Appendix: Infinity

You probably have never given much thought to characterising infinity other than the recognition that there is no end to it. Infinite quantities can be discomfitting, and most applications of mathematics shy away from dealing with them. You may be surprised to learn that there are several sizes of infinity. (In fact, there is an infinite number of them.) In order to see this, we must examine the way in which we compute sizes of finite sets and see how it generalises to the infinite case.

What is the size of the set maddabaybie, snoo, updok? How did you get that answer? You counted. The process of counting is an act of putting the elements of a set into one-to-one correspondence with a sequence of natural numbers 1, 2, 3, .... The last number in the sequence is the number of elements in the set. The natural numbers, therefore, are nothing but abstract properties of sets of objects. A couple of lovers and a pair of socks are both instances of the number two, because they both can be mapped in a one-to-one manner to the set 1, 2. Any finite set can have its elements marked off like this, placed in a one-to-one correspondence with a sequence of natural numbers. If this can be done for any finite set, it is natural to ask whether the definition of number can be extended so that there are numbers that describe infinite sets, for there do exist one-to-one correspondences that map between infinite sets.

Often it is easier to show that two sets are the same size than to specify the particular size that they have in common. Russell[50] uses the example of the relationship between husbands and wives. The relationship, marriage, constitutes a one-to-one mapping between the set of husbands and the set of wives. (We assume for the sake of this example that there is no polygamy.) Since there is exactly one husband for each wife, and exactly one wife for each husband, it can be inferred that the set of wives and the set of husbands have the same size. Thus if we know the size of the set of wives, then we also know the size of the set of husbands, and vice versa. In this example, the size is finite. But it need not be so.

The simplest infinity is countable infinity or enumerable infinity. The members of a countably infinite set can be counted. That is, it is possible to place them in one-to-one correspondence with the natural numbers--not a finite sequence of natural numbers, not just a part of the set of natural numbers, but all the natural numbers. All the sets that can be placed into one-to-one correspondence with the natural numbers share the property of countability. The natural numbers 0, 1, 2, 3, ... are of course countable, because any set can be placed in one-to-one correspondence with itself. So are the integers 0, 1, -1, 2, -2, 3, -3, .... Another term for the size of a set is cardinality. The cardinality of a finite set of size n is just the number n. The cardinality of a countably infinite set is 0. (`' is aleph, the first letter of the Hebrew alphabet. `0' is read "aleph-null".) 0 is a cardinal number. You have probably been taught (or perhaps you will yet be taught) in your mathematics classes that it is a mistake to treat infinity as a number. This is true in the context of arithmetic on the real or complex numbers. But numeration can be extended to infinite quantities; it is only a matter of making consistent definitions. Infinite cardinal numbers do not obey the same algebraic rules as finite numbers. For example, there is no distributive property; 0+0 = 0. (Think of the cardinality of the natural numbers in comparison with that of the integers.) The only distinction that will be made between types of infinity here is the distinction between countable and uncountable infinity.

The rational numbers are also countable, but this proof takes more work. The rational numbers are the numbers that can be expressed as quotients of integers. They can be generated in an infinite matrix whose rows and columns are labelled with the generating integers. The column number is the numerator, and the row number is the denominator.

0 1 -1 2 -2 3 -3 ...

1 0/1 1/1 -1/1 2/1 -2/1 3/1 -3/1 ...

-1 0/-1 1/-1 -1/-1 2/-1 -2/-1 3/-1 -3/-1 ...

2 0/2 1/2 -1/2 2/2 -2/2 3/2 -3/2 ...

-2 0/-2 1/-2 -1/-2 2/-2 -2/-2 3/-2 -3/-2 ...

3 0/3 1/3 -1/3 2/3 -2/3 3/3 -3/3 ...

-3 0/-3 1/-3 -1/-3 2/-3 -2/-3 3/-3 -3/-3 ...

: : : : : : : :

Obviously, there are many duplicate values constructed by this table, but these can be excluded in the enumeration. At any particular point in the enumeration, the list of rationals built up so far is finite, so it can be searched in finite time. The value of the ratio currently under consideration can be compared with all the values in the list, and added to the list if and only if it is not already present. All that remains is to specify an order in which to consider all the ratios in the table. Here is one. Therefore the rationals are countably infinite. We can think of taking this squiggly line (ray, actually) that touches all the rationals, removing duplicates from it, and stretching it out so that it looks just like the natural number line.

The matrix used above to generate the rationals can be extended to infinity in all directions and viewed as a 2-space of integers. An enumeration scheme can be constructed to spiral outward from the origin in such a 2-space. Such an enumeration will eventually reach any coördinate in the space. Therefore, the set of the integer pairs in 2-space is of cardinality 0. An enumeration scheme can be constructed for any number of dimensions. Therefore, the cardinality of the integer tuples in n-space, for all n, is 0.

So far the only property that we have proven about an infinite set is countability. Uncountability can also be proven. One common way of proving a set uncountable is Cantor diagonalisation, devised by Georg Cantor. Cantor diagonalisation is an indirect proof method; it supposes that the set in question is countable, and therefore that all the elements of the set can be included in an enumeration, and uses that assumption to construct a way of specifying for every possible enumeration an element that is in the set but is not included in the enumeration. The real numbers in the interval (0, 1) are uncountably infinite. To see that this is so, suppose that the reals in (0, 1) are countable. Then there must be some enumeration of all of them. Here is one possible enumeration:

0.042546...

0.583901...

0.839702...

0.043850...

0.374508...

0.572947...

:

This list can be viewed as a matrix, with the row index representing the position in the enumeration and the column index representing the decimal place. Observe the decimal generated by the diagonal of this matrix. A number that is not included in the enumeration can be generated by altering every digit on the diagonal. A number constructed in this way differs from the i[th] number in the enumeration in the i[th] decimal place. The example above cannot be a complete enumeration, because there is some number 0.190918... that is not included in it. This construction is valid for any proposed enumeration. Suppose an adversary in this argument says, "Well, if that number 0.190918... isn't in my enumeration, I'll just stick it in." The adversary is still no closer to constructing an enumeration of all the reals in (0, 1), because we can take the new enumeration, which includes 0.190918..., and use the same method as before to construct another number that is not included in it. No matter how many holes the adversary patches, there will always be more. Therefore, the reals are uncountable. The cardinality of the reals is denoted c. Obviously, c is greater than 0.

Two sets are of the same cardinality, whether that cardinality is finite or infinite, if and only if a one-to-one mapping between all their elements can be constructed. Thus the set of all integers is the same size as the set of all natural numbers:

0 1 2 3 4 5 6 ...

0 1 -1 2 -2 3 -3 ...

The one-to-one function here from any natural number n to any integer k is

n odd -> (n+1)/2

[] n even -> -n/2

Since this definition of equivalent size applies to infinite sets, it can be used to show that the cardinality of all the real numbers is the same as the cardinality of the reals in the interval (0, 1). Consider the function y = tan(.(x-1/2)) on the interval (0, 1), which has asymptotes at x=1 and at x=0 and constitutes a one-to-one mapping from the reals in (0, 1) to all the reals. (If you have not yet studied trigonometry, you will learn about the tangent function when you do. How it arises and what it is used for are not germane to this discussion; it is sufficient that it exists and is a mapping from (0, 1) to .)

The reals in the plane are of the same cardinality, c, as the reals on a line. A one-to-one correspondence between these two domains can be defined by interleaving the digits of a pair of coördinates. Any real number can be represented in decimal notation by a string of the form x.y, where x and y are the digit strings representing the integral part and the fractional part, respectively, of the the real number. Let xi denote the i[th ]digit in the string x. Then a one-to-one correspondence can be created by encoding the pair of reals (u.v, x.y) in the single real u0x0u1x1...u|u|-1x|x|-1 . v0y0v1y1...v|v|-1y|y|-1. This scheme interleaves the two integral parts and the two fractional parts separately and thereby preserves information on the location of the decimal point within each coördinate. (Numbers can be padded with 0's on the left and on the right, if necessary, to make both numbers contain the same number of digits.) Take a moment to think for yourself how this encoding scheme can be generalised to coördinates in n-space.

The complex numbers are all the numbers of the form a+bi, where i is the imaginary number -1. Since the complex numbers may be viewed as pairs of reals (a, b), they have the same cardinality as the reals in the plane, namely, c.


[50][Russell 1920]