Intuition of Epsilon-Delta Proofs

It took a while for me to grasp that epsilon-delta proofs are about control. We want to control the output of a function $f(x)$ by restricting the input. It's about whether bad behavior can be excluded by shrinking neighborhoods.

In an $\epsilon\text{-}\delta$ proof, $\epsilon$ represents how much error is allowed in the output, and $\delta$ represents how tightly we restrict the input.

Proving Existence of a Limit

$\forall\space \epsilon>0,\ \exists{}\delta>0,\ \forall\space x \in{}D(f),\ 0<|x-c|<\delta, \ \text{s.t.} \ |f(x)-L|<\epsilon$ $\epsilon$ is fixed first. It's arbitrary. $\delta$ is chosen in response to $\epsilon$, and we must handle all $x$ in the $\delta$-neighborhood.

Proving a Limit Doesn't Exist

$\exists\space \epsilon>0,\ \forall\delta>0,\ \exists\space x \in{}D(f),\ 0<|x-c|<\delta, \ \text{s.t.} \ |f(x)-L|\ge\epsilon$ $L$ is arbitrary, $\epsilon$ is chosen, $\delta$ is arbitrary, and $x$ is chosen to witness failure. Specifically, by "choosing an $\epsilon$" means to assert that the function fails by at least this much, no matter how close you go.

On Restricting Neighborhoods and Boundaries

In a continuity proof you often need two different conditions to be true at the same time:

1. A “geometry/safety” condition (trap $x$ in a nice region): Keep denominators away from 0, keep $x$ positive, keep derivatives bounded, etc.
2. The actual $\varepsilon$-condition: This is usually a simple inequality once the region is safe.

Each condition gives you an upper bound on how small $\delta$ must be:

• Condition 1 says “choose $\delta \le \delta_0(c)$”

• Condition 2 says “choose $\delta \le \delta_1(c,\varepsilon)$”

To satisfy both, you take $\delta := \min(\delta_0,\delta_1)$ You typically pick a convenient requirement like $|x-c|<1 \quad\text{or}\quad |x-c|<c/2$whose only purpose is to force $x$ to be in a region where you can bound the ugly term by a constant. This is not guessing but building a controlled local universe around $c$.

On the Structure of Logical Quantifiers

When we are given: $\forall \varepsilon > 0\; \exists \delta > 0\; \forall x$this literally means:

1. Someone gives you any $ε$. So $ε$ is chosen first.

2. You must respond with a $δ$. So $δ$ can depend on ε.

3. Then nature chooses any $x$ satisfying $|x−c|<δ$. $x$ is chosen after $δ$ so $δ$ cannot depend on $x$.

4. The inequality must hold.

It boils down to logic: A variable can depend on everything quantified before it. It cannot depend on anything quantified after it. Fundamentally, in real analysis we study different notions of continuity to express different notions of strictness; i.e., control of what can depend on what, and what cannot depend on what.