Introduction

In the previous lesson, we studied convergence in measure.

We learned that:

$$f_n \to f \quad \text{in measure}$$

means:

$$\mu\left({x:|f_n(x)-f(x)|>\varepsilon}\right)\to0$$

for every:

$$\varepsilon>0$$

We also studied almost everywhere convergence:

$$f_n(x)\to f(x)$$

for all points except possibly a set of measure zero.

Today we investigate a surprising question:

If a sequence converges almost everywhere, how close is it to converging uniformly?

At first glance, these notions seem completely different.

• Pointwise convergence is local.
• Uniform convergence is global.

Remarkably, on finite measure spaces, they are almost the same.

This is the content of Egoroff’s Theorem.

It is one of the most beautiful and subtle results in classical measure theory.

---

The Problem

Recall the hierarchy:

$$\text{Uniform Convergence} \implies \text{Almost Everywhere Convergence}$$

The converse is false.

---

Example

Consider:

$$f_n(x)=x^n$$

on:

$$[0,1]$$

We know:

$$f_n(x)\to0$$

for every:

$$x<1$$

Thus:

$$f_n\to0$$

almost everywhere.

However, the convergence is not uniform.

Indeed:

$$\sup_{x\in[0,1]}|x^n-0|=1$$

for every:

$$n$$

Therefore:

$$f_n \not\to 0$$

uniformly.

---

A Natural Question

Although uniform convergence fails on the entire space, perhaps it fails only because of a very small exceptional region near:

$$x=1$$

If we remove that region, does uniform convergence appear?

Egoroff’s Theorem says:

Yes.

---

Statement of Egoroff’s Theorem

Let:

$$\mu(X)<\infty$$

and suppose:

$$f_n \to f$$

almost everywhere on:

$$X$$

Then for every:

$$\varepsilon>0$$

there exists a measurable set:

$$E\subseteq X$$

such that:

$$\mu(E)<\varepsilon$$

and:

$$f_n\to f$$

uniformly on:

$$X\setminus E$$

---

What the Theorem Says

Suppose convergence is almost everywhere.

Then:

• Remove an arbitrarily small set.
• On what remains, convergence becomes uniform.

In symbols:

$$\text{Almost Everywhere Convergence} \implies \text{Almost Uniform Convergence}$$

on finite measure spaces.

---

Why This Is Amazing

Uniform convergence is much stronger than almost everywhere convergence.

Yet Egoroff’s Theorem says:

Almost everywhere convergence is nearly uniform convergence.

The only obstruction is a set of arbitrarily small measure.

This is one of the first indications that measure theory often ignores tiny exceptional sets.

---

Visual Interpretation

Imagine:

$$X=[0,1]$$

Suppose convergence behaves badly only near a few troublesome points.

Egoroff’s Theorem allows us to place all bad behavior inside a tiny set:

$$E$$

After removing:

$$E$$

the sequence behaves uniformly.

---

Example 1: Powers of x

Consider:

$$f_n(x)=x^n$$

on:

$$[0,1]$$

We know:

$$f_n(x)\to0$$

for:

$$x<1$$

Choose:

$$E=[1-\delta,1]$$

Then:

$$\mu(E)=\delta$$

which can be made arbitrarily small.

On:

$$[0,1-\delta]$$

we have:

$$|x^n|\le(1-\delta)^n$$

and:

$$ (1-\delta)^n\to0 $$

uniformly.

Thus Egoroff’s conclusion holds.

---

Why Finite Measure Is Necessary

The theorem requires:

$$\mu(X)<\infty$$

This assumption cannot be removed.

---

Counterexample

Consider:

$$f_n(x)=\mathbf1_{[n,n+1]}(x)$$

on:

$$\mathbb R$$

For every fixed:

$$x$$

eventually:

$$f_n(x)=0$$

Thus:

$$f_n\to0$$

everywhere.

However:

No matter what finite set we remove,

the moving spike continues to wander through the remaining space.

Uniform convergence never appears.

The reason is that:

$$\lambda(\mathbb R)=\infty$$

Egoroff’s Theorem fails.

---

Why the Proof Works

The proof is a masterpiece of measure-theoretic reasoning.

---

Step 1

Since:

$$f_n\to f$$

almost everywhere,

for each:

$$k$$

there exists:

$$N_k(x)$$

such that:

$$|f_n(x)-f(x)|<\frac1k$$

whenever:

$$n\ge N_k(x)$$

---

Step 2

Construct sets where convergence has not yet reached the desired accuracy.

---

Step 3

Show these sets shrink toward measure zero.

---

Step 4

Choose sufficiently large indices so that the total exceptional measure becomes smaller than:

$$\varepsilon$$

---

Step 5

Outside the exceptional set, uniform convergence follows.

The proof is a beautiful application of countable additivity and finite measure.

---

Almost Uniform Convergence

Egoroff’s Theorem motivates a new definition.

We say:

$$f_n\to f$$

almost uniformly if for every:

$$\varepsilon>0$$

there exists:

$$E$$

with:

$$\mu(E)<\varepsilon$$

such that:

$$f_n\to f$$

uniformly on:

$$X\setminus E$$

Egoroff’s Theorem states:

$$\text{Almost Everywhere Convergence} \implies \text{Almost Uniform Convergence}$$

when:

$$\mu(X)<\infty$$

---

Relationship to Convergence in Measure

Recall:

On finite measure spaces:

$$\text{Almost Everywhere Convergence} \implies \text{Convergence in Measure}$$

Egoroff’s Theorem tells us even more:

$$\text{Almost Everywhere Convergence} \implies \text{Almost Uniform Convergence}$$

Thus almost everywhere convergence is surprisingly powerful.

---

Why Analysts Love Egoroff’s Theorem

Many proofs require uniform convergence because it is easier to work with.

Unfortunately, only almost everywhere convergence may be available.

Egoroff’s Theorem provides a bridge.

It allows us to replace:

$$\text{a.e. convergence}$$

with:

$$\text{uniform convergence outside a tiny set}$$

This is often enough for the proof to proceed.

---

Example in Probability

Suppose:

$$X_n\to X$$

almost surely.

Almost sure convergence is simply almost everywhere convergence on a probability space.

Since:

$$P(\Omega)=1$$

Egoroff’s Theorem applies.

For every:

$$\varepsilon>0$$

there exists:

$$A$$

such that:

$$P(A)<\varepsilon$$

and:

$$X_n\to X$$

uniformly on:

$$\Omega\setminus A$$

This result is frequently used in asymptotic statistics.

---

Philosophical Meaning

One of the deepest ideas in measure theory is:

Small sets often do not matter.

Egoroff’s Theorem illustrates this beautifully.

By sacrificing an arbitrarily small set, we gain a dramatically stronger form of convergence.

This philosophy appears repeatedly throughout modern analysis.

---

Connection to Lusin’s Theorem

The next lesson presents a complementary result.

Egoroff’s Theorem says:

Measurable convergence is almost uniform.

Lusin’s Theorem says:

Measurable functions are almost continuous.

Together, these theorems form a powerful pair.

They reveal that measurable objects behave much more like continuous objects than one might initially expect.

---

Connection to Alain Connes

Egoroff’s Theorem teaches a lesson that becomes increasingly important throughout modern mathematics:

Exact behavior everywhere is often less important than behavior outside a negligible set.

In classical geometry, every point matters.

In measure theory, tiny exceptional sets can often be ignored.

Later, in operator algebras and noncommutative geometry, even the notion of individual points begins to disappear.

Egoroff’s Theorem is one of the earliest examples of this shift in perspective.

---

Key Concepts Learned

By the end of this lesson you should understand:

• Egoroff’s Theorem requires:

$$\mu(X)<\infty$$

• If:

$$f_n\to f$$

almost everywhere, then for every:

$$\varepsilon>0$$

there exists:

$$E$$

with:

$$\mu(E)<\varepsilon$$

such that convergence is uniform on:

$$X\setminus E$$

• Almost everywhere convergence implies almost uniform convergence on finite measure spaces.
• The theorem fails on infinite measure spaces.
• Egoroff’s Theorem creates a bridge between pointwise and uniform convergence.
• Small exceptional sets can often be ignored in analysis.

---

Looking Ahead

Measure Theory Lesson 27: Lusin’s Theorem

In the next lesson, we will prove one of the most surprising results in measure theory:

Every measurable function is almost continuous.

Lusin’s Theorem shows that although measurable functions may appear highly irregular, they can be approximated by continuous functions on sets of arbitrarily large measure. This theorem is one of the deepest links between measure theory and topology.