Direct comparison test: Visual walkthrough — Hinglish deep dive

4.3.8 · D2 · Maths › Calculus III — Sequences & Series › Direct comparison test

Step 0 — "Series" hota kya hai (pile draw karo)

Kisi bhi test se pehle, hum us cheez ko picture karte hain jo hum test kar rahe hain.

KYA HAI. Ek sequence bas ek endless list of numbers hai, har counting position $n = 1, 2, 3, \dots$ ke liye ek number. Hum $n$ -we number ko $a_{n}$ likhte hain (padho "a-sub-n"). Toh $a_{1}$ pehla number hai, $a_{2}$ doosra, aur aage bhi aise hi.

Ek series ka matlab hai hum unhe sab jod dete hain, hamesha ke liye:

1st bar a_{1} + 2nd bar a_{2} + 3rd bar a_{3} + \dots

BARS KYO DRAW KAREIN. Hum literally haath se infinitely many numbers nahi jod sakte. Toh hum har $a_{n}$ ko ek patli bar ka area imagine karte hain jiska height $a_{n}$ hai aur jo position $n$ par khada hai. Series ko add karna = saari bars ke neeche ka total area collect karna.

PICTURE. Figure mein, har burnt-orange bar ki height $a_{n}$ hai. Dhyan do ki bars shrink ho rahi hain — ek hint (guarantee nahi!) ki total area finite reh sakta hai.

Step 1 — Partial sum: ek stopping point tak jodna

Hum ek infinite sum kabhi finish nahi kar sakte, toh hum dekhte hain ki yeh ek bar ek ek bar badhta hai.

KYA HAI. Partial sum $S_{n}$ hai $n$ bars ke baad running total:

S_{n} = pehle n terms, bas itne hi a_{1} + a_{2} + \dots + a_{n} .

Yahan $S_{n}$ ek single number hai: pehle $n$ bars ka accumulated area. Jaise $n$ badhta hai, $S_{n}$ bhi badhta hai — hume totals ka ek naya sequence milta hai $S_{1}, S_{2}, S_{3}, \dots$ .

KYO. "Kya series kisi finite number mein add hoti hai?" yeh ek saaf sawaal ban jaata hai: kya sequence $S_{n}$ kisi fixed number ki taraf settle karta hai jab $n \to \infty$ ? ( $n \to \infty$ ka matlab hai "jaise $n$ hamesha ke liye chalte rehta hai.") Humne ek impossible infinite addition ko ordinary numbers ki limit se replace kar diya.

PICTURE. Plum staircase: $S_{n}$ se $S_{n + 1}$ tak har step exactly ek naya bar hai jiska height $a_{n + 1}$ hai.

Definition Arrow "

$\to$ " ka matlab asal mein kya hai

$S_{n} \to L$ (padho " $S_{n}$ , $L$ ki taraf converge karta hai") ka matlab hai: $L$ ke aas paas koi bhi chota sa tolerance band choose karo, chahe kitna bhi patla — maano half-width $ε$ ki ek strip (ek chhota positive number). Toh kisi position $N$ ke baad se, har total $S_{n}$ us band ke andar aata hai. Formally: har $ε > 0$ ke liye ek $N$ hai jisme $∣ S_{n} - L ∣ < ε$ jab bhi $n \geq N$ ho. Staircase $L$ par lock ho jaati hai aur phir kabhi bhatakti nahi.
$S_{n} \to \infty$ (padho "diverge hota hai infinity tak") ka matlab hai: koi bhi height $M$ lo, chahe kitna bhi bada; kisi position $N$ ke baad se har total use exceed karta hai, $S_{n} > M$ for $n \geq N$ . Staircase har ceiling ko paar kar jaati hai.

Toh "series converges" bas " $S_{n} \to L$ kisi finite $L$ ke liye" hai, aur "diverges" hai " $S_{n}$ ka koi finite limit nahi hai."

Step 2 — Staircase sirf upar jaati hai (monotone)

KYA HAI. Kyunki har bar height $a_{n} \geq 0$ hai, gap $S_{n} - S_{n - 1} = a_{n} \geq 0$ hai. Toh:

S_{1} \leq S_{2} \leq S_{3} \leq \dots

Yeh chain padho "har total pehle wale se chhota nahi hai." Jo sequence kabhi decrease nahi karta use monotone increasing kehte hain.

YEH NON-NEGATIVITY KA PAYOFF KYO HAI. Agar koi bar negative ho sakta, toh staircase neeche step kar sakti, wobble kar sakti, aur kabhi settle nahi hoti — poora argument collapse ho jaata. Isliye DCT negative terms forbid karta hai. (Sign-changing sums ke liye aap Absolute convergence ya Alternating Series Test use karte.)

PICTURE. Left panel: saari bars $\geq 0$ hain, staircase steadily rise karta hai. Right panel: ek sneaky negative bar staircase ko dip karta hai — woh counter-example jo hum rule out kar rahe hain.

Step 3 — Domination: ek series doosre ke neeche baithti hai, term by term

Ab comparison partner lao — yahan $b_{n}$ enter karta hai.

KYA HAI. Ek doosra non-negative sequence introduce karo, jise $b_{n}$ likhte hain (benchmark ya comparison sequence), jo $a_{n}$ ko dominate karta hai:

0 \leq hamari bar a_{n} \leq unchi bar b_{n} har n ke liye .

"Dominates" ka matlab bas hai: har position par, $b$ -bar, $a$ -bar se kam se kam utni hi unchi hai. Yeh pehli baar hai $b_{n}$ appear karta hai — yeh ek aisi series hai jo hum already samajhte hain, purpose se choose ki gayi hai us series ke saath sandwich karne ke liye jo hum nahi samajhte.

KYO. Hum usually apni series nahi samajhte, lekin hum ek benchmark jaise p-series ya Geometric series zaroor samajhte hain. Apni bars ko kisi benchmark ke neeche (ya upar) pin karke, benchmark ka jaana-maana behaviour hamare upar leak hota hai.

PICTURE. Har $n$ par, ek chhota orange bar $a_{n}$ ek unchi teal bar $b_{n}$ ke andar nested. Orange area, teal area ke andar trapped hai, column by column.

Step 4 — Ceiling: agar badi series converge karti hai, toh ek finite roof appear hota hai

KYA HAI. Maan lo unchi series converge karti hai — uske totals $T_{n}$ ek finite number ki taraf climb karte hain jise hum naam denge:

B = k = 1 \sum \infty b_{k} (ek finite ceiling) .

Step 3 ke saath chain karke:

S_{n} \leq T_{n} \leq B .

$T_{n} \leq B$ KYO. Beech wali inequality ko ek baat chahiye. Kyunki har $b_{k} \geq 0$ hai, teal staircase $T_{n}$ khud increasing hai (same argument jaisa Step 2 mein tha, ab $b$ 's par apply ho raha hai). Ek increasing sequence kabhi us value ko overshoot nahi karta jis ki taraf woh climb karta hai, toh $T_{n} \leq lim_{m \to \infty} T_{m} = B$ har $n$ ke liye. Yeh loop close karta hai: teal staircase apni limit $B$ ke neeche rehti hai, aur orange wali teal ke neeche — toh orange $B$ ke neeche rehti hai.

Yeh haari orange staircase ke upar height $B$ par ek fixed horizontal ceiling install karta hai. Hamari staircase bounded above hai — is phrase ka matlab hai "kisi fixed line se upar nahi jaati."

PICTURE. Orange staircase rising, teal staircase uske upar rising, aur ek dashed plum line height $B$ par dono ko cap karti hui. Orange steps kabhi plum line ke through nahi ja sakti.

Step 5 — Engine: Monotone Convergence case close karta hai

KYA HAI. Ab hamare paas orange staircase $S_{n}$ ke liye theorem ke exactly do hypotheses hain:

yeh increasing hai (Step 2), aur
yeh $B$ se bounded above hai (Step 4).

YEH FORCED KYO HAI. Ek hamesha-rising staircase jo kabhi ceiling cross nahi kar sakti, kahin bhaag nahi sakti. Woh ek corner mein squeeze ho jaati hai aur kisi limiting height $L \leq B$ par settle ho jaana zaroor hai. Isliye $S_{n} \to L$ , matlab $\sum a_{n}$ converge karta hai. ∎

PICTURE. Orange steps limit line $L$ ke karib se karib aate hain jo ceiling $B$ ke bilkul neeche hai — steps ke beech gaps khatam ho jaate hain.

Step 6 — Mirror case: divergence UP flow karta hai (contrapositive)

KYA HAI. Ab maan lo iska ulta — chhoti series $\sum a_{n}$ diverge karti hai — uski staircase $S_{n} \to \infty$ ho jaati hai (bina bound ke badhti hai). Kyunki Step 3 ne $T_{n} \geq S_{n}$ diya tha, unchi totals aur bhi upar push ho jaate hain:

T_{n} \geq S_{n} \to \infty \Rightarrow T_{n} \to \infty.

Toh $\sum b_{n}$ bhi diverge karta hai. ∎

YEH SIRF DOOSRI USEFUL DIRECTION KYO HAI. Chhoti wali ka blow up karna badi wali (jo uske upar baithti hai) ko bhi blow up karne ke liye force karta hai. Yeh logically Step 5 jaisa hi statement hai, ulta padha gaya (uska contrapositive).

PICTURE. Ek infinite floor: orange staircase har horizontal line ko paar karte hue upar rocket karti hai; teal staircase, hamesha uske upar, bhi upar drag ho jaati hai.

Step 7 — Degenerate & edge cases (jo logon ko trip karate hain)

Case A — inequality sirf "eventually" hold karti hai, aur index shift karna. DCT ko $a_{n} \leq b_{n}$ sirf $n \geq N$ ke liye chahiye (koi starting index). Sum ko $N$ par split karo:

k = 1 \sum \infty a_{k} = ek fixed finite number k = 1 \sum N - 1 a_{k} + tail k = N \sum \infty a_{k} .

Convergence poori tarah tail se decide hoti hai: ek shifted partial sum $S_{n}^{'} = \sum_{k = N}^{n} a_{k}$ define karo jo $N$ se count shuru karta hai. Ek sequence ke har term mein ek constant (finite front chunk) add karna kabhi nahi badalti ki woh finite limit tak settle karta hai ya nahi — bas poori staircase ek fixed amount se upar slide ho jaati hai. Toh front chunk harmless hai, aur tail par inequality $a_{k} \leq b_{k}$ poori tarah hold karti hai. Isliye "eventually" kaafi hai.

Case B — ek zero bar. Agar kuch $a_{n} = 0$ hai, toh staircase us step par simply nahi rise karti. Bilkul allowed: $0 \leq b_{n}$ phir bhi hold karta hai. Zeros harmless hain.

Case C — messy numerator trick. Jab koi term wobble kare, toh ek clean partner paane ke liye use uske maximum se bound karo. Parent se example: kyunki $- 1 \leq sin n \leq 1$ hai,

\frac{2 + sin n}{n ^{2}} \leq \frac{3}{n ^{2}},

jahan $3$ woh sabse bada value hai jo numerator $2 + sin n$ le sakta hai. Ab p-series rule yaad karo: $\sum \frac{1}{n ^{p}}$ converge karta hai exactly jab $p > 1$ aur diverge karta hai jab $p \leq 1$ ho. Yahan $p = 2 > 1$ hai, toh $\sum \frac{1}{n ^{2}}$ converge karta hai, isliye $\sum \frac{3}{n ^{2}} = 3 \sum \frac{1}{n ^{2}}$ bhi converge karta hai, aur hamari wobbly series converge karti hai.

Case D — negative terms present hain. Tab Step 2 bilkul fail ho jaata hai: staircase dip kar sakti hai aur Monotone Convergence ka koi grip nahi hai. DCT simply applicable nahi hai — Absolute convergence ya Limit comparison test ki taraf jao.

PICTURE. Char mini-panels: (A) ignored early bars grey shaded, (B) ek flat zero-height step, (C) wobbly $sin$ bars ek flat $3/ n^{2}$ lid se capped, (D) ek red-crossed dipping staircase.

Ek-picture summary

Upar sab kuch ek single frame mein compress kiya gaya: do staircases, ek doosre ke neeche nested, aur do arrows.

Orange = hamari unknown series $\sum a_{n}$ . Teal = trusted benchmark $\sum b_{n}$ .
Bars satisfy karte hain $0 \leq a_{n} \leq b_{n}$ , toh orange staircase $\leq$ teal staircase har step par.
Down-arrow (convergence): teal ek finite ceiling $B$ tak pahunchti hai ⇒ orange, neeche trapped, zaroor settle karegi. ✅
Up-arrow (divergence): orange $\infty$ tak rocket karti hai ⇒ teal, uske upar forced, zaroor rocket karegi. ✅
Engine throughout: increasing + bounded ⇒ converges, isliye har bar $\geq 0$ honi chahiye.

Recall Feynman: poori walk ek 12-saal ke bachche ko batao

Socho hamesha ke liye coins stack karte rehna, ek din mein ek shrinking stack, aur jaanna chahte ho ki tower kabhi growna band ho jaayegi. Apni tower measure karne ki bajay (impossible), use apne ek dost ki tower ke paas khada karo jise tum already samajhte ho. Pehli baat: kyunki tum sirf add karte ho coins, tumhari tower sirf upar ja sakti hai, kabhi neeche nahi. Doosri baat: agar tumhari tower har din ek dost se chhoti hai jiska tower ceiling par ruk jaata hai, tumhari woh ceiling paar nahi kar sakti — woh bhi ruk jaati hai. Woh hai convergence neeche flow karna. Teesri baat: agar tumhari tower har din ek dost se unchi hai jiska tower sky tak shoot karta hai, tumhari bhi sky tak shoot kar jaati hai. Woh hai divergence upar flow karna. Woh magic rule jo kehti hai "hamesha-rising-but-capped matlab settle ho jaayegi" Monotone Convergence Theorem hai — aur woh sirf isliye kaam karta hai kyunki tum kabhi coins nahi nikalte, matlab har din ka pile $\geq 0$ hai. Woh sirf do comparisons hain jo tumhe kuch batate hain; dono doosri directions khamosh hain.

Recall checks

S_{n}

ki kaunsi do properties Monotone Convergence ko proof finish karne deti hain?

Yeh increasing hai (from

a_{n} \geq 0

) aur bounded above hai (by

B = \sum b_{n}

Terms ki non-negativity essential kyon hai?

Sirf tabhi partial-sum staircase monotone increasing hai; ek negative bar ise dip karne deta hai aur engine fail ho jaata hai.

Agar

\sum b_{n}

converge karta hai aur

0 \leq a_{n} \leq b_{n}

hai, toh

S_{n}

ki ceiling kya hai?

B = \sum_{k = 1}^{\infty} b_{k}

, kyunki

S_{n} \leq T_{n} \leq B

T_{n} \leq B

kyon hai?

Kyunki

b_{k} \geq 0

T_{n}

ko increasing banata hai, aur ek increasing sequence kabhi apni limit

B = lim T_{m}

ko overshoot nahi karta.

Convergence kis direction mein flow karta hai, small-to-big ya big-to-small?

Big-to-small: ek convergent badi series chhoti wali ko converge hone ke liye trap karti hai.

Kaunse do comparison directions koi information nahi dete?

Chhoti series ka converge karna, aur badi series ka diverge karna.

Wobbly

\frac{2 + s i n n}{n ^{2}}

ko hum kaise handle karte hain?

Numerator ko uske max se bound karo (

2 + sin n \leq 3

) taaki

\leq \frac{3}{n ^{2}}

mile, jo ek convergent

p

-series ka multiple hai.

Kin

p

ke liye

p

-series

\sum 1/ n^{p}

converge karta hai?

Exactly jab

p > 1

ho (yeh

p \leq 1

ke liye diverge karta hai).

S_{n} \to L

formally kya matlab hai?

Har

ε > 0

ke liye ek

N

hai jisme

∣ S_{n} - L ∣ < ε

for all

n \geq N

— totals

L

ke aas paas kisi bhi tolerance band mein lock ho jaate hain.