ΘρϵηΠατπ

Pointwise Convergence of Functions
Suppose that X is a set, f1,f2,:X is a sequence of functions, and f:X. We say that f1,f2, converges pointwise on X to f if limkfk(x)=f(x) for every xX.
Uniform Convergence of Functions
Suppose that X is a set, f1,f2,:X is a sequence of functions, and f:X. We say that f1,f2, converges uniformly on X to f if for every ϵ+, there exists n+ such that |fk(x)f(x)|<ϵ for every kn and every xX.
Uniform Limit of Continuous Functions is Continuous
Suppose that B, f1,f2,:B converges uniformly on B to f:B, and bB. If each fk is continuous at b, then f is continuous at b.

Let ϵ+. By uniform convergence, choose N such that |fk(x)f(x)|<ϵ/3 for every kN and every xB. Since fN is continuous at b, choose δ>0 such that |fN(x)fN(b)|<ϵ/3 whenever xB and |xb|<δ. The triangle inequality gives |f(x)f(b)|<ϵ, so f is continuous at b.

Egorov's Theorem
Suppose that (X,𝒮,μ) is a measure space with μ(X)<. If f1,f2,:X are 𝒮-measurable and converge pointwise on X to f:X, then for every ϵ+, there exists E𝒮 such that μ(XE)<ϵ and f1,f2, converges uniformly to f on E.

For m,n1, set Em,n:={xX:|fj(x)f(x)|<1m for every jn}. These sets are measurable, increase in n, and their union is X by pointwise convergence. Since μ(X)<, the measure of an increasing union lets us choose nm such that μ(XEm,nm)<ϵ/2m. Let E=m=1Em,nm. Then μ(XE)<ϵ by countable subadditivity, and the definition of the sets gives uniform convergence on E.

Simple Function
A function is simple if it takes only finitely many values.
Approximation by Simple Functions
Suppose that (X,𝒮) is a measurable space and f:X[,] is 𝒮-measurable. Then there exists a sequence f1,f2,:X such that
  • each fk is simple and 𝒮-measurable,
  • |fk(x)||fk+1(x)||f(x)| for every k+ and every xX,
  • limkfk(x)=f(x) for every xX,
  • if f is bounded, then f1,f2, converges uniformly to f on X.

For f0, divide the range into dyadic intervals of length 2k up to height k, and let fk(x) be the left endpoint of the interval containing f(x), truncated at height k. For general f, apply this construction to the positive and negative parts and subtract. Measurability of the level sets follows from the measurability of f, each fk has finite range, and the dyadic mesh tends to 0, giving pointwise convergence. If f is bounded, the truncation height eventually exceeds the bound, so the dyadic mesh gives uniform convergence.

Luzin's Theorem
Suppose that g: is Borel measurable. For every ϵ+, there exists a closed set F such that omes(F)<ϵ and g|F is continuous on F.

First prove the claim for simple Borel functions by choosing a closed subset of each level set whose complement has arbitrarily small outer measure, using the regularity characterization of Lebesgue measurable sets. On the union of those closed pieces, the simple function is continuous. For a general Borel measurable g, use approximation by simple functions and Egorov's theorem to find a closed set outside a set of outer measure less than ϵ on which the simple approximants converge uniformly. The uniform limit of continuous functions is continuous, so g|F is continuous.

Continuous Extensions from Closed Subsets of the Real Line
Suppose that F is closed and g:F is continuous. Then there exists a continuous function h: such that h|F=g.

The complement of a closed subset of is a countable union of disjoint open intervals. On each bounded complementary interval (a,b), define h by linear interpolation between the endpoint values g(a) and g(b); on unbounded complementary intervals, extend constantly from the finite endpoint. Set h=g on F. The endpoint definitions and the continuity of g on F make h continuous on all of .

Luzin's Theorem, Extension Version
Suppose that E and g:E is Borel measurable. For every ϵ+, there exists a closed set FE and a continuous function h: such that omes(EF)<ϵandh|F=g|F.

Extend g from E to a Borel measurable function on , for example by assigning value 0 off E. By Luzin's theorem, choose a closed set K such that the exceptional set has outer measure less than ϵ and the extension is continuous on K. Put F=EK, using the regularity characterization to replace it by a closed subset of E without increasing the exceptional outer measure. Then apply continuous extension from closed subsets to obtain h with h|F=g|F.

Lebesgue Measurable Function
Suppose that A. A function f:A is Lebesgue measurable if f1(B) is a Lebesgue measurable set for every Borel set B.
Every Lebesgue Measurable Function is Almost Borel Measurable
Suppose that f: is Lebesgue measurable. Then there exists a Borel measurable function g: such that omes({x:g(x)f(x)})=0.

By the definition of Lebesgue measurable set, each inverse image f1(B) differs by a null set from a Borel set. Applying this to the inverse images of intervals with rational endpoints and using countable unions, construct a Borel measurable g that agrees with f outside the union of countably many null exceptional sets. By countable subadditivity, that union has outer measure 0, which is exactly the displayed conclusion.