Wikipedia 10K Redux by Reagle from Starling archive. Bugs abound!!!

<-- Previous | Newer --> | Current: 980728898 JoshuaGrosse at Mon, 29 Jan 2001 00:41:38 +0000.


Sets are one of the base concepts of mathematics.  A set is, more or less, just a collection of objects, called its elements.  Standard notation uses braces around the list of elements, as in:

   {red, green, blue}
   {x : x is a primary color}

If every x in some set A is contained in some set B, then A is said to be a SubSet of B.  Every set has at least two subsets: itself, called the improper subset, and the empty set {}.  The union of some sets is the set of all elements contained in at least one, and the intersection is the set of all elements contained in them all.  These are denoted

   A1 u A2 u A3 ...
   A1 n A2 n A3 ...

respectively.  If you don't mind jumping ahead a bit, the subsets of a given set form a BooleanAlgebra under these operations.  The set of all subsets of X is called its PowerSet and is denoted 2^X.

'''Axioms for set theory'''

The most common axioms for set theory are those of Zermelo and Fraenkel, which do a good job handling most sets, and are equivalent to several other axiomatizations.  However, they aren't the only possible axiom schema, and in some senses are to restrictive.  For instance, they provide a way to go from any set X to a set {X} and thereby prevent the existence of a set of all sets, which some other models allow.

The axioms of ZF are as follows:
   Two sets are the same iff they have the same elements.
   No set can be recursively defined.  This is to prevent RusselsParadox.
   There is an empty set {}.
   If X,Y are sets, so is {X,Y}.
   For any set X, there is another set composed of the union of its elements.
   There exists an infinite set.
   Axiom of replacement - roughly that sets can be imaged under functions.

The AxiomOfChoice leads to the existence of several sets, like a well-ordering for the real numbers, that can't be constructed in ordinary ZF set theory.  However, it is impossible to prove they don't exist - the axiom is independent of the axioms of ZF - and so we are free to add it to our axioms if we want.  This has become more or less conventional, since most of the sets it adds are handy to have, even if we can never find an explicit representation for any of them.