Sequences — convergence, divergence, boundedness, monotonicity: Visual walkthrough

4.3.1 · D2 · Maths › Calculus III — Sequences & Series › Sequences — convergence, divergence, boundedness, monotonici

Yeh walkthrough completeness axiom par lean karta hai aur us loop ko close karta hai jo parent ne khola tha. Yeh parent topic ka visual companion hai.

Step 1 — Ek sequence kya hai, number line par dots ki tarah?

KYA HAI. Ek sequence ek infinite ordered list hai $a_{1}, a_{2}, a_{3}, \dots$ — har counting number $n = 1, 2, 3, \dots$ ke liye ek real number. Ise graph ki tarah draw karne ki jagah, hum har value ko ek horizontal number line par dot ki tarah drop karte hain. Dot par label batata hai list mein uski position; line par jagah batati hai uski value.

YEH PICTURE KYUN. Convergence ek statement hai values kahan pile up hoti hain ke baare mein, na ki graph kaise utha hai ke baare mein. Dots-on-a-line "ek point ke paas pile up hona" literally visible banata hai.

PICTURE. Neeche, $a_{1}$ bilkul left mein hai, aur har baad ka dot thoda sa uske right mein baithta hai. Har dot ke upar label $n$ sirf uska index hai — "main $n$ -waan term hoon."

Figure — Sequences — convergence, divergence, boundedness, monotonicity

Step 2 — "Monotone increasing" kaisa dikhta hai?

KYA HAI. Increasing matlab hai har naya term pichle wale se kam se kam utna hi right mein hai: $a_{n + 1} \geq a_{n} for every n .$

$a_{n + 1}$ — agla term (aane wala dot).
$a_{n}$ — current term.
$\geq$ — "right mein baitha hai, ya uske upar" — dots kabhi left nahi jaate.

YEH KYUN MATTER KARTA HAI. Agar dots sirf rightward march kar sakte hain, toh sirf ek hi cheez unhe rok sakti hai — right side par ek wall. Yahi ek restriction limit ko trap karegi.

PICTURE. Har arrow right ki taraf point kar raha hai. Koi backtracking nahi — yeh monotone increasing ki pehchaan hai.

Recall Dono directions ko ek saath kyun handle nahi karte?

Increasing aur decreasing mirror images hain. Hum increasing carefully prove karte hain; number line ko flip karna (replace $a_{n}$ by $- a_{n}$ ) decreasing case free mein de deta hai. Hum woh mirror Step 8 mein dikhate hain.

Step 3 — "Bounded above" picture mein kya dalta hai?

KYA HAI. Bounded above matlab hai ek number $M$ exist karta hai jise koi bhi term kabhi pass nahi karta: $a_{n} \leq M for every n .$

$M$ — ek upper bound, right side par ek wall.
$a_{n} \leq M$ — har dot hamesha wall ke left mein ya uske upar rehta hai.

KYUN. Step 2 ne rightward-only motion di; Step 3 ne wall di. Ek dot jo hamesha right move karna chahta hai lekin wall cross nahi kar sakta, uske paas koi option nahi siwa iske ki woh bunch up ho — yahi intuition poora proof ban jaata hai.

PICTURE. $M$ par red wall ek barrier hai. Note karo ki $M$ koi ek wall hai, necessarily closest wali nahi — infinitely many walls hain (koi bhi bada number bhi wall hai).

Step 4 — Sabse tight wall: supremum

KYA HAI. Un tamam walls $M$ mein se jo kaam karti hain, ek least (leftmost) hoti hai. Use $L$ kahte hain: $L = sup {a_{1}, a_{2}, a_{3}, \dots} .$

$sup$ ("supremum") — least upper bound: sabse leftmost wall jise abhi bhi koi term cross nahi karti.

YEH KYUN EXIST KARTA HAI — AUR HUMEIN EK AXIOM KYUN CHAHIYE. Ek sabse choti wall kyun exist karni chahiye? Rationals par yeh fail ho sakta hai: terms $1, 1.4, 1.41, 1.414, \dots$ $2$ ki taraf chadh rahe hain, aur har rational wall ko thoda aur tight kiya ja sakta hai bina kisi rational wall ke the smallest hue. Completeness of $\mathbb{R}$ exactly yeh promise hai ki real line par least wall hamesha exist karta hai. Yeh woh deep fact hai jis par MCT tika hua hai — completeness nahi, theorem nahi.

$sup$ KYUN, AVERAGE YA MAX KYUN NAHI? Kyunki hum woh point chase kar rahe hain jis par dots approach karti hain. Ek monotone-increasing pile sirf apni ceiling approach kar sakti hai; ceiling uska least upper bound hai. Max exist nahi kar sakta (dots shayad kabhi pahunch na sakein), lekin $sup$ hamesha karta hai.

PICTURE. Wall ko tab tak left push karo jab tak koi dot poke through na kar le — woh limiting position $L$ hai.

Step 5 — "Least wall" ko ek captured term mein badlo

KYA HAI. Koi bhi tolerance $ε > 0$ fix karo — ise ek tiny gap ki tarah socho. Kyunki $L - ε$ wall nahi hai (Step 4, property 2), koi term us se past slip kar chuki hai: $\exists N : a_{N} > L - ε .$

$ε$ — challenger ki tolerance, ek positive gap chahe kitna bhi chhota ho.
$L - ε$ — ceiling ke thoda left mein ek point.
$a_{N}$ — pehla witness term jo $L - ε$ ke right mein hai.

KYUN. Yeh engine hai. "Least upper bound" decoration nahi hai — yeh ek witness guarantee karta hai. Agar koi term $L - ε$ se past na gai hoti, toh $L - ε$ khud ek wall hoti $L$ se chhoti, jo contradict karta $L$ ke least wall hone ko.

PICTURE. Green band $(L - ε, L]$ target zone hai. Dot $a_{N}$ pehla hai jo us mein land karta hai.

Step 6 — Monotonicity poori tail ko band mein khench laati hai

KYA HAI. Step 5 se, $a_{N} > L - ε$ . Ab Step 2 use karo: sequence kabhi left nahi jaati, isliye $a_{N}$ ke baad ki har term kam se kam $a_{N}$ jaisi hai: $n > N ⟹ a_{n} \geq a_{N} > L - ε .$ Aur Step 4 (property 1) se, koi term ceiling cross nahi karti: $a_{n} \leq L < L + ε$ . Combine karo: $n > N ⟹ L - ε < a_{n} \leq L < L + ε .$

Left inequality: monotonicity (dots witness ke past peeche nahi ja sakte).
Right inequality: wall $L$ (dots ceiling cross nahi kar sakte).

Milkar har tail dot $(L - ε, L + ε)$ mein squeeze ho jaata hai.

YEH FINISH KYUN HAI. Poori tail ko $L$ ke aas-paas half-width $ε$ ke band ke andar squeeze karna word-for-word parent note ki convergence ki $ε$ – $N$ definition hai.

PICTURE. Jab ek dot green band mein enter kar leta hai, monotonicity har baad wale dot ko usi mein rehne par majboor karti hai — left mein witness aur right mein ceiling ke beech trapped.

Step 7 — Degenerate & edge cases (koi gap mat chhodho)

KYA AUR KYUN. Ek careful walkthrough ko har weird input par survive karna chahiye.

Constant sequence $a_{n} = c$ . Yeh increasing hai ( $a_{n + 1} \geq a_{n}$ equality ke saath) aur bounded hai. $sup = c$ , aur har term already $c$ par hai. $c$ par converge karta hai. ✓
Eventually monotone. Sirf tail convergence decide karti hai. Agar sequence pehle million terms mein wobble karti hai aur tab ek wall ke neeche monotonically chadh'ti hai, toh finite head ko chop karo — argument tail par unchanged chalta hai. ✓
Witness $a_{N}$ exactly $L$ ke barabar hai. Allowed: agar koi term actually ceiling tak pahunch jaata hai, monotonicity + wall saare baad wale terms ko exactly $L$ par pin kar deti hai. Phir bhi har band ke andar. ✓
Terms $L$ ke paas aate hain lekin kabhi touch nahi karte (e.g. $a_{n} = 1 - \frac{1}{n}$ , $sup = 1$ , kabhi reach nahi hua). Proof ko kabhi zaroorat nahi padi ki sup attained ho — sirf approached ho. ✓ Isliye humne $sup$ choose kiya, $max$ nahi.

PICTURE. Teen rows: ek flat constant line, ek wobble-then-climb, aur dots ek ceiling ke paas crowd ho rahe hain jo woh kabhi reach nahi karte. Teeno land karte hain.

Recall Agar sequence

bounded above nahi hai toh kya hoga? Tab koi wall exist nahi karti, koi $sup$ nahi, aur ek increasing sequence $+ \infty$ tak march karti hai — yeh infinity par diverge karti hai. MCT apply nahi hota, aur sahi bhi hai: wall hato aur trap khatam.

Step 8 — Decreasing mirror (case closed)

KYA HAI. Ek sequence ke liye jo sirf left move karti hai (decreasing) aur floor $m$ se bounded below hai, poori line reflect karo: $b_{n} = - a_{n}$ study karo. Tab $b_{n}$ increasing aur $- m$ se bounded above hai, isliye Step 6 se $b_{n} \to sup b_{n}$ milta hai. Wapas flip karo: $a_{n} \to - sup {- a_{n}} = in f {a_{n}} .$

$in f$ ("infimum") — greatest lower bound, $sup$ ka mirror; sabse uuncha floor jise koi term cross karke neeche nahi jaati.

KYUN. Koi naya kaam nahi: reflection "least wall on the right" ko "greatest floor on the left" mein badal deta hai. MCT ke dono halves ab hold karte hain.

PICTURE. Upar: decreasing dots $in f$ par floor ki taraf gir rahe hain. Neeche: reflected increasing copy $- in f$ tak chadh rahi hai. Same proof, mirror mein dekha hua.

Ek-picture summary

Sab ek saath: rightward-only dots (monotone), ek red ceiling (bound), least wall $L = sup$ , ek green $ε$ -band, witness $a_{N}$ jo $L - ε$ ko pierce karta hai, aur poori tail band ke andar crush ho gayi.

Recall Feynman retelling — saada language mein poora proof

Socho beads ek wire par slide kar rahi hain, har naya bead pichle se right mein rakha gaya hai, lekin ek wall kisi bead ko ek certain point se aage nahi jaane deti. Beads right jaati rehti hain phir bhi wall nahi tod sakti — toh unhe kahin na kahin crowd up hona hi hoga. Kahan? Wall ko tab tak left slide karo jab tak koi bead poke through na kar le; woh limiting spot $L$ hai. Ab ek game khelo: ek challenger ek tiny gap $ε$ name karta hai aur wall ke bilkul left mein us width ki green strip paint karta hai. Kyunki $L$ sabse tight wall hai, $ε$ se peeche hatna ab wall nahi raha — toh kam se kam ek bead, use $N$ -wein kahte hain, already strip mein enter kar chuki hai. Aur kyunki beads sirf right move karti hain aur kabhi wall nahi todti, har bead $N$ -wein ke baad us entering bead aur wall ke beech stuck hai — strip ke andar. Challenger chahe strip kitni bhi patli banaye, poori tail us mein end up ho jaati hai. "Poori tail har strip ke andar" exactly wahi hai jo "converges to $L$ " ka matlab hai. Ek fact jo humne bahar se liya: ki sabse tight wall exist karti hi hai — yeh real numbers ki completeness hai.

Recall check

Least upper bound

L

kyun exist karna chahiye?

Completeness axiom guarantee karta hai ki

R

mein har nonempty set jo bounded above ho uska ek supremum hoga.

Step 5 mein koi term

L - ε

se kyun exceed karta hai?

Kyunki

L

least wall hai, isliye

L - ε

wall nahi hai — koi term zaroor us se past ja chuki hogi.

Step 6 mein monotonicity kya contribute karti hai?

Yeh witness

a_{N}

ke baad har term ko

a_{N}

ke right mein drag karti hai, isliye poori tail

L - ε

se upar rehti hai.

sup

kyun,

max

kyun nahi?

Limit approach ki ja sakti hai lekin kabhi attain nahi (e.g.

1 - \frac{1}{n} \to 1

);

sup

hamesha exist karta hai,

max

nahi kar sakta.

Decreasing case kaise handle hota hai?

b_{n} = - a_{n}

ke zariye reflect karo; yeh increasing bounded above ban jaata hai, phir wapas flip karo

in f

par convergence pane ke liye.

Agar ek increasing sequence unbounded above ho toh kya hoga?

Koi wall nahi, koi

sup

nahi — yeh

+ \infty

par diverge karta hai.