For every , left multiplication gives a bijection
whose inverse is . Hence every left coset has the same cardinality as . If is an infinite set, every left coset is infinite; if is a finite set, every left coset is finite. These are the only possibilities.
Suppose were finite, say . There are at most left cosets, each with elements by part (i), and they partition . Hence would be a finite set with at most elements. This contradicts the assumption that is an infinite group. Therefore .
Every left coset with is also a left coset of in . Moreover equality of two such cosets means the same thing in and in :
Thus the map from the set of left cosets of in to the set of left cosets of in , taking to itself, is injective. Consequently the index of a subgroup satisfies
These are sets of cosets; no normal subgroup assumption is needed.
Define a map between sets of left cosets by
It is well defined: replacing by for changes neither component. It is injective because
The Cartesian product on the right has at most elements. Therefore
This proves the basic bound for a finite intersection of finite-index subgroups.
Put . Repeated application of part (iv), or mathematical induction on , gives
Part (ii) therefore implies that is an infinite set. If is nonempty, choose . We claim its left coset is wholly outside . If for some , then and closure under products and inverses would give , a contradiction.
Thus . Part (i) makes infinite, so the complement is infinite. If there is no such , the complement is empty. This proves the complement of a finite union of finite-index subgroups property:

Articles by others on the same topic (0)

There are currently no matching articles.