opener

Image source: Marco Orazi / Flickr.

There is more than one ‘infinity’—in fact, there are infinitely-many infinities, each one larger than before!

Beyond infinity

Expert reviewers

Dr Mary Coupland

Senior Lecturer, School of Mathematical and Physical Sciences

University of Technology Sydney

Dr Melissa Tacy

Research Fellow

Australian National University

Essentials

  • Sets that have the same size as the set of natural numbers are called ‘countably infinite’. Examples include the set of even numbers and the set of rational numbers (numbers that can be written as fractions).
  • The set of real numbers (numbers that live on the number line) is the first example of a set that is larger than the set of natural numbers—it is ‘uncountably infinite’.
  • There is more than one ‘infinity’—in fact, there are infinitely-many infinities, each one larger than before!

Previously, we looked at methods of comparing the sizes of sets and discovered that many infinite sets could be paired up exactly, such as the set of natural numbers GLOSSARY natural numbersThe set of all positive whole numbers, which can be used to count individual objects. Depending on which mathematician you ask, this set may or may not include the number 0. Examples: 1, 2, 3, … and the set of even numbers. These sets both have the same size, or ‘cardinality’—namely, ℵ0, our familiar ‘infinity’.

Does every infinite set have the same size? Can they all be matched up exactly?

A pattern of light that seems to repeat to infinity
Despite intuitions, we can count and compare the size of infinite sets. Image source: benjgibbs / Flickr.

For a long time, it seemed as though all infinite sets would be able to be paired up with each other—meaning that every infinite set had the same size, ℵ0. This intuitive idea was shattered in 1874 by the discovery of larger and larger infinities. In a five-page paper that shook the mathematical world (and gave rise to set theory in the process), the German mathematician Georg Cantor laid out his discovery of new horizons ‘beyond infinity’.

How is it possible to construct a ‘larger’ infinite set than the set of all natural numbers? Clearly, we’re going to need to add extra ingredients to the mix.

Rational thinking

The first way we can expand our number system is by introducing all the negative numbers, forming the set of integers GLOSSARY integersThe set of all whole numbers, both positive and negative. The number 0 is definitely a part of this set. The natural numbers are a part of the integers. Examples: -2,-1,0,1,2, … , ℤ. However, it soon becomes clear that this won’t help us reach a larger cardinality—we’ve effectively doubled the set ℕ by adding in a negative number for each positive number, and we know from our experience with combining the odd and even numbers that doubling an infinite set is not going to change its cardinality.

The next thing we can try is to include all the numbers that can be written as fractions, creating the set of rational numbers GLOSSARY rational numbersThe set of all numbers, both positive and negative, that can be written as a fraction. Equivalently, the set of all numbers whose decimal expansions either terminate or eventually repeat. The natural numbers and the integers are a part of the rational numbers. Examples: -3, 1/3, 0.5, … , ℚ. We seem to have an awful lot of numbers in ℚ—for starters, there’s infinitely-many fractions between any two whole numbers! Clearly, it’s going to be difficult to pair off the set of natural numbers:

$$\mathbb{N}\text{ = \{0, 1, 2, 3, 4, 5, 6, … \}}$$

with the set of rational numbers:

$$\mathbb{Q}\text{ = \{ 0,} \frac{1}{2},\frac{1}{3}, \frac{2}{3}, \frac{1}{4}, \frac{3}{4}, … , 1, \frac{11}{2}, \frac{11}{3} … \}$$

Difficult … but not impossible. Cantor showed how this could be done by thinking about things a little differently. The key is to realise that in order to pair off the rational numbers with the natural numbers, we just need to find a way to list the rational numbers in some sort of order. We can then pair the first rational number in our list with the natural number 1, the second rational in the list with 2, and so on.

On the surface, this seems a tall order—how can we can list the rational numbers, which include fractions, in an order that captures every single one? There’s no such thing as the smallest fraction bigger than 0, for starters! But by doing a few mental zigzags, it turns out that it is possible to make such a list.

To start, we notice that every fraction is made from a pair of natural numbers—one on the top of the fraction (the numerator) and one on the bottom (the denominator). There are two caveats. First, we cannot have 0 on the denominator of the fraction. Second, we can write the same fraction in infinitely-many different ways using different pairs of numbers (for example, 12 is the same as 24, and 36, and so on).

  • Why you can’t divide by zero

    Aside from calculators throwing up errors, and jokes about workbooks catching fire and black holes spontaneously forming, there’s a very valid (and very simple) reason why we cannot divide any number by zero.

    We learn early in school that there are four basic operations in mathematics: addition, subtraction, multiplication and division. To mathematicians, however, there are only two basic operations: addition and multiplication. We get subtraction and division for free, as the opposites of addition and multiplication.

    More specifically, every subtraction question can be thought of as an addition question, and every division question can be thought of as a multiplication question. For instance, a typical multiplication question might look like:

    $$\text{2 × 3 = ?}$$

    Whereas a typical division problem might look like:

    $$\text{6 ÷ 2 = ?}$$

    But what does this mean? When we ask, ‘What is six divided by two?’, we are really asking, ‘What do we multiply two by, in order to get six?’. In other words,

    $$\text{6 ÷ 2 = ?}$$

    is the same problem as:

    $$\text{2 × ? = 6}$$

    So a division problem is really a multiplication problem with the unknown number on the other side of the equation. With that in mind, let’s think about dividing by zero.

    $$\text{1 ÷ 0 = ?}$$

    is the same problem as asking, ‘What do we multiply zero by, to get one?’.

    $$\text{0 × ? = 1}$$

    But there is no number that works! No matter what we multiply zero by, we’re always going to get zero. That’s why we can’t divide one, or two or any other number by zero—there’s no sensible answer, because there’s no number that we can multiply by zero in order to get one, or two, or anything else—aside from zero.

    Which leads to one final point: what about dividing zero by zero?

    $$\text{0 ÷ 0 = ?}$$

    is the same problem as asking, ‘What do we multiply zero by, in order to get zero?’.

    $$\text{0 × ? = 0}$$

    This time the problem is not that there’s no number that works—it’s that every number works! Since multiplying any number by zero gives you zero, it follows that zero divided by zero could be literally any number at all. And, for that reason, it also doesn’t have a sensible answer.

    If we tried to be accommodating and let 0 ÷ 0 be any number you choose, then we’d quickly run into the following problem:

    Since 0 ÷ 0 = 1 and 0 ÷ 0 = 2, that means 1 = 0 ÷ 0 = 2.

    As 1 and 2 are most definitely not the same number, we can’t sensibly define 0 ÷ 0.

Since every rational number can be thought of as a pair of natural numbers, we can draw up the following infinite table to capture every possible rational number:

A diagram showing a table of fractions. Top headings: numerators from 0 to 6. Left headings: denominators from 1 to 6. You can then fill out the inside of the graph by dividing the numerator by the denominator, providing a list of all fractions. You can continue beyond 6 to infinity.
Above: The start of an infinite table that lists every possible fraction in its most simplified form. Each fraction is obtained by dividing the column number (the numerator) by the row number (the denominator).

We only want each fraction to appear once in the table, and at the moment this table captures each fraction over and over again, infinitely-many times. So we take care to eliminate all the duplicates by crossing out each entry in the table where the fraction can be simplified (all the entries where the numerator and denominator share a common factor are removed).

The same diagram as above, but now the duplicate entries are deleted. For example, 0 is 0/1, 0/2, 0/3 and so on, and 1/2 is 1/2, 2/4, 3/6 and so on.
Above: The infinite table of fractions, with duplicate entries deleted.

Having done that, we can now zigzag our way through the table to catch every single fraction, like so:

The same diagram as above, but now a zig-zag line is drawn from the top left of the table to the immediate right, then down one, then up and right, then down and left, then up and right, and so on. This is done to put all the fractions into some order.
Above: Zig-zagging through the table of fractions, in order to list every possible fraction in some kind of order (the order in which we meet them in our zig-zags).

Then all we have to do then is list every fraction in the order we encountered it during the zigzag:

Table: one possible bijection between the natural numbers and the rational numbers.
Natural number Rational number
1 0
2 1
3 12
4 2
5 13

And, just like that, we’ve made our bijection between the rational numbers ℚ and the natural numbers ℕ! We can check that every natural number has a partner (the list of fractions never ends) and that every rational number has a partner (every rational appears exactly once in the table, after we carefully remove the duplicates). Although we didn’t capture all the negative rational numbers in our mapping, the same idea can be easily modified to take them into account.

So it turns out that the set of rational numbers is countable too—it has cardinality ℵ0. It may seem like we’ve run out of options; that there really is only one type of infinity. But Cantor had one final trick up his sleeve, which he published in his famous 1874 paper. There is one more source of numbers we can draw on.

The real surprise

Apart from natural numbers and rational numbers, there is one more set of numbers that most people are familiar with—the set of real numbers GLOSSARY real numbersThe set of all numbers that live on the number line. The natural numbers, integers and rational numbers are a part of the real numbers—but the real numbers also include numbers that cannot be written as a fraction, like π and the square root of pi, whose decimal expansions never terminate or repeat. Examples: -2, 0, 1/4, √3, log 2, … , ℝ. The set of real numbers includes all the whole numbers and all the fractions, as well as lots of other numbers that can’t be written as a fraction, like π (pi) and the square root of 2.

Cantor was able to prove that it was impossible to create a bijection between the set of real numbers ℝ and the set of natural numbers ℕ. No matter how hard you try, you’ll always end up using up all of the natural numbers and still having real numbers left over, unpaired. There are definitely more real numbers, in a mathematically precise way, than there are natural numbers.

To prove this, Cantor used an elegant mathematical technique known as ‘proof by contradiction’. Cantor began by assuming that you could create the bijection, listing all real numbers in some order, just like we did for the rational numbers, in order to match them with the natural numbers. He then showed that no matter how you had created the list, he’d be able to show you a real number that didn’t appear on your list, thus contradicting your result and proving it couldn’t be done.

Even more incredible, Cantor’s ‘diagonalisation argument’ doesn’t even need all the real numbers to be considered. It turns out that any piece of the number line, no matter how tiny, contains ‘more’ numbers in it than all of the natural numbers (and all of the rationals too)! To illustrate Cantor’s argument, we’ll just take the set of real numbers between 0 and 1 and show that you can’t possibly match them up with the set of all natural numbers.

Suppose that we’ve created our bijection, producing a list containing every single real number between 0 and 1 in some order. Cantor then hands us another real number between 0 and 1 that can’t possibly be on our list. To create this number, he takes the first decimal place of our first number on our list and adds 1 to it, and uses the result as the first decimal place of his new number. (If the first decimal place of our first number was 9, he changes it to a 0). He then takes the second decimal place of our second number and adds 1 to it, using the result as the second decimal place of his new number. This process continues, resulting in a real number that Cantor claims won’t be on our list.

Interactive

Some infinities are greater than others

No matter how you pair them together, you can always find a real number that isn’t paired with a natural number. This means there are more real than natural numbers—even though there’s an infinite amount of both!

Use this interactive to pair up some natural numbers with the beginning of real numbers. We’ll always be able to generate a new number that’s not on your list. We do this by taking one digit from each real number (in a diagonal pattern) and incrementing it by one.

Natural numberReal number
10.01234
20.12345
30.23456
40.34567
50.45678
Unmapped real number
0.13579
Natural numberReal number
10.10000
20.01000
30.00100
40.00010
50.00001
Unmapped real number
0.22222
Natural numberReal number
10.22222
20.02000
30.00200
40.00020
50.00002
Unmapped real number
0.33333
Natural numberReal number
10.13254
20.24365
30.35476
40.46586
50.57698
Unmapped real number
0.25599
Natural numberReal number
10.
20.
30.
40.
50.
Unmapped real number
0.00000
Insert digits in each section then press ‘update result’
Reset

Is he right? Well, let’s try and find Cantor’s number in our list. It can’t be the first number, because the first decimal place (at least) doesn’t match up. It can’t be the second number on our list, because the second decimal place doesn’t match. It can’t be the millionth number on our list, because the millionth decimal place doesn’t match, and so on. No matter how we go about making the list, we can’t even capture every real number between 0 and 1!

Cantor’s argument works for the real numbers because the decimal expansion of a real number doesn’t have to have a pattern to it. This is in contrast to the decimal expansions of rational numbers, which must eventually repeat themselves.

So at last we have finally found a larger infinity than ℵ0! Perhaps not surprisingly, this new infinity—the cardinality of the set of real numbers ℝ—is called ℵ1. It’s the second transfinite cardinal number, and our first example of a bigger infinity than the ℵ0 infinity we know and love.

 
Video: How to count past infinity (Vsauce / YouTube). View video details.

Conclusion

So although Buzz Lightyear’s goals of travelling ‘to infinity and beyond’ doesn’t make sense in the literal sense—infinity is not a place, nor a destination, nor a final point—you could argue that mathematics allows us to explore infinity and beyond. Different infinite sets can have different cardinalities, and some are larger than others. Beyond the infinity known as ℵ0 (the cardinality of the natural numbers) there is ℵ1 (which is larger) … ℵ2 (which is larger still) … and, in fact, an infinite variety of different infinities.

It may seem esoteric, but the understanding of infinity—and set theory—is vital to understanding the very foundations of mathematics. It’s knowledge that was hard won over millennia, and that knowledge provides us with tools for understanding very real complexities in the world around us.