std::bitset is one. But you can just use an unsigned int, or array of them. In C++ or C, set intersection is &, union is |, complement is ~. Cardinality is __builtin_popcount, or std::bitset<>::count(). Membership test for m is (s & (1<<m)). For a larger set, represented as an array of unsigned, it's (s[m>>5] & (1<<(m&0x1f)). std::bitset encapsulates all this, optimally. There are similarly efficient implementations for lots of other useful operations: consult HAKMEM.

Bitsets are useful in more places than you might guess, because they can be used to filter out, with typically a single instruction, a majority of uninteresting cases in searches, leaving fewer cases to be examined with more precise and expensive operations. You record interesting properties of elements of a collection upon insertion as bits in a word stored alongside (or, better, in an index); and then first check the bits during any search. It is easy to speed up an amazing variety of programs by an order of magnitude, sometimes much more, with a tiny change.

For example, you can store an int value for each of a (possibly very large) collection of lines of text, representing the set of less-common letters that appear in the line. Searching for lines that contain a string, you first check it against the set for the string. Any line that doesn't have them all certainly doesn't have the string.

Leibniz (inventor of calculus) used this method very heavily in his own work. Before computers--and even into the 1960s--it was the most important way of automating data processing. Back then, you used cards that had a row of either holes or notches along one edge, and selected matching cards from a stack by inserting a rod through the stack at a selected spot, and lifting.