Real analysis — rigorous epsilon-delta, metric spaces

1. The epsilon-delta definition of a limit

WHAT each piece means

$|f(x)-L|<\varepsilon$ : output is in the target tube of half-width $\varepsilon$ around $L$ .
$0<|x-a|<\delta$ : input is in a punctured neighborhood of $a$ (we ignore $x=a$ itself — the limit is about the neighbourhood, not the point).
The order of quantifiers is sacred: $\varepsilon$ comes first (the challenge), then $\delta$ may depend on $\varepsilon$ .

2. Deriving a $\delta$ from scratch (the working method)

To prove $\lim_{x\to a}f(x)=L$ you must produce $\delta$ as a function of $\varepsilon$ . The strategy:

Start from the goal $|f(x)-L|<\varepsilon$ .
Factor out $|x-a|$ : write $|f(x)-L| = |x-a|\cdot(\text{stuff})$ .
Bound the "stuff" by a constant $M$ on a small neighborhood (this often needs a preliminary restriction like $|x-a|<1$ ).
Then $|f(x)-L| < M|x-a| < \varepsilon$ holds when $|x-a| < \varepsilon/M$ .
Choose $\delta=\min(1,\varepsilon/M)$ . Done.

Worked example Prove

$\lim_{x\to 3}x^2 = 9$ Goal: $|x^2-9|<\varepsilon$ . Step — factor. $|x^2-9| = |x-3|\,|x+3|$ . Why? We need $|x-3|$ explicit because $\delta$ controls $|x-3|$ . Step — bound $|x+3|$ . Restrict $|x-3|<1 \Rightarrow 2<x<4 \Rightarrow |x+3|<7$ . Why first restrict? Because $|x+3|$ is unbounded globally; we must cage it near $x=3$ . Step — combine. $|x^2-9| < 7|x-3|$ . To beat $\varepsilon$ take $|x-3|<\varepsilon/7$ . Step — choose $\delta$ . $\delta=\min(1,\varepsilon/7)$ . Why min? Both restrictions ( $<1$ and $<\varepsilon/7$ ) must hold simultaneously. ✅ Then $0<|x-3|<\delta \Rightarrow |x^2-9|<7\cdot\frac{\varepsilon}{7}=\varepsilon.$

Figure — Real analysis — rigorous epsilon-delta, metric spaces

3. Continuity (limit that hits the point)

4. Metric spaces — abstracting "distance"

Standard metrics

Space	Metric $d(x,y)$	Name
$\mathbb{R}$	$\lvert x-y\rvert$	usual
$\mathbb{R}^n$	$\sqrt{\sum (x_i-y_i)^2}$	Euclidean
$\mathbb{R}^n$	$\sum\lvert x_i-y_i\rvert$	taxicab/ $\ell^1$
$\mathbb{R}^n$	$\max_i\lvert x_i-y_i\rvert$	Chebyshev/ $\ell^\infty$
any set	$1$ if $x\ne y$ , else $0$	discrete

5. Forecast-then-Verify

Recall Forecast before reading on

Q: For $\lim_{x\to 4}\sqrt{x}=2$ , predict a working $\delta(\varepsilon)$ . Verify: $|\sqrt x-2| = \frac{|x-4|}{\sqrt x+2} \le \frac{|x-4|}{2}$ (since $\sqrt x\ge0$ ). So $|x-4|<2\varepsilon$ works: $\delta=2\varepsilon$ . Why divide? Rationalise $\sqrt x-2$ by multiplying by $\frac{\sqrt x+2}{\sqrt x+2}$ to expose $|x-4|$ .

Flashcards

The rigorous definition of

\lim_{x\to a}f(x)=L

\forall\varepsilon>0\,\exists\delta>0:\ 0<|x-a|<\delta\Rightarrow|f(x)-L|<\varepsilon

Which quantifier comes first,

\varepsilon

\delta

, and why

\varepsilon

first;

\delta

may depend on

\varepsilon

(challenge then response)

Why "

0<|x-a|

" (punctured) in the limit

The limit ignores

x=a

; it's about behaviour near

a

, not the value at

a

General recipe to find

\delta

Factor

|f(x)-L|=|x-a|\cdot M

, cap with

|x-a|<1

, then

\delta=\min(1,\varepsilon/M)

\delta

for

\lim_{x\to3}x^2=9

\delta=\min(1,\varepsilon/7)

Difference: continuity vs uniform continuity

Continuity allows

\delta

to depend on the point; uniform continuity needs one

\delta

for all points

The three metric axioms

d(x,y)=0\!\iff\!x=y

; symmetry; triangle inequality

Discrete metric

d(x,y)=0

x=y

, else

1

Definition of open ball

B(a,r)

\{x: d(x,a)<r\}

Limit in a metric space

\forall\varepsilon\,\exists N:\ n\ge N\Rightarrow d(x_n,L)<\varepsilon

Inequality proving the Euclidean triangle inequality

Cauchy–Schwarz,

x\cdot y\le\|x\|\|y\|

Recall Feynman: explain to a 12-year-old

Imagine a dartboard. A friend says "I bet you can't land within 1 cm of the bullseye." You say "fine, just let me stand close enough to the board." If for every distance they pick — 1 cm, 1 mm, even a hair's width — you can always find a spot to stand that guarantees you hit that close, then we say your darts truly approach the bullseye. That's a limit. $\varepsilon$ is how close they demand; $\delta$ is how close you stand. A "metric" is just any fair ruler for measuring how far two things are, as long as the ruler is honest: zero distance only for identical things, same both ways, and no shortcut beats going straight (triangle inequality).

Connections

Limits and Continuity — the calculus this rigorises
Sequences and Series — convergence is the $N$ – $\varepsilon$ version
Cauchy Sequences and Completeness — when limits exist inside the space
Topology — open sets generalise away from any metric
Cauchy–Schwarz Inequality — engine of the Euclidean triangle inequality
Uniform Continuity and Compactness — Heine–Cantor theorem

Concept Map

Hinglish (regional understanding)

Intuition Hinglish mein samjho

Dekho, calculus mein hum bolte hain " $x$ , $a$ ke paas jaa raha hai, toh $f(x)$ , $L$ ke paas jaata hai" — par "paas jaana" ka exact matlab kya? Yahin epsilon-delta ka kamaal aata hai. Isko ek challenge game samjho: tumhara dost challenge deta hai " $f(x)$ ko $L$ ke $\varepsilon$ (chhoti) doori ke andar lao". Tum jawab dete ho "theek hai, bas $x$ ko $a$ ke $\delta$ doori ke andar rakho, main guarantee deta hoon". Limit tabhi exist karti hai jab har $\varepsilon$ ke liye, chahe kitna bhi micro ho, tum koi $\delta$ dhoond pao. Sabse important baat — pehle $\varepsilon$ aata hai, phir $\delta$ ( $\delta$ usually $\varepsilon$ par depend karta hai).

$\delta$ nikaalne ka trick simple hai: $|f(x)-L|$ ko $|x-a|$ se factor karo, baaki "stuff" ko ek constant $M$ se bound karo (iske liye pehle $|x-a|<1$ maan lo taaki stuff bound rahe), phir $\delta=\min(1,\varepsilon/M)$ . Jaise $x^2$ wale example mein $M=7$ aaya, toh $\delta=\min(1,\varepsilon/7)$ . Yaad rakho — straight $\delta=\varepsilon$ sirf $f(x)=x$ ke liye chalta hai; jahan function steep hota hai wahan $\delta$ ko chhota karna padta hai.

Ab metric space ka idea: humne in saare proofs mein sirf "doori" $|x-y|$ use ki. Toh kyun na doori ko abstract kar dein? Metric $d$ bas teen rule maanta hai — doori zero sirf same point ke liye, dono taraf same (symmetry), aur triangle inequality ( $d(x,z)\le d(x,y)+d(y,z)$ , yaani seedha jaana hi sabse chhota). Bas itne se hi poori analysis vectors, functions, kahin bhi chal jaati hai. Yeh elite-level isliye matter karta hai kyunki engineering, ML, optimisation — sab jagah "convergence" aur "distance" yahi rigorous foundation pe khade hain.

Real analysis — rigorous epsilon-delta, metric spaces

1. The epsilon-delta definition of a limit

2. Deriving a $\delta$ from scratch (the working method)

3. Continuity (limit that hits the point)

4. Metric spaces — abstracting "distance"

5. Forecast-then-Verify

Flashcards

Connections

Concept Map

Hinglish (regional understanding)

Go deeper — visual, from zero

Test yourself — Advanced Topics (Elite Level)

Connections