[Math] Why is the distributive property so pervasive in mathematics

abstract-algebramath-historyphilosophysoft-question

I just read this post which gives a geometric argument for the distributive law for real numbers, which I liked: https://math.stackexchange.com/a/466397/241685

However the distributive law comes up everywhere, not just for numbers. Set intersection distributes over union, inner products distribute over vector addition, wedge products distribute over vector addition, ring multiplication distributes over ring addition, matrix multiplication distributes over matrix addition, etc.

Is it that we are intentionally studying systems which generalize the distributive law for numbers, or is it that the systems we study which are interesting happen to generalize the distributive law? And either way, why is this the case?

Best Answer

Let's start very fundamentally. A lot of mathematics is concerned about sets of objects and sets of functions mapping one or more of those objects to another one.

Now let's look at the case that we have a set $X$ and a function $f$ takes two elements of that set and gives a third one, that is, $f:X\times X\to X$. Note that this is just an arbitrary function acting on two elements of an arbitrary set. Now if we need such a function often, we like to write $f(x,y)$ in a slightly simpler form. This usually is something like $x*y$, $x\cdot y$ or simply $xy$, but especially if $f(x,y)=f(y,x)$ (and in some rare cases, even if not), it is not uncommon to write $f(x,y)=x+y$. Well, we usually want a few other conditions, but at the moment we don't even need those. So at this point, all we care about is that $x+y$ takes two elements from some set $X$, and gives another element of the same set $X$.

Now let's look at two sets $X$ and $Y$, both coming with their own $+_X$ and $+_Y$ (this includes the case that $X=Y$ and $+_X=+_Y$). Now as with any two sets, we can consider functions from $X$ to $Y$. However, there are certain functions that are special: Namely those functions $\phi:X\to Y$ which respect our additions. That is, $\phi(x+y)=\phi(x)+\phi(x)$. Now if we do this a lot, we may like to omit the parentheses wherever possible, that is, write the function application as product of the function and the argument. The most prominent example of this is operators in linear algebra. If we do that, the above law reads: $$\phi(x+y) = \phi x + \phi y$$ Voila, a distributive law!

But wait, there is more: Given two functions $\phi:X\to Y$ and $\chi:X\to Y$, it is natural to ask about a function $x\mapsto \phi(x)+\chi(x)$. Now this gives a function operating on functions from $X\to Y$, that takes two functions and returns a new function. That new function is exactly the function that applies $\phi$ and $\chi$ to its argument and then adds that result. It is natural to consider that function on functions also as addition (the technical term is "pointwise addition"), and again denote it with $+$. So, using the product notation for function application, we have, by definition, $$(\phi + \chi)x = \phi x + \chi x$$ Voila, another distributive law!

OK, so now where do the numbers come in? Well, let's now consider one of the additional requirements which I didn't talk about above: Namely we also require of an operation we want to call $+$ that it is associative, that is, $(a+b)+c = a+(b+c)$. This means that when you add up many things, you can basically omit the parentheses.

This in particular means that if you repeatedly add something to itself, like $b+b+b+\dots+b$, all that matters is how many $b$s you've got in that sum. Therefore we again introduce a new multiplication, this time with a positive integer $n$: $$nb = \underbrace{b+b+\dots+b}_{n\text{terms}}$$ Note that this could also be seen as interpreting the integer $n$ as a function that takes an argument $b$ and returns a sum of $n$ $b$s. That is, we have functions \begin{align} 1&:x\mapsto x\\ 2&:x\mapsto x+x\\ 3&:x\mapsto x+x+x\\ &\vdots \end{align} Now we can ask: Are these "number functions" functions that respect the addition structure? Well, let's try e.g. with $3$: \begin{align} 3(x+y) &= (x+y)+(x+y)+(x+y) && \text{Definition of multiplication with $3$}\\ &= x+y+x+y+x+y && \text{because we require associativity}\\ &= x+x+x+y+y+y && \text{because we earlier required *commutativity* ($x+y=y+x$)}\\ &= 3x + 3y && \text{again, definition of multiplication} \end{align} The same works of course with any $n$ (a strict mathematical proof is slightly more involved). So we have: $$n(x+y) = nx + ny$$ Again, a distributive law.

Now by interpreting numbers as functions as above, we get an addition, namely pointwise addition. And of course we have the normal number addition. But it is not hard to check that those two additions indeed give identical results, that is, they can be regarded as the same addition.

But for pointwise addition, we already know that there is a distributive law, which therefore carries over also to multiplications with numbers: $$(m+n)x = mx + nx$$

OK, now let's consider the case where $X$ is actually a set of numbers itself, and $+$ is the normal addition of numbers. Then quite obviously, you recover the usual multiplication, and the corresponding distributive law.

So distribution laws occur quite naturally, in many contexts by requiring only that a binary operation exists without making any further requirements, and asking about operations that respect that operation, and for operations involving numbers, with only the additional requirement that addition is associative and commutative (two things we commonly demand from operations we call "addition").