. One can therefore easily combine comonads using products (if the need arises to solve the coexpression problem). Just like the monad is a monoid in the category of endofunctors, a comonad is a comonoid in the same category. The functor that generates a cofree comonad has the form: type ComonadF f g = Identity :*: Compose f g  where the product of functors is defined as: data (f :*: g) e = Both (f e) (g e) infixr 6 :*:  Here’s the more familiar form of this functor: data ComonadF f g e = e :< f (g e)  It is indeed a higher order functor, as witnessed by this instance: instance Functor f => HFunctor (ComonadF f) where hfmap nat (e :< fge) = e :< fmap nat fge ffmap h (e :< fge) = h e :< fmap (fmap h) fge  A cofree comonad is the terminal coalgebra for this functor and can be written as a fixed point: type Cofree f = FixH (ComonadF f)  Indeed, for any functor f, Cofree f is a comonad: instance Functor f => Comonad (Cofree f) where extract (InH (e :< fge)) = e duplicate fr@(InH (e :< fge)) = InH (fr :< fmap duplicate fge)  ## Cofree Comonad Example The canonical example of a cofree comonad is an infinite stream: type Stream = Cofree Identity  We can use this stream to sample a function. We’ll encapsulate this function inside the following functor (in fact, itself a comonad): data Store a x = Store a (a -> x) deriving Functor  We can use a higher order coalgebra to unpack the Store into a stream: streamCoa :: HCoalgebra (ComonadF Identity)(Store Int) streamCoa (Store n f) = f n :< (Identity$ Store (n + 1) f)


The actual unpacking is a higher order anamorphism:

stream :: Store Int a -> Stream a
stream = hana streamCoa


We can use it, for instance, to generate a list of squares of natural numbers:

stream (Store 0 (^2))


Since, in Haskell, the same fixed point defines a terminal coalgebra as well as an initial algebra, we are free to construct algebras and catamorphisms for streams. Here’s an algebra that converts a stream to an infinite list:

listAlg :: HAlgebra (ComonadF Identity) []
listAlg(a :< Identity as) = a : as

toList :: Stream a -> [a]
toList = hcata listAlg


# Future Directions

In this paper I concentrated on one type of higher order functor:

$1 + a \otimes x$

and its dual. This would be equivalent to studying folds for lists and unfolds for streams. But the structure of the functor category is richer than that. Just like basic data types can be combined into algebraic data types, so can functors. Moreover, besides the usual sums and products, the functor category admits at least two additional monoidal structures generated by functor composition and Day convolution.

Another potentially fruitful area of exploration is the profunctor category, which is also equipped with two monoidal structures, one defined by profunctor composition, and another by Day convolution. A free monoid with respect to profunctor composition is the basis of Haskell Arrow library [jaskelioff]. Profunctors also play an important role in the Haskell lens library [kmett].

## Bibliography

1. Erik Meijer, Maarten Fokkinga, and Ross Paterson, Functional Programming with Bananas, Lenses, Envelopes and Barbed Wire
2. Conor McBride, Ross Paterson, Idioms: applicative programming with effects
3. Paolo Capriotti, Ambrus Kaposi, Free Applicative Functors
4. Wouter Swierstra, Data types a la carte
5. Exequiel Rivas and Mauro Jaskelioff, Notions of Computation as Monoids
6. Edward Kmett, Lenses, Folds and Traversals
7. Richard Bird and Lambert Meertens, Nested Datatypes
8. Patricia Johann and Neil Ghani, Initial Algebra Semantics is Enough!

## The Free Theorem for Ends

In Haskell, the end of a profunctor p is defined as a product of all diagonal elements:

forall c. p c c

together with a family of projections:

pi :: Profunctor p => forall c. (forall a. p a a) -> p c c
pi e = e

In category theory, the end must also satisfy the edge condition which, in (type-annotated) Haskell, could be written as:

dimap f idb . pib = dimap ida f . pia

for any f :: a -> b.
Using a suitable formulation of parametricity, this equation can be shown to be a free theorem. Let’s first review the free theorem for functors before generalizing it to profunctors.

## Functor Characterization

You may think of a functor as a container that has a shape and contents. You can manipulate the contents without changing the shape using fmap. In general, when applying fmap, you not only change the values stored in the container, you change their type as well. To really capture the shape of the container, you have to consider not only all possible mappings, but also more general relations between different contents.

A function is directional, and so is fmap, but relations don’t favor either side. They can map multiple values to the same value, and they can map one value to multiple values. Any relation on values induces a relation on containers. For a given functor F, if there is a relation a between type A and type A':

A <=a=> A'

then there is a relation between type F A and F A':

F A <=(F a)=> F A'

We call this induced relation F a.

For instance, consider the relation between students and their grades. Each student may have multiple grades (if they take multiple courses) so this relation is not a function. Given a list of students and a list of grades, we would say that the lists are related if and only if they match at each position. It means that they have to be equal length, and the first grade on the list of grades must belong to the first student on the list of students, and so on. Of course, a list is a very simple container, but this property can be generalized to any functor we can define in Haskell using algebraic data types.

The fact that fmap doesn’t change the shape of the container can be expressed as a “theorem for free” using relations. We start with two related containers:

xs :: F A
xs':: F A'

where A and A' are related through some relation a. We want related containers to be fmapped to related containers. But we can’t use the same function to map both containers, because they contain different types. So we have to use two related functions instead. Related functions map related types to related types so, if we have:

f :: A -> B
f':: A'-> B'

and A is related to A' through a, we want B to be related to B' through some relation b. Also, we want the two functions to map related elements to related elements. So if x is related to x' through a, we want f x to be related to f' x' through b. In that case, we’ll say that f and f' are related through the relation that we call a->b:

f <=(a->b)=> f'

For instance, if f is mapping students’ SSNs to last names, and f' is mapping letter grades to numerical grades, the results will be related through the relation between students’ last names and their numerical grades.

To summarize, we require that for any two relations:

A <=a=> A'
B <=b=> B'

and any two functions:

f :: A -> B
f':: A'-> B'

such that:

f <=(a->b)=> f'

and any two containers:

xs :: F A
xs':: F A'

we have:

if       xs <=(F a)=> xs'
then   F xs <=(F b)=> F xs'

This characterization can be extended, with suitable changes, to contravariant functors.

## Profunctor Characterization

A profunctor is a functor of two variables. It is contravariant in the first variable and covariant in the second. A profunctor can lift two functions simultaneously using dimap:

class Profunctor p where
dimap :: (a -> b) -> (c -> d) -> p b c -> p a d

We want dimap to preserve relations between profunctor values. We start by picking any relations a, b, c, and d between types:

A <=a=> A'
B <=b=> B'
C <=c=> C'
D <=d=> D'


For any functions:

f  :: A -> B
f' :: A'-> B'
g  :: C -> D
g' :: C'-> D'

that are related through the following relations induced by function types:

f <=(a->b)=> f'
g <=(c->d)=> g'

we define:

xs :: p B C
xs':: p B'C'

The following condition must be satisfied:

if             xs <=(p b c)=> xs'
then   (p f g) xs <=(p a d)=> (p f' g') xs'


where p f g stands for the lifting of the two functions by the profunctor p.

Here’s a quick sanity check. If b and c are functions:

b :: B'-> B
c :: C -> C'

than the relation:

xs <=(p b c)=> xs'

becomes:

xs' = dimap b c xs


If a and d are functions:

a :: A'-> A
d :: D -> D'


then these relations:

f <=(a->b)=> f'
g <=(c->d)=> g'

become:

f . a = b . f'
d . g = g'. c

and this relation:

(p f g) xs <=(p a d)=> (p f' g') xs'

becomes:

(p f' g') xs' = dimap a d ((p f g) xs)

Substituting xs', we get:

dimap f' g' (dimap b c xs) = dimap a d (dimap f g xs)

and using functoriality:

dimap (b . f') (g'. c) = dimap (f . a) (d . g)


which is identically true.

## Special Case of Profunctor Characterization

We are interested in the diagonal elements of a profunctor. Let’s first specialize the general case to:

C = B
C'= B'
c = b

to get:

xs = p B B
xs'= p B'B'

and

if             xs <=(p b b)=> xs'
then   (p f g) xs <=(p a d)=> (p f' g') xs'


Chosing the following substitutions:

A = A'= B
D = D'= B'
a = id
d = id
f = id
g'= id
f'= g

we get:

if              xs <=(p b b)=> xs'
then   (p id g) xs <=(p id id)=> (p g id) xs'


Since p id id is the identity relation, we get:

(p id g) xs = (p g id) xs'

or

dimap id g xs = dimap g id xs'

## Free Theorem

We apply the free theorem to the term xs:

xs :: forall c. p c c

It must be related to itself through the relation that is induced by its type:

xs <=(forall b. p b b)=> xs

for any relation b:

B <=b=> B'

Universal quantification translates to a relation between different instantiations of the polymorphic value:

xsB <=(p b b)=> xsB'

Notice that we can write:

xsB = piB xs
xsB'= piB'xs

using the projections we defined earlier.

We have just shown that this equation leads to:

dimap id g xs = dimap g id xs'

which shows that the wedge condition is indeed a free theorem.

## Natural Transformations

Here’s another quick application of the free theorem. The set of natural transformations may be represented as an end of the following profunctor:

type NatP a b = F a -> G b
instance Profunctor NatP where
dimap f g alpha = fmap g . alpha . fmap f

The free theorem tells us that for any mu :: NatP c c:

(dimap id g) mu = (dimap g id) mu

which is the naturality condition:

mu . fmap g = fmap g . mu

It’s been know for some time that, in Haskell, naturality follows from parametricity, so this is not surprising.

## Acknowledgment

I’d like to thank Edward Kmett for reviewing the draft of this post.

## Bibliography

1. Bartosz Milewski, Ends and Coends
2. Edsko de Vries, Parametricity Tutorial, Part 1, Part 2, Contravariant Functions.

There are many intuitions that we may attach to morphisms in a category, but we can all agree that if there is a morphism from the object a to the object b than the two objects are in some way “related.” A morphism is, in a sense, the proof of this relation. This is clearly visible in any poset category, where a morphism is a relation. In general, there may be many “proofs” of the same relation between two objects. These proofs form a set that we call the hom-set. When we vary the objects, we get a mapping from pairs of objects to sets of “proofs.” This mapping is functorial — contravariant in the first argument and covariant in the second. We can look at it as establishing a global relationship between objects in the category. This relationship is described by the hom-functor:

C(-, =) :: Cop × C -> Set

In general, any functor like this may be interpreted as establishing a relation between objects in a category. A relation may also involve two different categories C and D. A functor, which describes such a relation, has the following signature and is called a profunctor:

p :: Dop × C -> Set

Mathematicians say that it’s a profunctor from C to D (notice the inversion), and use a slashed arrow as a symbol for it:

C ↛ D

You may think of a profunctor as a proof-relevant relation between objects of C and objects of D, where the elements of the set symbolize proofs of the relation. Whenever p a b is empty, there is no relation between a and b. Keep in mind that relations don’t have to be symmetric.

Another useful intuition is the generalization of the idea that an endofunctor is a container. A profunctor value of the type p a b could then be considered a container of bs that are keyed by elements of type a. In particular, an element of the hom-profunctor is a function from a to b.

In Haskell, a profunctor is defined as a two-argument type constructor p equipped with the method called dimap, which lifts a pair of functions, the first going in the “wrong” direction:

class Profunctor p where
dimap :: (c -> a) -> (b -> d) -> p a b -> p c d

The functoriality of the profunctor tells us that if we have a proof that a is related to b, then we get the proof that c is related to d, as long as there is a morphism from c to a and another from b to d. Or, we can think of the first function as translating new keys to the old keys, and the second function as modifying the contents of the container.

For profunctors acting within one category, we can extract quite a lot of information from diagonal elements of the type p a a. We can prove that b is related to c as long as we have a pair of morphisms b->a and a->c. Even better, we can use a single morphism to reach off-diagonal values. For instance, if we have a morphism f::a->b, we can lift the pair <f, idb> to go from p b b to p a b:

dimap f id pbb :: p a b

Or we can lift the pair <ida, f> to go from p a a to p a b:

dimap id f paa :: p a b

## Dinatural Transformations

Since profunctors are functors, we can define natural transformations between them in the standard way. In many cases, though, it’s enough to define the mapping between diagonal elements of two profunctors. Such a transformation is called a dinatural transformation, provided it satisfies the commuting conditions that reflect the two ways we can connect diagonal elements to non-diagonal ones. A dinatural transformation between two profunctors p and q, which are members of the functor category [Cop × C, Set], is a family of morphisms:

αa :: p a a -> q a a

for which the following diagram commutes, for any f::a->b:

Notice that this is strictly weaker than the naturality condition. If α were a natural transformation in [Cop × C, Set], the above diagram could be constructed from two naturality squares and one functoriality condition (profunctor q preserving composition):

Notice that a component of a natural transformation α in [Cop × C, Set] is indexed by a pair of objects α a b. A dinatural transformation, on the other hand, is indexed by one object, since it only maps diagonal elements of the respective profunctors.

## Ends

We are now ready to advance from “algebra” to what could be considered the “calculus” of category theory. The calculus of ends (and coends) borrows ideas and even some notation from traditional calculus. In particular, the coend may be understood as an infinite sum or an integral, whereas the end is similar to an infinite product. There is even something that resembles the Dirac delta function.

An end is a generalization of a limit, with the functor replaced by a profunctor. Instead of a cone, we have a wedge. The base of a wedge is formed by diagonal elements of a profunctor p. The apex of the wedge is an object (here, a set, since we are considering Set-valued profunctors), and the sides are a family of functions mapping the apex to the sets in the base. You may think of this family as one polymorphic function — a function that’s polymorphic in its return type:

α :: forall a . apex -> p a a

Unlike in cones, within a wedge we don’t have any functions that would connect the vertices that form the base. However, as we’ve seen earlier, given any morphism f::a->b in C, we can connect both p a a and p b b to the common set p a b. We therefore insist that the following diagram commute:

This is called the wedge condition. It can be written as:

p ida f ∘ αa = p f idb ∘ αb

dimap id f . alpha = dimap f id . alpha

We can now proceed with the universal construction and define the end of p as the universal wedge — a set e together with a family of functions π such that for any other wedge with the apex a and a family α there is a unique function h::a->e that makes all triangles commute:

πa ∘ h = αa

The symbol for the end is the integral sign, with the “integration variable” in the subscript position:

∫c p c c

Components of π are called projection maps for the end:

πa :: ∫c p c c -> p a a

Note that if C is a discrete category (no morphisms other than the identities) the end is just a global product of all diagonal entries of p across the whole category C. Later I’ll show you that, in the more general case, there is a relationship between the end and this product through an equalizer.

In Haskell, the end formula translates directly to the universal quantifier:

forall a. p a a

Strictly speaking, this is just a product of all diagonal elements of p, but the wedge condition is satisfied automatically due to parametricity (I’ll explain it in a separate blog post). For any function f :: a -> b, the wedge condition reads:

dimap f id . pi = dimap id f . pi

or, with type annotations:

dimap f idb . pib = dimap ida f . pia

where both sides of the equation have the type:

Profunctor p => (forall c. p c c) -> p a b

and pi is the polymorphic projection:

pi :: Profunctor p => forall c. (forall a. p a a) -> p c c
pi e = e

Here, type inference automatically picks the right component of e.

Just as we were able to express the whole set of commutation conditions for a cone as one natural transformation, likewise we can group all the wedge conditions into one dinatural transformation. For that we need the generalization of the constant functor Δc to a constant profunctor that maps all pairs of objects to a single object c, and all pairs of morphisms to the identity morphism for this object. A wedge is a dinatural transformation from that functor to the profunctor p. Indeed, the dinaturality hexagon shrinks down to the wedge diamond when we realize that Δc lifts all morphisms to one identity function.

Ends can also be defined for target categories other than Set, but here we’ll only consider Set-valued profunctors and their ends.

## Ends as Equalizers

The commutation condition in the definition of the end can be written using an equalizer. First, let’s define two functions (I’m using Haskell notation, because mathematical notation seems to be less user-friendly in this case). These functions correspond to the two converging branches of the wedge condition:

lambda :: Profunctor p => p a a -> (a -> b) -> p a b
lambda paa f = dimap id f paa

rho :: Profunctor p => p b b -> (a -> b) -> p a b
rho pbb f = dimap f id pbb

Both functions map diagonal elements of the profunctor p to polymorphic functions of the type:

type ProdP p = forall a b. (a -> b) -> p a b

Functions lambda and rho have different types. However, we can unify their types, if we form one big product type, gathering together all diagonal elements of p:

newtype DiaProd p = DiaProd (forall a. p a a)

The functions lambda and rho induce two mappings from this product type:

lambdaP :: Profunctor p => DiaProd p -> ProdP p
lambdaP (DiaProd paa) = lambda paa

rhoP :: Profunctor p => DiaProd p -> ProdP p
rhoP (DiaProd paa) = rho paa

The end of p is the equalizer of these two functions. Remember that the equalizer picks the largest subset on which two functions are equal. In this case it picks the subset of the product of all diagonal elements for which the wedge diagrams commute.

## Natural Transformations as Ends

The most important example of an end is the set of natural transformations. A natural transformation between two functors F and G is a family of morphisms picked from hom-sets of the form C(F a, G a). If it weren’t for the naturality condition, the set of natural transformations would be just the product of all these hom-sets. In fact, in Haskell, it is:

forall a. f a -> g a

The reason it works in Haskell is because naturality follows from parametricity. Outside of Haskell, though, not all diagonal sections across such hom-sets will yield natural transformations. But notice that the mapping:

<a, b> -> C(F a, G b)

is a profunctor, so it makes sense to study its end. This is the wedge condition:

Let’s just pick one element from the set ∫c C(F c, G c). The two projections will map this element to two components of a particular transformation, let’s call them:

τa :: F a -> G a
τb :: F b -> G b

In the left branch, we lift a pair of morphisms <ida, G f> using the hom-functor. You may recall that such lifting is implemented as simultaneous pre- and post-composition. When acting on τa the lifted pair gives us:

G f ∘ τa ∘ ida

The other branch of the diagram gives us:

idb ∘ τb ∘ F f

Their equality, demanded by the wedge condition, is nothing but the naturality condition for τ.

## Coends

As expected, the dual to an end is called a coend. It is constructed from a dual to a wedge called a cowedge (pronounced co-wedge, not cow-edge).

An edgy cow?

The symbol for a coend is the integral sign with the “integration variable” in the superscript position:

∫ c p c c

Just like the end is related to a product, the coend is related to a coproduct, or a sum (in this respect, it resembles an integral, which is a limit of a sum). Rather than having projections, we have injections going from the diagonal elements of the profunctor down to the coend. If it weren’t for the cowedge conditions, we could say that the coend of the profunctor p is either p a a, or p b b, or p c c, and so on. Or we could say that there exists such an a for which the coend is just the set p a a. The universal quantifier that we used in the definition of the end turns into an existential quantifier for the coend.

This is why, in pseudo-Haskell, we would define the coend as:

exists a. p a a

The standard way of encoding existential quantifiers in Haskell is to use universally quantified data constructors. We can thus define:

data Coend p = forall a. Coend (p a a)

The logic behind this is that it should be possible to construct a coend using a value of any of the family of types p a a, no matter what a we chose.

Just like an end can be defined using an equalizer, a coend can be described using a coequalizer. All the cowedge conditions can be summarized by taking one gigantic coproduct of p a b for all possible functions b->a. In Haskell, that would be expressed as an existential type:

data SumP p = forall a b. SumP (b -> a) (p a b)

There are two ways of evaluating this sum type, by lifting the function using dimap and applying it to the profunctor p:

lambda, rho :: Profunctor p => SumP p -> DiagSum p
lambda (SumP f pab) = DiagSum (dimap f id pab)
rho    (SumP f pab) = DiagSum (dimap id f pab)

where DiagSum is the sum of diagonal elements of p:

data DiagSum p = forall a. DiagSum (p a a)

The coequalizer of these two functions is the coend. A coequilizer is obtained from DiagSum p by identifying values that are obtained by applying lambda or rho to the same argument. Here, the argument is a pair consisting of a function b->a and an element of p a b. The application of lambda and rho produces two potentially different values of the type DiagSum p. In the coend, these two values are identified, making the cowedge condition automatically satisfied.

The process of identification of related elements in a set is formally known as taking a quotient. To define a quotient we need an equivalence relation ~, a relation that is reflexive, symmetric, and transitive:

a ~ a
if a ~ b then b ~ a
if a ~ b and b ~ c then a ~ c

Such a relation splits the set into equivalence classes. Each class consists of elements that are related to each other. We form a quotient set by picking one representative from each class. A classic example is the definition of rational numbers as pairs of whole numbers with the following equivalence relation:

(a, b) ~ (c, d) iff a * d = b * c

It’s easy to check that this is an equivalence relation. A pair (a, b) is interpreted as a fraction a/b, and fractions whose numerator and denominator have a common divisor are identified. A rational number is an equivalence class of such fractions.

You might recall from our earlier discussion of limits and colimits that the hom-functor is continuous, that is, it preserves limits. Dually, the contravariant hom-functor turns colimits into limits. These properties can be generalized to ends and coends, which are a generalization of limits and colimits, respectively. In particular, we get a very useful identity for converting coends to ends:

Set(∫ x p x x, c) ≅ ∫x Set(p x x, c)

Let’s have a look at it in pseudo-Haskell:

(exists x. p x x) -> c ≅ forall x. p x x -> c

It tells us that a function that takes an existential type is equivalent to a polymorphic function. This makes perfect sense, because such a function must be prepared to handle any one of the types that may be encoded in the existential type. It’s the same principle that tells us that a function that accepts a sum type must be implemented as a case statement, with a tuple of handlers, one for every type present in the sum. Here, the sum type is replaced by a coend, and a family of handlers becomes an end, or a polymorphic function.

## Ninja Yoneda Lemma

The set of natural transformations that appears in the Yoneda lemma may be encoded using an end, resulting in the following formulation:

∫z Set(C(a, z), F z) ≅ F a

There is also a dual formula:

∫ z C(z, a) × F z ≅ F a

This identity is strongly reminiscent of the formula for the Dirac delta function (a function δ(a - z), or rather a distribution, that has an infinite peak at a = z). Here, the hom-functor plays the role of the delta function.

Together these two identities are sometimes called the Ninja Yoneda lemma.

To prove the second formula, we will use the consequence of the Yoneda embedding, which states that two objects are isomorphic if and only if their hom-functors are isomorphic. In other words a ≅ b if and only if there is a natural transformation of the type:

[C, Set](C(a, -), C(b, =))

that is an isomorphism.

We start by inserting the left-hand side of the identity we want to prove inside a hom-functor that’s going to some arbitrary object c:

Set(∫ z C(z, a) × F z, c)

Using the continuity argument, we can replace the coend with the end:

∫z Set(C(z, a) × F z, c)

We can now take advantage of the adjunction between the product and the exponential:

∫z Set(C(z, a), c(F z))

We can “perform the integration” by using the Yoneda lemma to get:

c(F a)

(Notice that we used the contravariant version of the Yoneda lemma, since c(F z) is contravariant in z.)
This exponential object is isomorphic to the hom-set:

Set(F a, c)

Finally, we take advantage of the Yoneda embedding to arrive at the isomorphism:

∫ z C(z, a) × F z ≅ F a

## Profunctor Composition

Let’s explore further the idea that a profunctor describes a relation — more precisely, a proof-relevant relation, meaning that the set p a b represents the set of proofs that a is related to b. If we have two relations p and q we can try to compose them. We’ll say that a is related to b through the composition of q after p if there exist an intermediary object c such that both q b c and p c a are non-empty. The proofs of this new relation are all pairs of proofs of individual relations. Therefore, with the understanding that the existential quantifier corresponds to a coend, and the cartesian product of two sets corresponds to “pairs of proofs,” we can define composition of profunctors using the following formula:

(q ∘ p) a b = ∫ c p c a × q b c

Here’s the equivalent Haskell definition from Data.Profunctor.Composition, after some renaming:

data Procompose q p a b where
Procompose :: q a c -> p c b -> Procompose q p a b


This is using generalized algebraic data type, or GADT syntax, in which a free type variable (here c) is automatically existentially quanitified. The (uncurried) data constructor Procompose is thus equivalent to:

exists c. (q a c, p c b)

The unit of so defined composition is the hom-functor — this immediately follows from the Ninja Yoneda lemma. It makes sense, therefore, to ask the question if there is a category in which profunctors serve as morphisms. The answer is positive, with the caveat that both associativity and identity laws for profunctor composition hold only up to natural isomorphism. Such a category, where laws are valid up to isomorphism, is called a bicategory (which is more general than a 2-category). So we have a bicategory Prof, in which objects are categories, morphisms are profunctors, and morphisms between morphisms (a.k.a., two-cells) are natural transformations. In fact, one can go even further, because beside profunctors, we also have regular functors as morphisms between categories. A category which has two types of morphisms is called a double category.

Profunctors play an important role in the Haskell lens library and in the arrow library.

## Challenge

1. Write explicitly the cowedge condition for a coend.

Next: Kan extensions.

If there is one structure that permeates category theory and, by implication, the whole of mathematics, it’s the monoid. To study the evolution of this concept is to study the power of abstraction and the idea of getting more for less, which is at the core of mathematics. When I say “evolution” I don’t necessarily mean chronological development. I’m looking at a monoid as if it were a life form evolving through various eons of abstraction.

It’s an ambitious project and I’ll have to cover a lot of material. I’ll start slowly, with the definitions of magmas and monoids, but then I will accelerate. A lot of concepts will be introduced in one or two sentences, mainly to familiarize the reader with the notation. I’ll dwell a little on monoidal categories, then breeze through ends, coends, and profunctors. I’ll show you how monads, arrows, and applicative functors arise from monoids in various monoidal categories.

## The Magmas of the Hadean Eon

Monoids evolved from more primitive life forms feeding on sets. So, before even touching upon monoids, let’s talk about cartesian products, relations, and functions. You take two sets a and b (or, in the simplest case, two copies of the same set a) and form pairs of elements. That gives you a set of pairs, a.k.a., the cartesian product a×b. Any subset of such a cartesian product is called a relation. Two elements x and y are in a relation if the pair <x, y> is a member of that subset.

A function from a to b is a special kind of relation, in which every element x in the set a has one and only one element y in the set b that’s related to it. (Sometimes this is called a total function, since it’s defined for all elements of a).

Even before there were monoids, there was magma. A magma is a set with a binary operation and nothing else. So, in particular, there is no assumption of associativity, and there is no unit. A binary operation is simply a function from the cartesian product of a with itself back to a

a × a -> a

It takes a pair of elements <x, y>, both coming from the set a, and maps it to an element of a.

It’s tempting to quote the Haskell definition of a magma:

class Magma a where
(<>) :: a -> a -> a

but this definition is already tainted with some higher concepts like currying. An alternative would be:

class Magma a where
(<>) :: (a, a) -> a

Here, we at least see a pair of elements that are being “multiplied.” But the pair type (a, a) is also a higher-level concept. I’ll come back to it later.

Lack of associativity means that we cannot identify (x<>y)<>z with x<>(y<>z). You have to keep the parentheses.

You might have heard of quaternions — their multiplication is associative. But not many people have heard of octonions, which are not associative. In fact Hamilton, who discovered quaternions, invented the word associative to disassociate himself from octonions, which are not.

If you’re familiar with continuous groups, you might know that Lie algebras are not associative.

Closer to home — most operations on floating-point numbers are not associative on modern computers because of rounding errors.

But, really, most interesting binary operations are associative. So out of the magma emerges a semigroup. In a semigroup you can drop parentheses. A non-trivial (that is, non-monoidal) example of a semigroup is the set of integers with max binary operation. A maximum of three numbers is the same no matter in which order you pair them. But there is no integer that’s less or equal to any other integer, so this is not a monoid.

## Monoids of the Archean Eon

But, really, most interesting binary operations are both associative and unital. There usually is a “do nothing” element with respect to most binary operations. So life as we know it begins with a monoid.

A monoid is a set with a binary operation that is associative, and with a special element called the unit e that is neutral with respect to the binary operation. To be precise, these are the three monoid laws:

(x <> y) <> z = x <> (y <> z)
e <> x = x
x <> e = x

In Haskell, the traditional definition of a monoid uses mempty for the unit and mappend for the binary operation:

class Monoid a where
mempty  :: a
mappend :: a -> a -> a

As with the magma, the definition of mappend is curried. Equivalently, it could have been written as:

mappend :: (a, a) -> a

I’ll come back to this point later.

There are plenty of examples of monoids. Non-negative integers with addition, or positive integers with multiplication are the obvious ones. Strings with concatenation are interesting too, because concatenation is not commutative.

Just like pairs of elements from two sets a and b organize themselves into a set a×b, which is their cartesian product; functions between two sets organize themselves into a set — the set of functions from a to b, which we sometimes write as a->b.

This organizing principle is characteristic of sets, where everything you can think of is a set. Except when it’s more than just a set — for instance when you try to organize all sets into one large collection. This collection, or “class,” is not itself a set. You can’t have a set of all sets, but you can have a category Set of “small” sets, which are sets that belong to a “universe.” In what follows, I will confine myself to a single universe in order to dodge questions from foundational mathematicians.

Let’s now pop one level up and look at cartesian product as an operation on sets. For any two sets a and b, we can construct the set a×b. If we view this as “multiplication” of sets, we can say that sets form a magma. But do they form a monoid? Not exactly! To begin with, cartesian product is not associative. We can see it in Haskell: the type ((a, b), c) is not the same as the type (a, (b, c)). They are, however, isomorphic. There is an invertible function called the associator, from one type to the other:

alpha :: ((a, b), c) -> (a, (b, c))
alpha ((x, y), z) = (x, (y, z))

It’s just a repackaging of containers (such repackaging is, by the way, called a natural transformation).

For the unit of this “multiplication” we can pick the singleton set. In Haskell, this is the type called unit and it’s denoted by an empty pair of parentheses (). Again, the unit laws are valid up to isomorphism. There are two such isomorphisms called left and right unitors:

lambda :: ((), a) -> a
lambda ((), x) = x
rho :: (a, ()) -> a
rho (x, ()) -> x

We have just exposed monoidal structure in the category Set. Set is not strictly a monoid because monoidal laws are satisfied only up to isomorphism.

There is another monoidal structure in Set. Just like cartesian product resembles multiplication, there is an operation on sets that resembles addition. It’s called disjoint sum. In Haskell it’s embodied in the type Either a b . Just like cartesian product, disjoint sum is associative up to isomorphism. The unit (or the “zero”) of this sum type is the empty set or, in Haskell, the Void type — also up to isomorphism.

## The Cambrian Explosion of Categories

The first rule of abstraction is, You do not talk about Fight Club. In the category Set, for instance, we are not supposed to admit that sets have elements. An object in Set is really a set, but you never talk about its elements. We still have functions between sets, but they become abstract morphisms, of which we only know how they compose.

Composition of functions is associative, and there is an identity function for every set, which serves as a unit of composition. We can write these rules compactly as:

(f ∘ g) ∘ h = f ∘ (g ∘ h)
id ∘ f = f
f ∘ id = f

These look exactly like monoid laws. So do functions form a monoid with respect to composition? Not quite, because you can’t compose any two functions. They must be composable, which means their endpoints have to match. In Haskell, we can compose g after f, or g ∘ f, only if:

f :: a -> b
g :: b -> c

Also, there is no single identity function, but a whole family of functions ida, one for each set a. In Haskell, we call that a polymorphic function.

But notice what happens if we restrict ourselves to just a single object a in Set. Every morphism from a back to a can be composed with any other such morphism (their endpoints always match). Moreover, we are guaranteed that among those so called endomorphisms there is one identity morphism ida, which acts as a unit of composition.

Notice that I switched from the set/function nomenclature to the more general object/morphism naming convention of category theory. We can now forget about sets and functions and define an arbitrary category as a collection (a set in a given universe) of objects, and sets of morphisms that go between them. The only requirements are that any two composable morphisms compose, and that there is an identity morphism for every object. And that composition must be associative.

We can now forget about sets and define a monoid as a category that has only one object. The binary operation is just the composition of (endo-)morphisms. It works! We have defined a monoid without a set. Or have we?

No, we haven’t! We have just swept it under the rug — the rug being the set of morphisms. Yes, morphisms between any two objects form a set called the hom-set. In a category C, the hom-set between objects a and b is denoted by C(a, b). So we haven’t completely eliminated sets from the picture.

In the single object category M, we have only one hom-set M(a, a). The elements of this set — and we are allowed to call them elements because it’s a set — are morphisms like f and g. We can compose them, and we can call this composition “multiplication,” thus recovering our previous definition of the monoid as a set. We get associativity for free, and we have the identity morphism ida serving as the unit.

It might seem at first that we haven’t made progress and, in fact, we might have made some things more complicated by forgetting the internal structure of objects. For instance, in the category Set, it’s no longer obvious what an empty set is. You can’t say it’s a set with no elements because of the Fight Club rule. Similarly with the singleton set. Fortunately, it turns out that both these sets can be uniquely described in terms of their interactions with other sets. By that I mean the kind of functions/morphisms that connect them to other objects in Set. These object-opaque definitions are called universal constructions. For instance, the empty set is the only set that has a unique morphism going from it to every other set. The advantage of this characterization is that it can now be applied to any category. One may ask this question in any category: Is there an object that has this property? If there is, we call it the initial object. The empty set is the initial object in Set. Similarly, a singleton set is the terminal object in Set (and it’s unique up to unique isomorphism).

A cartesian product of two sets can also be defined using a universal construction, one which doesn’t mention elements (or pairs of elements). And again, this construction may be used to define a (categorical) product in other categories. Of particular interest are categories where a product exists for every pair of objects (it does in Set).

In such categories there is actually an even better way of defining a product using an adjunction. But before we can get to adjunctions, let me summarize a few millions of years of evolution in a few terse paragraphs.

A functor is a mapping of categories that preserves their structure. It maps objects to objects and morphisms to morphisms. In Haskell we define a functor (really, an endofunctor) as a type constructor f (a mapping from types to types) that can be lifted to functions that go between these types:

class Functor f where
fmap :: (a -> b) -> (f a -> f b)

The mapping of morphisms must also preserve composition and identity. Functors may collapse multiple objects into one, and multiple morphisms into one, but they never break connections. You may also think of functors as embedding one category inside another.

Finally, functors can be composed in the obvious way, and there is an identity endofunctor that maps a category onto itself. It follows that categories (at least the small ones) form a category Cat in which functors serve as morphisms.

There may be many ways of embedding one category inside another, and it’s extremely useful to be able to compare such embeddings by defining mappings between them. If we have two functors F and G between two categories C and D we define a natural transformation between these functors by picking a morphism between a pair F a and G a, for every a.

In Haskell, a natural transformation between two functors f and g is a polymorphic function:

type Nat f g = forall a. f a -> g a

Natural transformations may be composed, and there is an identity natural transformations from any functor to itself. It follows that functors between any two categories C and D form a category denoted by [C, D], where natural transformations play the role of morphisms. A hom-set in such a category is a set of natural transformations between two functors F and G denoted by [C, D](F, G).

An invertible natural transformation is called a natural isomorphism. If two functors are naturally isomorphic they are essentially the same.

Using a pair of functors that are the inverse of each other we may define equivalence of categories, but there is an even more useful concept of adjoint functors that compare the structures of two non-equivalent categories. The idea is that we have a “right” functor R going from category C to D and a “left” functor L going in the other direction, from D to C.

There are two possible compositions of these functors, both resulting in round trips or endofunctors. The categories would be equivalent if those endofunctors were naturally isomorphic to identity endofunctors. But for an adjunction, we impose weaker conditions. We require that there be two natural transformations (not necessarily isomorphisms):

η :: ID -> R ∘ L
ε :: L ∘ R -> IC

The first transformation η is called the unit; and the second ε, the counit of the adjunction.

In a small category objects form sets, so it’s possible to form a cartesian product of two small categories C and D. Object in such a category C×D are pairs of objects <c, d>, and morphisms are pairs of morphisms <f, g>.

After these preliminaries, we are ready to define the categorical product in C using an adjunction. We chose C×C as the left category. The left functor is the diagonal functor Δ that maps any object c to a pair <c, c> and any morphism f to a pair of morphisms <f, f>. Its right adjoint, if it exists, maps a pair of objects <a, b> to their categorical product a×b.

Interestingly, the terminal object can also be defined using an adjunction. This time we chose, as the left category, a singleton category with one object and one (identity) morphism. The left functor maps any object c to the singleton object. Its right adjoint, if it exists, maps the singleton object to the terminal object in C.

A category with all products and the terminal object is called a cartesian category, or cartesian monoidal category. Why monoidal? Because the operation of taking the categorical product is monoidal. It’s associative, up to isomorphism; and its unit is the terminal object.

Incidentally, this is the same monoidal structure that we’ve seen in Set, but now it’s generalized to the level of other categories. There was another monoidal structure in Set induced by the disjoint sum. Its categorical generalization is given by the coproduct, with the initial object playing the role of the unit.

But what about the set of morphisms? In Set, morphisms between two sets a and b form a hom-set, which is the object of the same category Set. In an arbitrary category C, a hom-set C(a, b) is still a set — but now it’s not an object of C. That’s why it’s called the external hom-set. However, there are categories in which each external hom-set has a corresponding object called the internal hom. This object is also called an exponential, ba. It can be defined using an adjunction, but only if the category supports products. It’s an adjunction in which the left and right categories are the same. The left endofunctor takes an object b and maps it to a product b×a, where a is an arbitrary fixed object. Its adjoint functor maps an object b to the exponential ba. The counit of this adjunction:

ε :: ba × a -> b

is the evaluation function. In Haskell it has the following signature:

eval :: (a -> b, a) -> b

The Haskell function type a->b is equivalent to the exponential ba.

A category that has all products and exponentials together with the terminal object is called cartesian closed. Cartesian closed categories, or CCCs, play an important role in the semantics of programming languages.

## Tensorosaurus Rex

We have already seen two very similar monoidal structures induced by products and coproducts. In mathematics, two is a crowd, so let’s look for a pattern. Both product and coproduct act as bifunctors C×C->C. Let’s call such a bifunctor a tensor product and write it as an infix operator a ⊗ b. As a bifunctor, the tensor product can also lift pairs of morphisms:

f :: a -> a'
g :: b -> b'
f ⊗ g :: a ⊗ b -> a' ⊗ b'

To define a monoid on top of a tensor product, we will require that it be associative — up to isomorphism:

α :: (a ⊗ b) ⊗ c -> a ⊗ (b ⊗ c)

We also need a unit object, which we will call i. The two unit laws are:

λ :: i ⊗ a -> a
ρ :: a ⊗ i -> a

A category with a tensor product that satisfies the above properties, plus some additional coherence conditions, is called a monoidal category.

We can now specialize the tensor product to categorical product, in which case the unit object is the terminal object; or to coproduct, in which case we chose the initial object as the unit. But there is an even more interesting operation that has all the properties of the tensor product. I’m talking about functor composition.

## Functorosaurus

Functors between any two categories C and D form a functor category [C, D] with natural transformations playing the role of morphisms. In general, these functors don’t compose (their endpoints don’t match) unless we pick the target category to be the same as the source category.

### Endofunctor Composition

In the endofunctor category [C, C] any two functors can be composed. But in [C, C] functors are objects, so functor composition becomes an operation on objects. For any two endofunctors F and G it produces a new endofunctor F∘G. It’s a binary operation, so it’s a potential candidate for a tensor product. Indeed, it is a bifunctor: it can be lifted to natural transformations, which are morphisms in [C, C]. It’s associative — in fact it’s strictly associative, the associator α is the identity natural transformation. The unit with respect to endofunctor composition is the identity functor I. So the category of endofunctors is a monoidal category.

Unlike product and coproduct, which are symmetric up to isomorphism, endofunctor composition is not symmetric. In general, there is no relation between F∘G and G∘F.

### Profunctor Composition

Different species of functors came up with their own composition strategies. Take for instance the profunctors, which are functors Cop×D->Set. They generalize the idea of relations between objects in C and D. The sets they map to may be thought of as sets of proofs of the relationship. An empty set means that the two objects are not related. If you want to compose two relations, you have to find an element that’s common to the image of one relation and the source of the other (relations are not, in general, symmetric). The proofs of the new composite relation are pairs of proofs of individual relations. Symbolically, if p and q are such profunctors/relations, their composition can be written as:

exists x. (p a x, q x b)

Existential quantification in Haskell translates to polymorphic construction, so the actual definition is:

data PCompose p q a b = forall x . PCompose (p a x) (q x b)

In category theory, existential quantification is encoded as the coend, which is a generalization of a colimit for profunctors. The coend formula for the composition of two profunctors reads:

(p ⊗ q) a b = ∫ z p a z × q z b

The product here is the cartesian product of sets.

Profunctors, being functors, form a category in which morphisms are natural transformations. As long as the two categories that they relate are the same, any two profunctors can be composed using a coend. So profunctor composition is a good candidate for a tensor product in such a category. It is indeed associative, up to isomorphism. But what’s the unit of profunctor composition? It turns out that the simplest profuctor — the hom-functor — because of the Yoneda lemma, is the unit of composition:

∫ z C(a, z) × p z b ≅ p a b
∫ z p a z × C(z, b) ≅ p a b

Thus profunctors Cop×C->Set form a monoidal category.

### Day Convolution

Or consider Set-valued functors. They can be composed using Day convolution. For that, the category C must itself be monoidal. Day convolution of two functors C->Set is defined using a coend:

(f ★ g) a = ∫ x y f x × g y × C(x ⊗ y, a)

Here, the tensor product of x ⊗ y comes from the monoidal category C, the other products are just cartesian products of sets (one of them being the hom-set).

As before, in Haskell, the coend turns into existential quantifier, which can be written symbolically:

Day f g a = exists x y. ((f x, g y), (x, y) -> a)

and encoded as a polymorphic constructor:

data Day f g a = forall x y. Day (f x) (g y) ((x, y) -> a)

We use the fact that the category of Haskell types is monoidal with respect to cartesian product.

We can build a monoidal category based on Day convolution. The unit with respect to Day convolution is C(i, -), the hom-functor applied to i — the unit in the monoidal category C. For instance, the left identity can be derived from:

(C(i, -) ★ g) a = ∫ x y C(i, x) × g y × C(x ⊗ y, a)

Applying the Yoneda lemma, or “integrating over x,” we get:

∫y g y × C(i ⊗ y, a)

Considering that i is the unit of the tensor product, we can perform the second integration to get g a.

## The Monozoic Era

Monoidal categories are important because they provide rich grazing grounds for monoids. In a monoidal category we can define a more general monoid. It’s an object m with some special properties. These properties replace the usual definitions of multiplication and unit.

First, let’s reformulate the definition of a set-based monoid, taking into account the fact that Set is a monoidal category with respect to cartesian product.

A monoid is a set, so it’s an object in Set — let’s call it m. Multiplication maps pairs of elements of m back to m. These pairs are just elements of the cartesian product m × m. So multiplication is defined as a function:

μ :: m × m -> m

Unit of multiplication is a special element of m. We can select this element by providing a special morphism from the singleton set to m:

η :: () -> m

We can now express associativity and unit laws as properties of these two functions. The beauty of this formulation is that it generalizes easily to any cartesian category — just replace functions with morphisms and the unit () with the terminal object. There’s no reason to stop there: we can lift this definition all the way up to a monoidal category.

A monoid in a monoidal category is an object m together with two morphisms:

μ :: m ⊗ m -> m
η :: i -> m

Here i is the unit object with respect to the tensor product ⊗. Monoidal laws can be expressed using the associator α and the two unitors, λ and ρ, of the monoidal category:

Having previously defined several interesting monoidal categories, we can now go digging for new monoids.

Let’s start with the category of endofunctors where the tensor product is functor composition. A monoid in the category of endofunctors is an endofunctor m and two morphism. Remember that morphisms in a functor category are natural transformations. So we end up with two natural transformations:

μ :: m ∘ m -> m
η :: I -> m

where I is the identity functor. Their components at an object a are:

μa :: m (m a) -> m a
ηa :: a -> m a

This construct is easily recognizable as a monad. The associativity and unit laws are just monad laws. In Haskell, μa is called join and ηa is called return.

### Arrows

Let’s switch to the category of profunctors Cop×C->Set with profunctor composition as the tensor product. A monoid in that category is a profunctor ar. Multiplication is defined by a natural transformation:

μ :: ar ⊗ ar -> ar

Its component at a, b is:

μa b :: (∫ z ar a z × ar z b) -> ar a b

To simplify this formula we need a very useful identity that relates coends to ends. A hom-set that starts at a coend is equivalent to an end of the hom set:

C(∫ z p z z, y) ≅ ∫ z C(p z z, y)

Or, replacing external hom-sets with internal homs:

(∫ z p z z) -> y ≅ ∫ z (p z z -> y)

In Haskell, this formula is used to turn functions that take existential types to functions that are polymorphic:

(exists z. p z z) -> y ≅ forall z. (p z z -> y)

Intuitively, it makes perfect sense. If you want to define a function that takes an existential type, you have to be prepared to handle any type.

Using that identity, our multiplication formula can be rewritten as:

μa b :: ∫ z ((ar a z × ar z b) -> ar a b)

In Haskell, this derivation uses the existential quantifier:

mu a b = (exists z. (ar a z, ar z b)) -> ar a b

As we discussed, a function from an existential type is equivalent to a polymorphic function:

forall z. (ar a z, ar z b) -> ar a b

or, after currying and dropping the redundant quantification:

ar a z -> ar z b -> ar a b

This looks very much like a composition of morphisms in a category. In Haskell, this function is known in the infix-operator form as:

(>>>) :: ar a z -> ar z b -> ar a b

Let’s see what we get as the monoidal unit. Remember that the unit object in the profunctor category is the hom-functor C(a, b).

ηa b :: C(a, b) -> ar a b

In Haskell, this polymorphic function is traditionally called arr:

arr :: (a -> b) -> ar a b

The whole construct is known in Haskell as a pre-arrow. The full arrow is defined as a monoid in the category of strong profunctors, with strength defined as a natural transformation:

sta b :: p a b -> p (a, x) (b, x)

In Haskell, this function is called first.

### Applicatives

There are several categorical formulations of what’s called in Haskell the applicative functor. To first approximaton, Haskell’s type system is the category Set. To translate Haskell constructs to category theory, the safest approach is to just play with endofunctors in Set. But both Set and its endofunctors have a lot of extra structure, so I’d like to start in a slightly more general setting.

Let’s have a look at the monoidal category of functors [C, Set], with Day convolution as the tensor product, and C(i, -) as unit. A monoid in this category is a functor f with multiplication given by the natural transformation:

μ :: f ★ f -> f

and unit given by:

η :: C(i, -) -> f

It turns out that the existence of these two natural transformations is equivalent to the requirement that f be a lax monoidal functor, which is the basis of the definition of the applicative functor in Haskell.

A monoidal functor is a functor that maps monoidal structure of one category to the monoidal structure of another category. It maps the tensor product, and it maps the unit object. In our case, the source category C has the monoidal structure given by the tensor product ⊗, and the target category Set is monoidal with respect to the cartesian product ×. A functor is monoidal if it doesn’t matter whether we first map two object and then multiply them, or first multiply them and then map the result:

f x × f y ≅ f (x ⊗ y)

Also, the unit object in Set should be isomporphic to the result of mapping the unit object in C:

() ≅ f i

Here, () is the terminal object in Set and i is the unit object in C.

These conditions are relaxed in the definition of a lax monoidal functor. A lax monoidal functor replaces isomorphisms with regular unidirectional morphisms:

f x × f y -> f (x ⊗ y)
() -> f i

It can be shown that the monoid in the category[C, Set], with Day convolution as the tensor product, is equivalent to the lax monoidal functor.

The Haskell definition of Applicative doesn’t look like Day convolution or like a lax monoidal functor:

class Functor f => Applicative f where
(<*>) :: f (a -> b) -> (f a -> f b)
pure :: a -> f a

You may recognize pure as a component of η, the natural transformation defining the monoid with respect to Day convolution. When you replace the category C with Set, the unit object C(i, -) turns into the identity functor. However, the operator <*> is lifted from the definition of yet another lax functor, the lax closed functor. It’s a functor that preserves the closed structure defined by the internal hom functor. In Set, the internal hom functor is just the arrow (->), hence the definition:

class Functor f => Closed f where
(<*>) :: f (a -> b) -> (f a -> f b)
unit :: f ()

As long as the internal hom is defined through the adjunction with the product, a lax closed functor is equivalent to a lax monoidal functor.

## Conclusion

It is pretty shocking to realize how many different animals share the same body plan — I’m talking here about the monoid as the skeleton of a myriad of different mathematical and programming constructs. And I haven’t even touched on the whole kingdom of enriched categories, where monoidal categories form the reservoir of hom-objects. Virtually all notions I’ve discussed here can be generalized to enriched categories, including functors, profunctors, the Yoneda lemma, Day convolution, and so on.

## Glossary

• Hadean Eon: Began with the formation of the Earth about 4.6 billion years ago. It’s the period before the earliest-known rocks.
• Archean Eon: During the Archean, the Earth’s crust had cooled enough to allow the formation of continents.
• Cambrian explosion: Relatively short evolutionary event, during which most major animal phyla appeared.
• Arthropods: from Greek ἄρθρωσις árthrosis, “joint”
• Tensor, from Latin tendere “to stretch”
• Functor: from Latin fungi, “perform”

## Bibliograhpy

1. Moggi, Notions of Computation and Monads.
2. Rivas, Jaskelioff, Notions of Computation as Monoids.

Unlike monads, which came into programming straight from category theory, applicative functors have their origins in programming. McBride and Paterson introduced applicative functors as a programming pearl in their paper Applicative programming with effects. They also provided a categorical interpretation of applicatives in terms of strong lax monoidal functors. It’s been accepted that, just like “a monad is a monoid in the category of endofunctors,” so “an applicative is a strong lax monoidal functor.”

The so called “tensorial strength” seems to be important in categorical semantics, and in his seminal paper Notions of computation and monads, Moggi argued that effects should be described using strong monads. It makes sense, considering that a computation is done in a context, and you should be able to make the global context available under the monad. The fact that we don’t talk much about strong monads in Haskell is due to the fact that all functors in the category Set, which underlies the Haskell’s type system, have canonical strength. So why do we talk about strength when dealing with applicative functors? I have looked into this question and came to the conclusion that there is no fundamental reason, and that it’s okay to just say:

An applicative is a lax monoidal functor

In this post I’ll discuss different equivalent categorical definitions of the applicative functor. I’ll start with a lax closed functor, then move to a lax monoidal functor, and show the equivalence of the two definitions. Then I’ll introduce the calculus of ends and show that the third definition of the applicative functor as a monoid in a suitable functor category equipped with Day convolution is equivalent to the previous ones.

## Applicative as a Lax Closed Functor

class Functor f => Applicative f where
(<*>) :: f (a -> b) -> (f a -> f b)
pure :: a -> f a

At first sight it doesn’t seem to involve a monoidal structure. It looks more like preserving function arrows (I added some redundant parentheses to suggest this interpretation).

Categorically, functors that “preserve arrows” are known as closed functors. Let’s look at a definition of a closed functor f between two categories C and D. We have to assume that both categories are closed, meaning that they have internal hom-objects for every pair of objects. Internal hom-objects are also called function objects or exponentials. They are normally defined through the right adjoint to the product functor:

C(z × a, b) ≅ C(z, a => b)

To distinguish between sets of morphisms and function objects (they are the same thing in Set), I will temporarily use double arrows for function objects.

We can take a functor f and act with it on the function object a=>b in the category C. We get an object f (a=>b) in D. Or we can map the two objects a and b from C to D and then construct the function object in D: f a => f b.

We call a functor closed if the two results are isomorphic (I have subscripted the two arrows with the categories where they are defined):

f (a =>C b) ≅ (f a =>D f b)

and if the functor preserves the unit object:

iD ≅ f iC

What’s the unit object? Normally, this is the unit with respect to the same product that was used to define the function object using the adjunction. I’m saying “normally,” because it’s possible to define a closed category without a product.

Note: The two arrows and the two is are defined with respect to two different products. The first isomorphism must be natural in both a and b. Also, to complete the picture, there are some diagrams that must commute.

The two isomorphisms that define a closed functor can be relaxed and replaced by unidirectional morphisms. The result is a lax closed functor:

f (a => b) -> (f a => f b)
i -> f i

This looks almost like the definition of Applicative, except for one problem: how can we recover the natural transformation we call pure from a single morphism i -> f i.

One way to do it is from the position of strength. An endofunctor f has tensorial strength if there is a natural transformation:

stc a :: c ⊗ f a -> f (c ⊗ a)

Think of c as the context in which the computation f a is performed. Strength means that we can use this external context inside the computation.

In the category Set, with the tensor product replaced by cartesian product, all functors have canonical strength. In Haskell, we would define it as:

st (c, fa) = fmap ((,) c) fa

The morphism in the definition of the lax closed functor translates to:

unit :: () -> f ()

Notice that, up to isomorphism, the unit type () is the unit with respect to cartesian product. The relevant isomorphisms are:

λa :: ((), a) -> a
ρa :: (a, ()) -> a

Here’s the derivation from Rivas and Jaskelioff’s Notions of Computation as Monoids:

    a
≅  (a, ())   -- unit law, ρ-1
-> (a, f ()) -- lax unit
-> f (a, ()) -- strength
≅  f a       -- lifted unit law, f ρ

Strength is necessary if you’re starting with a lax closed (or monoidal — see the next section) endofunctor in an arbitrary closed (or monoidal) category and you want to derive pure within that category — not after you restrict it to Set.

There is, however, an alternative derivation using the Yoneda lemma:

f ()
≅ forall a. (() -> a) -> f a  -- Yoneda
≅ forall a. a -> f a -- because: (() -> a) ≅ a

We recover the whole natural transformation from a single value. The advantage of this derivation is that it generalizes beyond endofunctors and it doesn’t require strength. As we’ll see later, it also ties nicely with the Day-convolution definition of applicative. The Yoneda lemma only works for Set-valued functors, but so does Day convolution (there are enriched versions of both Yoneda and Day convolution, but I’m not going to discuss them here).

We can define the categorical version of the Haskell’s applicative functor as a lax closed functor going from a closed category C to Set. It’s a functor equipped with a natural transformation:

f (a => b) -> (f a -> f b)

where a=>b is the internal hom-object in C (the second arrow is a function type in Set), and a function:

1 -> f i

where 1 is the singleton set and i is the unit object in C.

The importance of a categorical definition is that it comes with additional identities or “axioms.” A lax closed functor must be compatible with the structure of both categories. I will not go into details here, because we are really only interested in closed categories that are monoidal, where these axioms are easier to express.

The definition of a lax closed functor is easily translated to Haskell:

class Functor f => Closed f where
(<*>) :: f (a -> b) -> f a -> f b
unit :: f ()

## Applicative as a Lax Monoidal Functor

Even though it’s possible to define a closed category without a monoidal structure, in practice we usually work with monoidal categories. This is reflected in the equivalent definition of the Haskell’s applicative functor as a lax monoidal functor. In Haskell, we would write:

class Functor f => Monoidal f where
(>*<) :: (f a, f b) -> f (a, b)
unit :: f ()

This definition is equivalent to our previous definition of a closed functor. That’s because, as we’ve seen, a function object in a monoidal category is defined in terms of a product. We can show the equivalence in a more general categorical setting.

This time let’s start with a symmetric closed monoidal category C, in which the function object is defined through the right adjoint to the tensor product:

C(z ⊗ a, b) ≅ C(z, a => b)

As usual, the tensor product is associative and unital — with the unit object i — up to isomorphism. The symmetry is defined through natural isomorphism:

γ :: a ⊗ b -> b ⊗ a

A functor f between two monoidal categories is lax monoidal if there exist: (1) a natural transformation

f a ⊗ f b -> f (a ⊗ b)

and (2) a morphism

i -> f i

Notice that the products and units on either side of the two mappings are from different categories.

A (lax-) monoidal functor must also preserve associativity and unit laws.

For instance a triple product

f a ⊗ (f b ⊗ f c)

may be rearranged using an associator α to give

(f a ⊗ f b) ⊗ f c

then converted to

f (a ⊗ b) ⊗ f c

and then to

f ((a ⊗ b) ⊗ c)

Or it could be first converted to

f a ⊗ f (b ⊗ c)

and then to

f (a ⊗ (b ⊗ c))

These two should be equivalent under the associator in C.

Similarly, f a ⊗ i can be simplified to f a using the right unitor ρ in D. Or it could be first converted to f a ⊗ f i, then to f (a ⊗ i), and then to f a, using the right unitor in C. The two paths should be equivalent. (Similarly for the left identity.)

We will now consider functors from C to Set, with Set equipped with the usual cartesian product, and the singleton set as unit. A lax monoidal functor is defined by: (1) a natural transformation:

(f a, f b) -> f (a ⊗ b)

and (2) a choice of an element of the set f i (a function from 1 to f i picks an element from that set).

We need the target category to be Set because we want to be able to use the Yoneda lemma to show equivalence with the standard definition of applicative. I’ll come back to this point later.

## The Equivalence

The definitions of a lax closed and a lax monoidal functors are equivalent when C is a closed symmetric monoidal category. The proof relies on the existence of the adjunction, in particular the unit and the counit of the adjunction:

ηa :: a -> (b => (a ⊗ b))
εb :: (a => b) ⊗ a -> b

For instance, let’s assume that f is lax-closed. We want to construct the mapping

(f a, f b) -> f (a ⊗ b)

First, we apply the lifted pair (unit, identity), (f η, f id)

(f a -> f (b => a ⊗ b), f id)

to the left hand side. We get:

(f (b => a ⊗ b), f b)

Now we can use (the uncurried version of) the lax-closed morphism:

(f (b => x), f b) -> f x

to get:

f (a ⊗ b)

Conversely, assuming the lax-monoidal property we can show that the functor is lax-closed, that is to say, implement the following function:

(f (a => b), f a) -> f b

First we use the lax monoidal morphism on the left hand side:

f ((a => b) ⊗ a)

and then use the counit (a.k.a. the evaluation morphism) to get the desired result f b

There is yet another presentation of applicatives using Day convolution. But before we get there, we need a little refresher on calculus.

## Calculus of Ends

Ends and coends are very useful constructs generalizing limits and colimits. They are defined through universal constructions. They have a few fundamental properties that are used over and over in categorical calculations. I’ll just introduce the notation and a few important identities. We’ll be working in a symmetric monoidal category C with functors from C to Set and profunctors from Cop×C to Set. The end of a profunctor p is a set denoted by:

∫a p a a

The most important thing about ends is that a set of natural transformations between two functors f and g can be represented as an end:

[C, Set](f, g) = ∫a C(f a, g a)

In Haskell, the end corresponds to universal quantification over a functor of mixed variance. For instance, the natural transformation formula takes the familiar form:

forall a. f a -> g a

The Yoneda lemma, which deals with natural transformations, can also be written using an end:

∫z (C(a, z) -> f z) ≅ f a

In Haskell, we can write it as the equivalence:

forall z. ((a -> z) -> f z) ≅ f a

which is a generalization of the continuation passing transform.

The dual notion of coend is similarly written using an integral sign, with the “integration variable” in the superscript position:

∫ a p a a

In pseudo-Haskell, a coend is represented by an existential quantifier. It’s possible to define existential data types in Haskell by converting existential quantification to universal one. The relevant identity in terms of coends and ends reads:

(∫ z p z z) -> y ≅ ∫ z (p z z -> y)

In Haskell, this formula is used to turn functions that take existential types to functions that are polymorphic:

(exists z. p z z) -> y ≅ forall z. (p z z -> y)

Intuitively, it makes perfect sense. If you want to define a function that takes an existential type, you have to be prepared to handle any type.

The equivalent of the Yoneda lemma for coends reads:

∫ z f z × C(z, a) ≅ f a

exists z. (f z, z -> a) ≅ f a

(The intuition is that the only thing you can do with this pair is to fmap the function over the first component.)

There is also a contravariant version of this identity:

∫ z C(a, z) × f z ≅ f a

where f is a contravariant functor (a.k.a., a presheaf). In pseudo-Haskell:

exists z. (a -> z, f z) ≅ f a

(The intuition is that the only thing you can do with this pair is to apply the contramap of the first component to the second component.)

Using coends we can define a tensor product in the category of functors [C, Set]. This product is called Day convolution:

(f ★ g) a = ∫ x y f x × g y × C(x ⊗ y, a)

It is a bifunctor in that category (read, it can be used to lift natural transformations). It’s associative and symmetric up to isomorphism. It also has a unit — the hom-functor C(i, -), where i is the monoidal unit in C. In other words, Day convolution imbues the category [C, Set] with monoidal structure.

Let’s verify the unit laws.

(C(i, -) ★ g) a = ∫ x y C(i, x) × g y × C(x ⊗ y, a)

We can use the contravariant Yoneda to “integrate over x” to get:

∫ y g y × C(i ⊗ y, a)

Considering that i is the unit of the tensor product in C, we get:

∫ y g y × C(y, a)

Covariant Yoneda lets us “integrate over y” to get the desired g a. The same method works for the right unit law.

## Applicative as a Monoid

Given a monoidal category, we can always define a monoid as an object m equipped with two morphisms:

μ :: m ⊗ m -> m
η :: i -> m

satisfying the laws of associativity and unitality.

We have shown that the functor category [C, Set] (with C a symmetric monoidal category) is monoidal under Day convolution. An object in this category is a functor f. The two morphisms that would make it a candidate for a monoid are natural transformations:

μ :: f ★ f -> f
η :: C(i, -) -> f

The a component of the natural transformation μ can be rewritten as:

(∫ x y f x × f y × C(x ⊗ y, a)) -> f a

which is equivalent to:

∫x y (f x × f y × C(x ⊗ y, a) -> f a)

or, upon currying:

∫x y (f x, f y) -> C(x ⊗ y, a) -> f a

It turns out that so defined monoid is equivalent to a lax monoidal functor. This was shown by Rivas and Jaskelioff. The following derivation is due to Bob Atkey.

The trick is to start with the whole set of natural transformation from f★f to f. The multiplication μ is just one of them. We’ll express the set of natural transformations as an end:

∫ a ((f ★ f) a -> f a)

Plugging in the formula for the a component of μ, we get:

∫ a x y (f x, f y) -> C(x ⊗ y, a) -> f a

The end over a does not involve the first argument, so we can move the integral sign:

∫ x y (f x, f y) -> ∫ a C(x ⊗ y, a) -> f a

Then we use the Yoneda lemma to “perform the integration” over a:

∫ x y (f x, f y) -> f (x ⊗ y)

You may recognize this as a set of natural transformations that define a lax monoidal functor. We have established a one-to-one correspondence between these natural transformations and the ones defining monoidal multiplication using Day convolution.

The remaining part is to show the equivalence between the unit with respect to Day convolution and the second part of the definition of the lax monoidal functor, the morphism:

1 -> f i

We start with the set of natural transformations that contains our η:

∫ a (i -> a) -> f a

By Yoneda, this is just f i. Picking an element from a set is equivalent to defining a morphism from the singleton set 1, so for any choice of η we get:

1 -> f i

and vice versa. The two definitions are equivalent.

Notice that the monoidal unit η under Day convolution becomes the definition of pure in the Haskell version of applicative. Indeed, when we replace the category C with Set, f becomes and endofunctor, and the unit of Day convolution C(i, -) becomes the identity functor Id. We get:

η :: Id -> f

or, in components:

pure :: a -> f a

So, strictly speaking, the Haskell definition of Applicative mixes the elements of the lax closed functor and the monoidal unit under Day convolution.

## Acknowledgments

I’m grateful to Mauro Jaskelioff and Exequiel Rivas for correspondence and to Bob Atkey, Dimitri Chikhladze, and Make Shulman for answering my questions on Math Overflow.

Now that we have covered monads, we can reap the benefits of duality and get comonads for free simply by reversing the arrows and working in the opposite category.

Recall that, at the most basic level, monads are about composing Kleisli arrows:

a -> m b

where m is a functor that is a monad. If we use the letter w (upside down m) for the comonad, we can define co-Kleisli arrows as morphism of the type:

w a -> b

The analog of the fish operator for co-Kleisli arrows is defined as:

(=>=) :: (w a -> b) -> (w b -> c) -> (w a -> c)

For co-Kleisli arrows to form a category we also have to have an identity co-Kleisli arrow, which is called extract:

extract :: w a -> a

This is the dual of return. We also have to impose the laws of associativity as well as left- and right-identity. Putting it all together, we could define a comonad in Haskell as:

class Functor w => Comonad w where
(=>=) :: (w a -> b) -> (w b -> c) -> (w a -> c)
extract :: w a -> a

In practice, we use slightly different primitives, as we’ll see shortly.

The question is, what’s the use for comonads in programming?

Let’s compare the monad with the comonad. A monad provides a way of putting a value in a container using return. It doesn’t give you access to a value or values stored inside. Of course, data structures that implement monads might provide access to their contents, but that’s considered a bonus. There is no common interface for extracting values from a monad. And we’ve seen the example of the IO monad that prides itself in never exposing its contents.

A comonad, on the other hand, provides the means of extracting a single value from it. It does not give the means to insert values. So if you want to think of a comonad as a container, it always comes pre-filled with contents, and it lets you peek at it.

Just as a Kleisli arrow takes a value and produces some embellished result — it embellishes it with context — a co-Kleisli arrow takes a value together with a whole context and produces a result. It’s an embodiment of contextual computation.

Remember the reader monad? We introduced it to tackle the problem of implementing computations that need access to some read-only environment e. Such computations can be represented as pure functions of the form:

(a, e) -> b

We used currying to turn them into Kleisli arrows:

a -> (e -> b)

But notice that these functions already have the form of co-Kleisli arrows. Let’s massage their arguments into the more convenient functor form:

data Product e a = Prod e a
deriving Functor

We can easily define the composition operator by making the same environment available to the arrows that we are composing:

(=>=) :: (Product e a -> b) -> (Product e b -> c) -> (Product e a -> c)
f =>= g = \(Prod e a) -> let b = f (Prod e a)
c = g (Prod e b)
in c

The implementation of extract simply ignores the environment:

extract (Prod e a) = a

Not surprisingly, the product comonad can be used to perform exactly the same computations as the reader monad. In a way, the comonadic implementation of the environment is more natural — it follows the spirit of “computation in context.” On the other hand, monads come with the convenient syntactic sugar of the do notation.

The connection between the reader monad and the product comonad goes deeper, having to do with the fact that the reader functor is the right adjoint of the product functor. In general, though, comonads cover different notions of computation than monads. We’ll see more examples later.

It’s easy to generalize the Product comonad to arbitrary product types including tuples and records.

## Dissecting the Composition

Continuing the process of dualization, we could go ahead and dualize monadic bind and join. Alternatively, we can repeat the process we used with monads, where we studied the anatomy of the fish operator. This approach seems more enlightening.

The starting point is the realization that the composition operator must produce a co-Kleisli arrow that takes w a and produces a c. The only way to produce a c is to apply the second function to an argument of the type w b:

(=>=) :: (w a -> b) -> (w b -> c) -> (w a -> c)
f =>= g = g ... 

But how can we produce a value of type w b that could be fed to g? We have at our disposal the argument of type w a and the function f :: w a -> b. The solution is to define the dual of bind, which is called extend:

extend :: (w a -> b) -> w a -> w b

Using extend we can implement composition:

f =>= g = g . extend f

Can we next dissect extend? You might be tempted to say, why not just apply the function w a -> b to the argument w a, but then you quickly realize that you’d have no way of converting the resulting b to w b. Remember, the comonad provides no means of lifting values. At this point, in the analogous construction for monads, we used fmap. The only way we could use fmap here would be if we had something of the type w (w a) at our disposal. If we coud only turn w a into w (w a). And, conveniently, that would be exactly the dual of join. We call it duplicate:

duplicate :: w a -> w (w a)

So, just like with the definitions of the monad, we have three equivalent definitions of the comonad: using co-Kleisli arrows, extend, or duplicate. Here’s the Haskell definition taken directly from Control.Comonad library:

class Functor w => Comonad w where
extract :: w a -> a
duplicate :: w a -> w (w a)
duplicate = extend id
extend :: (w a -> b) -> w a -> w b
extend f = fmap f . duplicate

Provided are the default implementations of extend in terms of duplicate and vice versa, so you only need to override one of them.

The intuition behind these functions is based on the idea that, in general, a comonad can be thought of as a container filled with values of type a (the product comonad was a special case of just one value). There is a notion of the “current” value, one that’s easily accessible through extract. A co-Kleisli arrow performs some computation that is focused on the current value, but it has access to all the surrounding values. Think of the Conway’s game of life. Each cell contains a value (usually just True or False). A comonad corresponding to the game of life would be a grid of cells focused on the “current” cell.

So what does duplicate do? It takes a comonadic container w a and produces a container of containers w (w a). The idea is that each of these containers is focused on a different a inside w a. In the game of life, you would get a grid of grids, each cell of the outer grid containing an inner grid that’s focused on a different cell.

Now look at extend. It takes a co-Kleisli arrow and a comonadic container w a filled with as. It applies the computation to all of these as, replacing them with bs. The result is a comonadic container filled with bs. extend does it by shifting the focus from one a to another and applying the co-Kleisli arrow to each of them in turn. In the game of life, the co-Kleisli arrow would calculate the new state of the current cell. To do that, it would look at its context — presumably its nearest neighbors. The default implementation of extend illustrates this process. First we call duplicate to produce all possible foci and then we apply f to each of them.

This process of shifting the focus from one element of the container to another is best illustrated with the example of an infinite stream. Such a stream is just like a list, except that it doesn’t have the empty constructor:

data Stream a = Cons a (Stream a)

It’s trivially a Functor:

instance Functor Stream where
fmap f (Cons a as) = Cons (f a) (fmap f as)

The focus of a stream is its first element, so here’s the implementation of extract:

extract (Cons a _) = a

duplicate produces a stream of streams, each focused on a different element.

duplicate (Cons a as) = Cons (Cons a as) (duplicate as)

The first element is the original stream, the second element is the tail of the original stream, the third element is its tail, and so on, ad infinitum.

Here’s the complete instance:

instance Comonad Stream where
extract (Cons a _) = a
duplicate (Cons a as) = Cons (Cons a as) (duplicate as)

This is a very functional way of looking at streams. In an imperative language, we would probably start with a method advance that shifts the stream by one position. Here, duplicate produces all shifted streams in one fell swoop. Haskell’s laziness makes this possible and even desirable. Of course, to make a Stream practical, we would also implement the analog of advance:

tail :: Stream a -> Stream a
tail (Cons a as) = as

but it’s never part of the comonadic interface.

If you had any experience with digital signal processing, you’ll see immediately that a co-Kleisli arrow for a stream is just a digital filter, and extend produces a filtered stream.

As a simple example, let’s implement the moving average filter. Here’s a function that sums n elements of a stream:

sumS :: Num a => Int -> Stream a -> a
sumS n (Cons a as) = if n <= 0 then 0 else a + sumS (n - 1) as

Here’s the function that calculates the average of the first n elements of the stream:

average :: Fractional a => Int -> Stream a -> a
average n stm = (sumS n stm) / (fromIntegral n)

Partially applied average n is a co-Kleisli arrow, so we can extend it over the whole stream:

movingAvg :: Fractional a => Int -> Stream a -> Stream a
movingAvg n = extend (average n)

The result is the stream of running averages.

A stream is an example of a unidirectional, one-dimensional comonad. It can be easily made bidirectional or extended to two or more dimensions.

Defining a comonad in category theory is a straightforward exercise in duality. As with the monad, we start with an endofunctor T. The two natural transformations, η and μ, that define the monad are simply reversed for the comonad:

ε :: T -> I
δ :: T -> T2

The components of these transformations correspond to extract and duplicate. Comonad laws are the mirror image of monad laws. No big surprise here.

Then there is the derivation of the monad from an adjunction. Duality reverses an adjunction: the left adjoint becomes the right adjoint and vice versa. And, since the composition R ∘ L defines a monad, L ∘ R must define a comonad. The counit of the adjunction:

ε :: L ∘ R -> I

is indeed the same ε that we see in the definition of the comonad — or, in components, as Haskell’s extract. We can also use the unit of the adjunction:

η :: I -> R ∘ L

to insert an R ∘ L in the middle of L ∘ R and produce L ∘ R ∘ L ∘ R. Making T2 from T defines the δ, and that completes the definition of the comonad.

We’ve also seen that the monad is a monoid. The dual of this statement would require the use of a comonoid, so what’s a comonoid? The original definition of a monoid as a single-object category doesn’t dualize to anything interesting. When you reverse the direction of all endomorphisms, you get another monoid. Recall, however, that in our approach to a monad, we used a more general definition of a monoid as an object in a monoidal category. The construction was based on two morphisms:

μ :: m ⊗ m -> m
η :: i -> m

The reversal of these morphisms produces a comonoid in a monoidal category:

δ :: m -> m ⊗ m
ε :: m -> i

One can write a definition of a comonoid in Haskell:

class Comonoid m where
split   :: m -> (m, m)
destroy :: m -> ()

but it is rather trivial. Obviously destroy ignores its argument.

destroy _ = ()

split is just a pair of functions:

split x = (f x, g x)

Now consider comonoid laws that are dual to the monoid unit laws.

lambda . bimap destroy id . split = id
rho . bimap id destroy . split = id

Here, lambda and rho are the left and right unitors, respectively (see the definition of monoidal categories). Plugging in the definitions, we get:

lambda (bimap destroy id (split x))
= lambda (bimap destroy id (f x, g x))
= lambda ((), g x)
= g x

which proves that g = id. Similarly, the second law expands to f = id. In conclusion:

split x = (x, x)

which shows that in Haskell (and, in general, in the category Set) every object is a trivial comonoid.

Fortunately there are other more interesting monoidal categories in which to define comonoids. One of them is the category of endofunctors. And it turns out that, just like the monad is a monoid in the category of endofunctors,

The comonad is a comonoid in the category of endofunctors.

Another important example of a comonad is the dual of the state monad. It’s called the costate comonad or, alternatively, the store comonad.

We’ve seen before that the state monad is generated by the adjunction that defines the exponentials:

L z = z × s
R a = s ⇒ a

We’ll use the same adjunction to define the costate comonad. A comonad is defined by the composition L ∘ R:

L (R a) = (s ⇒ a) × s

Translating this to Haskell, we start with the adjunction between the Product functor on the left and the Reader functor or the right. Composing Product after Reader is equivalent to the following definition:

data Store s a = Store (s -> a) s

The counit of the adjunction taken at the object a is the morphism:

εa :: ((s ⇒ a) × s) -> a

counit (Prod (Reader f) s) = f s

This becomes our extract:

extract (Store f s) = f s

The unit of the adjunction is:

unit :: a -> Reader s (Product a s)
unit a = Reader (\s -> Prod a s)

We construct δ, or duplicate, as the horizontal composition:

δ :: L ∘ R -> L ∘ R ∘ L ∘ R
δ = L ∘ η ∘ R

We have to sneak η through the leftmost L, which is the Product functor. It means acting with η, or Store f, on the left component of the pair (that’s what fmap for Product would do). We get:

duplicate (Store f s) = Store (Store f) s

(Remember that, in the formula for δ, L and R stand for identity natural transformations whose components are identity morphisms.)

Here’s the complete definition of the Store comonad:

instance Comonad (Store s) where
extract (Store f s) = f s
duplicate (Store f s) = Store (Store f) s

You may think of the Reader part of Store as a generalized container of as that are keyed using elements of the type s. For instance, if s is Int, Reader Int a is an infinite bidirectional stream of as. Store pairs this container with a value of the key type. For instance, Reader Int a is paired with an Int. In this case, extract uses this integer to index into the infinite stream. You may think of the second component of Store as the current position.

Continuing with this example, duplicate creates a new infinite stream indexed by an Int. This stream contains streams as its elements. In particular, at the current position, it contains the original stream. But if you use some other Int (positive or negative) as the key, you’d obtain a shifted stream positioned at that new index.

In general, you can convince yourself that when extract acts on the duplicated Store it produces the original Store (in fact, the identity law for the comonad states that extract . duplicate = id).

The Store comonad plays an important role as the theoretical basis for the Lens library. Conceptually, the Store s a comonad encapsulates the idea of “focusing” (like a lens) on a particular substructure of the date type a using the type s as an index. In particular, a function of the type:

a -> Store s a

is equivalent to a pair of functions:

set :: a -> s -> a
get :: a -> s

If a is a product type, set could be implemented as setting the field of type s inside of a while returning the modified version of a. Similarly, get could be implemented to read the value of the s field from a. We’ll explore these ideas more in the next section.

## Challenges

1. Implement the Conway’s Game of Life using the Store comonad. Hint: What type do you pick for s?

## Acknowledgments

I’m grateful to Edward Kmett for reading the draft of this post and pointing out flaws in my reasoning.

Next: F-Algebras.

Next Page »