Measure Theory

Carathéodory criterion

Carathéodory’s criterion is the clever test that decides which sets an outer measure can handle additively. The slogan: a set E is measurable if and only if it splits every other set cleanly, with no slack. That is, using E as a knife to cut any test set must leave the two pieces adding back up to the whole, with respect to the outer measure.

Precisely, given an outer measure mu-star on X, a set E is Carathéodory measurable if for every test set A (every subset of X), mu-star(A) = mu-star(A intersect E) + mu-star(A minus E). Because subadditivity already gives the “at most” direction automatically, the content of the criterion is the reverse inequality: E must not waste any measure when it partitions A.

The remarkable theorem of Carathéodory is that the collection of all such measurable sets is automatically a sigma-algebra, and the outer measure restricted to it is a genuine, complete, countably additive measure. This is the engine that turns the easy-to-build outer measure into a real measure; applied to Lebesgue outer measure it yields exactly the Lebesgue measurable sets and Lebesgue measure.

The criterion may look unmotivated at first — why test against all sets A, not just simple ones? The universal quantifier is exactly what forces the family to be closed under complements and countable unions, which is why the resulting collection is a sigma-algebra for free.

Also called
Carathéodory measurability卡拉西奥多里可测性卡拉西奧多里可測性