Frobenius method — regular singular points: Visual walkthrough

4.6.18 · D2 · Maths › Ordinary Differential Equations › Frobenius method — regular singular points

Shuru karne se pehle, ek vaada: end tak aap dekhenge kyun exponent $r$ choose nahi kiya jaata balki equation khud force karti hai use, aur kyun sirf $x$ ki sabse choti power hi use dhundhne ke liye matter karti hai.

Step 1 — Ek plain power series yahan shuru bhi kyun nahi ho sakti

KYA. Hum ek aisi equation dekhte hain jiske coefficients $x = 0$ par blow up kar jaate hain, aur dekhte hain ek plain Taylor series kaise fail hoti hai.

Write any second-order linear ODE in standard form (divide karo jab tak $y^{''}$ akela na khade ho): $curvature y^{''} + friction coeff. P (x) slope y^{'} + spring coeff. Q (x) height y = 0.$

$y$ unknown function hai (socho: time $x$ par jhule ki height).
$y^{'}$ uska slope hai, $y^{''}$ uski curvature.
$P (x), Q (x)$ known functions hain — ye "coefficients" hain. Ordinary point par ye finite hote hain; yahan ye explode karte hain jab $x \to 0$ .

KYUN. Ek plain power series $y = \sum a_{n} x^{n}$ smooth hoti hai — $x = 0$ par uski finite value aur finite slope hoti hai. Lekin agar $P$ ya $Q$ wahan infinity pe jaate hain, toh asli solution khud behave kar sakti hai (jaise $x^{1/2}$ ya $x^{- 1}$ ), jo koi ordinary Taylor series represent nahi kar sakti. Isliye plain method shuru hone se pehle hi doomed hai.

PICTURE. Blue curve $x^{1/2}$ ka $x = 0$ par vertical tangent hai — infinite slope. Koi bhi polynomial-jaisi Taylor curve (amber) use match nahi kar sakti.

Figure — Frobenius method — regular singular points

Ise ordinary point se compare karo, jahan ek clean Taylor series kaam karti hai.

Step 2 — Coefficients ko tame karna: RSP test

KYA. Hum check karte hain ki blow-up itni mild hai ki fix ho sake, $P$ aur $Q$ ko exactly sahi powers of $x$ se multiply karke.

Tamed coefficients define karo: $p (x) = one factor x P (x), q (x) = two factors x^{2} Q (x) .$

$p (x) = x P (x)$ : $x$ ki ek power $P$ ke $1/ x$ blow-up ko cancel karti hai.
$q (x) = x^{2} Q (x)$ : do powers $Q$ ke $1/ x^{2}$ blow-up ko cancel karti hain.

Agar dono $p$ aur $q$ $x = 0$ par analytic (finite, Taylor-expandable) niklen, toh $x = 0$ ek regular singular point (RSP) hai aur Frobenius kaam karne ki guarantee hai.

In exact powers ki KYUN zaroorat? Kyunki sabse bada blow-up jo ek single factor $x^{r}$ equation $x^{2} y^{''} + x p y^{'} + q y = 0$ mein balance kar sakta hai woh hai $P \sim 1/ x$ aur $Q \sim 1/ x^{2}$ . Isse kam ya barabar — hum cope kar lete hain. Isse zyada — point irregular hai aur method kaam nahi karta.

PICTURE. $P (x) = 3/ (2 x)$ (cyan) infinity tak rocket karta hai; $x$ se multiply karo aur amber curve $x P$ $x = 0$ par flat hokar finite value $\frac{3}{2}$ par land karti hai. Yahi finite landing value hai jise hum $p_{0}$ kahenge.

Analyticity yahan matlab hai "genuinely radius of convergence hai".

Step 3 — Frobenius ansatz: Taylor idea churaao, usmein magic power daalo

KYA. Hum ek aisi solution shape guess karte hain jo ek Taylor series hai unknown power $x^{r}$ se multiply karke.

$y = \sum_{n = 0}^{\infty} a_{n} x^{n + r}, a_{0} \neq = 0.$

$x^{r}$ = magic leading power, $r$ unknown (ho sakta hai $\frac{1}{2}$ , $- 1$ , kuch bhi).
$\sum a_{n} x^{n}$ = upar sawaar ordinary smooth Taylor part.
$a_{0} \neq = 0$ : bilkul pehla coefficient nonzero hai, isliye $x^{r}$ genuinely series ki lowest power hai.

$a_{0} \neq = 0$ KYUN? Agar $a_{0} = 0$ hota, toh series actually $x^{r + 1}$ se shuru hoti — lekin tab hum use leading power kahte aur $r$ rename karte. $a_{0} \neq = 0$ insist karna $r$ ko series ke asli bottom par pin karta hai, koi ambiguity nahi.

PICTURE. Amber envelope $x^{r}$ overall shape set karta hai $0$ ke paas; cyan smooth wiggle $\sum a_{n} x^{n}$ usse decorate karta hai. Product (white) hamara candidate $y$ hai.

Step 4 — Ansatz ko term by term differentiate karo

KYA. Hum $y^{'}$ aur $y^{''}$ compute karte hain taaki ODE mein substitute kar sakein.

Har baar exponent niche laao (yahi hota hai jab ek power differentiate karte hain): $y^{'} = \sum_{n} a_{n} old exponent (n + r) x^{n + r - 1},$ $y^{''} = \sum_{n} a_{n} two exponents pulled down (n + r) (n + r - 1) x^{n + r - 2} .$

Har differentiation term $n$ ko uske current exponent se multiply karta hai aur woh exponent $1$ se kam ho jaata hai.
Toh $y^{''}$ mein product $(n + r) (n + r - 1)$ hota hai — yeh "do baar differentiate kiya" ka signature hai.

KYUN. ODE $y, y^{'}, y^{''}$ ke terms mein likhi hai. Ise unknowns $a_{n}$ aur $r$ ki equations mein convert karne ke liye, teeno ko same variable $x$ mein series ki tarah express karna zaroori hai.

PICTURE. Exponents ki ek "staircase": $y$ height $n + r$ par hai, $y^{'}$ ek step niche $n + r - 1$ par, $y^{''}$ do steps niche $n + r - 2$ par. Multipliers arrows par dikhte hain.

Step 5 — $x^{2}$ se multiply karo aur sirf LOWEST power padhо

KYA. ODE ko clean form $x^{2} y^{''} + x p (x) y^{'} + q (x) y = 0$ mein rakhte hain aur sirf $x$ ki sabse choti power collect karte hain, jo $x^{r}$ hai ( $n = 0$ term).

Derivatives ko $x^{2}$ aur $x$ se multiply karne par har exponent wapas $n + r$ par aa jaata hai: $x^{2} y^{''} \to a_{0} (r) (r - 1) x^{r}, x p y^{'} \to a_{0} p_{0} r x^{r}, q y \to a_{0} q_{0} x^{r},$ jahan $p_{0} = p (0) = lim_{x \to 0} x P$ aur $q_{0} = q (0) = lim_{x \to 0} x^{2} Q$ Step 2 se landing values hain (lowest power par har tamed series ka sirf constant term bachta hai).

$r (r - 1)$ = $n = 0$ par $x^{2} y^{''}$ ka contribution.
$p_{0} r$ = $n = 0$ par $x p y^{'}$ ka contribution.
$q_{0}$ = $n = 0$ par $q y$ ka contribution.

Sirf lowest power KYUN? $x$ ki har power ko alag alag vanish karna padega taaki poora sum zero ho. Sabse choti wali, $x^{r}$ , woh akeli jagah hai jahan $r$ akela appear karta hai (koi $a_{1}, a_{2}, \dots$ mix nahi hote). Isliye yeh $r$ ke liye ek clean equation deta hai.

PICTURE. Ek stacked-coefficient tower: $x^{r}$ floor teen pieces $r (r - 1)$ , $p_{0} r$ , $q_{0}$ collect karta hai; upar ke floors ( $x^{r + 1}, \dots$ ) mein baad wale $a_{n}$ hain aur unhein side mein rakh dete hain.

Step 6 — Indicial equation saamne aati hai

KYA. $x^{r}$ ka coefficient zero set karo. Kyunki $a_{0} \neq = 0$ , hum use divide out kar sakte hain.

$a_{0} [r (r - 1) + p_{0} r + q_{0}] = 0 ⟹ r (r - 1) + p_{0} r + q_{0} = 0$

$F (r)$ = indicial polynomial — $r$ mein ek plain quadratic.
Iske roots hi woh allowed magic powers hain; koi aur series lead nahi kar sakta.

Yeh quadratic KYUN hai. Ek second-order ODE ke do independent solutions hote hain. Ek quadratic ke do roots hote hain. Ek doosre ko balance karta hai: $x = 0$ ke paas do behaviours ke liye do exponents.

PICTURE. Parabola $F (r)$ horizontal axis ko $r_{1}$ aur $r_{2}$ par cross karti hai. Woh do crossings hi woh akele exponents hain jo equation allow karti hai.

Euler–Cauchy equation se compare karo, jiska characteristic equation exactly yahi indicial equation hai $p_{0}, q_{0}$ constant ke saath — RSP case uska curved generalisation hai.

Step 7 — Edge cases: root difference sab decide karta hai

KYA. Hum $r_{1} - r_{2}$ ke hisaab se classify karte hain, kyunki woh difference control karta hai ki do series collide karengi ya nahi.

Uper wale coefficients recurrence follow karte hain ( $x^{n + r}$ ka coefficient): $F (n + r) a_{n} = - \sum_{k = 1}^{n} [p_{k} (n - k + r) + q_{k}] a_{n - k} .$ Takleef sirf tab aati hai jab $F (n + r) = 0$ — yaani jab $n + r_{2}$ doosre root $r_{1}$ par land karta hai.

Case 1 · $r_{1} - r_{2} \in / Z$ : $F (n + r_{2})$ kabhi zero nahi hita → do clean independent series. Wronskian se check karo ki woh independent hain.
Case 2 · $r_{1} = r_{2}$ : sirf ek exponent exist karta hai; doosra solution forced hai $ln x$ carry karne ke liye.
Case 3 · $r_{1} - r_{2} = N \in Z^{+}$ : step $n = N$ par, $F (r_{2} + N) = F (r_{1}) = 0$ → possible division by zero → $ln x$ aa sakta hai (coefficient $κ$ zero bhi ho sakta hai).

Log KYUN? Jab $F (n + r_{2}) = 0$ hota hai toh recurrence zero se divide karne ki demand karta hai — power series doosre solution ke liye jagah se bahar ho jaati hai. Tab $y_{2} = v y_{1}$ ke saath Reduction of order mein $\int \frac{d x}{x} = ln x$ produce hota hai: mathematics khud genuinely alag doosra solution supply karta hai.

PICTURE. Exponents ki teen number lines. Case 1: powers ke do combs interleave karte hain, kabhi overlap nahi. Case 2: woh coincide karte hain (ek comb). Case 3: woh step $N$ par collide karte hain (amber marked) — danger point.

Equal-root Case 2 exactly wahi hai jo Bessel's equation of order 0 mein hota hai ( $r_{1} = r_{2} = 0$ ), jisse $J_{0}$ milta hai aur log-carrying $Y_{0}$ . Legendre's equation ordinary point par hai, isliye wahan koi $x^{r}$ ki zaroorat nahi.

Ek-picture summary

Yeh single blueprint poora pipeline compress karta hai: blow-up → tame with $x$ aur $x^{2}$ → RSP? → ansatz $x^{r} \sum a_{n} x^{n}$ → lowest power se $F (r) = 0$ milta hai → root difference case pick karta hai.

Recall Poore walkthrough ki Feynman retelling

Ek jhule ki zanjeer pivot $x = 0$ par hi tooti hui hai, isliye coefficients $P, Q$ wahan infinity tak chillate hain (Step 1). Hum test karte hain ki toot kitni buri hai: $P$ ko $x$ se aur $Q$ ko $x^{2}$ se multiply karo; agar dono finite numbers $p_{0}, q_{0}$ par settle ho jaayein, toh toot tame hai — ek regular singular point (Step 2). Ab hum ek clever cheat karte hain: guess karo ki motion $x^{r}$ (ek magic power) times ek ordinary smooth wiggle jaisi lagti hai, pehla coefficient nonzero ho taaki $x^{r}$ genuinely bottom ho (Step 3). Hum is guess ko differentiate karte hain (Step 4), use tidy $x^{2} y^{''} + x p y^{'} + q y = 0$ mein plug karte hain, aur sirf sabse choti power $x^{r}$ ko dekhte hain, kyunki wahi ek jagah hai jahan $r$ akela khada hota hai (Step 5). Uska coefficient zero set karne se ek tiny quadratic $r (r - 1) + p_{0} r + q_{0} = 0$ nikal aati hai — indicial equation (Step 6). Iske do roots hi allowed magic powers hain. Aakhir mein unka difference poori kahaani batata hai: zyada door (non-integer) → do clean series; equal ya poora-number gap → do solutions ek jaisi banne ki koshish karti hain, recurrence zero se divide karti hai, aur ek genuinely alag doosra solution bachane ke liye $ln x$ andar aa jaata hai (Step 7).

Active Recall

Recall Derivation par self-test

Kaunsi single power of $x$ indicial equation deti hai, aur kyun? ::: Sabse choti, $x^{r}$ ( $n = 0$ term), kyunki wahan $r$ bina kisi aur $a_{n}$ ke mix hue appear karta hai. $p_{0}$ aur $q_{0}$ limits ke roop mein kya hain? ::: $p_{0} = lim_{x \to 0} x P (x)$ , $q_{0} = lim_{x \to 0} x^{2} Q (x)$ . $a_{0} \neq = 0$ kyun hona chahiye? ::: Taaki $x^{r}$ genuinely lowest power ho; warna hum $r$ rename karte. Exactly kya cheez doosre solution mein $ln x$ trigger karti hai? ::: $F (n + r_{2}) = 0$ kisi step $n$ par, jo recurrence ko zero se divide karvaata hai.