Initial value problems (IVP) vs boundary value problems (BVP): Question bank

4.7.2 · D5 · Maths › Partial Differential Equations › Initial value problems (IVP) vs boundary value problems (BVP

Traps se pehle, us machinery ko pakka karte hain jis par neeche har sawaal tikaa hua hai.

Definition Is page mein use hone wale words aur symbols

Side conditions = woh extra data (values ya slopes) jo tum ek differential equation ke saath lagaate ho taaki poore family mein se ek solution chun sako. Jab sab ek hi point par hon toh unhe initial conditions kehte hain; jab woh do siron par baanthe hon toh unhe boundary conditions kehte hain. "Side conditions" bas dono ko cover karne wala umbrella word hai.
Independent variable = woh input jis par solution depend karta hai. Hum ise poore mein $x$ likhte hain, lekin $t$ (time) ka matlab bilkul wahi hai — $x$ aur $t$ interchangeable placeholders hain. Space problems mein aksar $x$ use hota hai, time problems mein $t$ ; maths identical hai.
$y_{1} (x), y_{2} (x)$ = homogeneous equation ke do independent basis solutions. 2nd-order linear ODE ka har solution inhin ka koi mix hota hai.
$c_{1}, c_{2}$ = do arbitrary constants (woh "mixing amounts") jo ek 2nd-order general solution carry karta hai. Side conditions theek inhi do numbers ko fix karne ke liye exist karti hain.
$a, b$ = do boundary points (interval ke left aur right ends), jahan $a \neq = b$ .
$α, β$ = do boundary-value targets: $α$ woh demanded value hai $y (a) = α$ left end par, $β$ woh demanded value hai $y (b) = β$ right end par. Yeh woh numbers hain jinhein curve ko hit karna hai.

Formula Linear system step by step banana

Step 1 — general solution se shuru karo. Ek 2nd-order linear ODE ka $y (x) = c_{1} y_{1} (x) + c_{2} y_{2} (x) + y_{p} (x),$ jahaan $y_{p}$ ek particular solution hai (homogeneous problems ke liye $y_{p} = 0$ lo). Jinhe hume fix karna hai woh unknowns hain $c_{1}, c_{2}$ .

Step 2 — left condition $y (a) = α$ impose karo. $x = a$ substitute karo: $c_{1} y_{1} (a) + c_{2} y_{2} (a) + y_{p} (a) = α ⟹ y_{1} (a) c_{1} + y_{2} (a) c_{2} = α - y_{p} (a) .$ Yeh $c_{1}, c_{2}$ mein ek linear equation hai — dhyan do ki $c_{1}, c_{2}$ unknowns hain aur $y_{i} (a)$ sirf numbers hain.

Step 3 — right condition $y (b) = β$ impose karo. $x = b$ substitute karo: $y_{1} (b) c_{1} + y_{2} (b) c_{2} = β - y_{p} (b) .$

Step 4 — dono equations ko ek matrix ki rows mein stack karo. Coefficients padh ke: $M (y_{1} (a) y_{1} (b) y_{2} (a) y_{2} (b)) c (c_{1} c_{2}) = b (α - y_{p} (a) β - y_{p} (b)) .$ $M$ ki har row ek boundary point se aayi hai; har column ek constant ko multiply karta hai. Isliye ek BVP ki kismat ek $2 \times 2$ matrix mein rehti hai.

Step 5 — linear algebra se verdict padho: $det M \neq = 0 \Rightarrow$ unique $c$ ; $det M = 0 \Rightarrow$ koi solution nahi ya infinitely many, yeh $b$ ki compatibility par depend karta hai.

IVP cousin. IVP ke liye dono conditions ek hi point $x_{0}$ par hoti hain: $y (x_{0}) = A$ aur $y^{'} (x_{0}) = B$ impose karo. Ab dono rows $y$ aur $y^{'}$ ko usi point par use karti hain, toh coefficient matrix ka determinant hoga $W = y_{1} (x_{0}) y_{2}^{'} (x_{0}) - y_{2} (x_{0}) y_{1}^{'} (x_{0}),$ yani Wronskian — independent $y_{1}, y_{2}$ ke liye kabhi zero nahi hota, isliye IVP system hamesha uniquely solvable hota hai.

Shuru karne se pehle teen anchor pictures apne dimaag mein rakho — literally inhe ek baar sketch karo:

IVP picture — axis par ek dot jisme se ek arrow nikalta hai: ek hi jagah sab data, aage badho. (Ek point banao, ek outgoing arrow.)
BVP picture — interval ke dono siron par do dots aur ek curve dono ko touch karti hui: edges par data, beech fill karo. (Do point banao, ek curve unhe bridge karti hui.)
Matrix $M$ — Step 4 se woh $2 \times 2$ coefficient grid jiska determinant BVP ki kismat decide karta hai; iska IVP cousin Wronskian $W$ hai, single initial point par evaluate kiya hua. (Grid banao, ek row per boundary point.)

Sach ya jhooth — justify karo

Yahan har "jawab" justification hai. Pehle sach/jhooth bolo, phir wajah check karo.

2 side conditions wali 2nd-order ODE hamesha uniquely solvable hoti hai.

Jhooth. Conditions ginना kaafi nahi — agar do conditions alag-alag points par hain toh matrix

M

singular ho sakta hai (

det M = 0

), jisse koi solution nahi ya infinitely many milte hain. Uniqueness sirf IVPs ke liye guaranteed hai, jahaan Wronskian nonzero hota hai.

y (a) = α

aur

y^{'} (a) = β

milke ek boundary value problem banate hain.

Jhooth. Dono conditions usi point

a

par hain, toh yeh ek IVP hai chahe ek derivative ho. "Boundary" ka matlab alag-alag locations hai,

y

y^{'}

use karna nahi.

"Initial" conditions

t = 0

par specify honi chahiye.

Jhooth. "Initial" ka matlab hai ek common point $x_{0}$ par saari conditions — woh point kuch bhi ho sakta hai (

x_{0} = 3

t = 7

, etc.). Defining feature hai ek location, number zero nahi.

Ek well-behaved ODE ka IVP solution dene mein fail ho sakta hai.

Jhooth (essentially). Mild continuity/Lipschitz conditions ke under, Picard–Lindelöf ek unique local solution guarantee karta hai — start point par nonzero Wronskian constants

c_{1}, c_{2}

ko uniquely pin kar deta hai.

Ek BVP ka hamesha solution hota hai kyunki tumne kaafi data diya.

Jhooth. Edges par data self-contradictory ho sakta hai (jaise

y^{''} + y = 0

y (0) = 0

y (π) = 1

force karta hai

0 = 1

). Conditions ki kaafi count system

M c = b

ki compatibility guarantee nahi karti.

Agar kisi BVP ke liye

det M = 0

ho, toh uske definitely infinitely many solutions hain.

Jhooth.

det M = 0

ka matlab hai ya toh koi solution nahi ya infinitely many — kaunsa wala depend karta hai right-hand side

b

par (target values

α, β

). Singular matrix ⇒ non-unique, lekin non-unique mein empty case bhi shamil hai.

Agar

det M \neq = 0

ho, toh BVP ka exactly ek solution hai.

Sach. Ek nonsingular

2 \times 2

system

M c = b

ka ek unique

(c_{1}, c_{2})

hota hai, isliye ek hi curve dono boundary conditions satisfy karti hai.

Do independent solutions ka Wronskian initial point par zero ho sakta hai.

Jhooth. Genuinely linearly independent solutions ke liye Wronskian interval mein har jagah nonzero hota hai (Abel's theorem: ya toh hamesha zero ya kabhi nahi). Woh non-vanishing hi exactly wajah hai ki

c_{1}, c_{2}

ke liye IVP system hamesha solvable kyun hota hai.

Ek IVP ke liye, relevant matrix

M

do points par evaluate hota hai.

Jhooth. IVP ke liye data ek point par hota hai, toh discriminating object Wronskian

y_{1} y_{2}^{'} - y_{2} y_{1}^{'}

hai us single point

x_{0}

par — ek matrix jiska rows

y

aur

y^{'}

use karte hain, do locations nahi.

Error dhundho

Har line mein ek galat claim hai; reveal galti ka naam batata hai aur use theek karta hai. Padhte waqt matrix $M$ picture (ek row per point) dimaag mein rakho — yahan ki zyaadatar errors mein ek row secretly collapse ho jaati hai.

y^{''} + y = 0

y (0) = 1

y (π) = 1

ek IVP hai jiska unique answer hai."

Error: yeh ek BVP hai (do distinct points

0

aur

π

). Basis

y_{1} = cos x, y_{2} = sin x

ke saath,

det M = cos 0 sin π - sin 0 cos π = 0

, toh yeh singular hai — bilkul bhi comfortably-unique IVP nahi.

"Kyunki

y^{''} + y = 0

y (0) = 0

y (π) = 0

mein

c_{1} = 0

force hota hai, toh sirf

y \equiv 0

solution hai."

Error:

y = c_{1} cos x + c_{2} sin x

ke saath, doosri condition

y (π) = - c_{1} = 0

wahi information repeat karti hai ki

c_{1} = 0

aur

c_{2}

ke baare mein kuch nahi kehti. Toh

y = c_{2} sin x

har

c_{2}

ke liye kaam karta hai — infinitely many solutions hain, sirf trivial nahi.

"Ek eigenvalue problem BVP se alag hota hai."

Error: ek eigenvalue problem ek BVP hi hai jahan nontrivial solutions pane ke liye

det M (λ) = 0

enforce kiya jaata hai. Woh

λ

jo matrix ko singular banate hain woh eigenvalues hain — yeh exactly Sturm-Liouville Theory hai.

"BVP solve karne ke liye main left boundary se IVP ki tarah aage badhunga."

Error: aage badhna sirf left-edge data use karta hai, lekin BVP ke liye tumhe

y^{'} (a)

pata nahi. Tumhe globally solve karna hoga taaki curve right-edge target par bhi utare — ek two-point matching, forward roll nahi.

det M = 0

buri arithmetic ki wajah se aaya; recompute karo toh nonzero hoga."

Error: khaas interval lengths ke liye (jaise

y^{''} + y = 0

ke saath

L = π

)

det M = 0

structurally exact hai, koi galti nahi. Yeh ek resonance/eigenvalue length mark karta hai — koi bhi recomputation ise nahi hataata.

y^{''} = 0

[0, L]

dono ends fixed ke saath, uniqueness fail ho sakti hai."

Error: basis

y_{1} = 1, y_{2} = x

ke saath,

det M = y_{1} (0) y_{2} (L) - y_{2} (0) y_{1} (L) = 1 \cdot L - 0 \cdot 1 = L \neq = 0

for

L > 0

, toh linear profile hamesha unique hai — steady heat conduction ka ek hi answer hota hai. (Agar rows ulte order mein likho toh

- L

milta hai; sirf sign flip hota hai, nonzero conclusion same rehti hai.)

"Kyun" wale sawaal

Kyun ek BVP ka koi solution nahi ho sakta jabki same-order IVP ka hamesha hota hai?

Ek IVP ek fully-known state (position + velocity) se aage evolve karta hai, toh nature baaki determine kar deti hai — Wronskian nonzero hota hai. Ek BVP demand karta hai ki curve ek door target hit kare, jo geometrically impossible ho sakta hai, isliye edge matrix

M

singular ho sakta hai aur system

M c = b

incompatible.

Kyun

det M = 0

eigenvalue behaviour signal karta hai?

Ek singular

M

matlab homogeneous BVP ka ek nonzero solution hai — ek nontrivial curve jo dono zero boundary conditions satisfy kare. Woh special parameters (interval length, ya

λ

) precisely eigenvalues hain, Sturm-Liouville Theory aur separation of variables ko feed karte hain.

Wronskian

det M

ka "IVP version" kyun hai?

Dono

M c = b

mein

c_{1}, c_{2}

fix karne wale coefficient matrix ka determinant hain. IVP ke liye dono rows

y

aur

y^{'}

se ek point par aati hain (woh hai Wronskian); BVP ke liye dono rows

y

se do points par aati hain (woh hai

M

). Same kaam, alag data layout.

Kyun conditions ka spatial separation trouble create karta hai, unki sankhya nahi?

Jab dono conditions ek hi point par hon, coefficient matrix ki rows "construction se independent" hoti hain (value vs slope). Jab alag-alag points par split hoon, dono rows accidentally align ho sakti hain (jaise

sin

0

aur

π

dono par zero hona), determinant collapse kar deta hai.

Kyun Heat Equation aur Wave Equation space mein naturally BVPs produce karte hain lekin time mein IVPs?

Space ke physical edges hote hain (rod ke ends, string ke ends) jahaan tum values fix karte ho — boundaries. Time ka ek starting instant hota hai jahaan tum state jaante ho aur use evolve hone dete ho — ek initial point. Isliye PDEs typically initial-boundary value problems hote hain.

Kyun "kitni conditions hain" check karna students ko solvability ke baare mein mislead karta hai?

Kyunki IVPs ke liye count kaafi hai, students overgeneralise karte hain. BVPs ke liye count sirf guarantee karta hai ki tum system

M c = b

likh sakte ho; yeh solvable aur unique hai ya nahi yeh poori tarah

det M

aur target vector

b

par depend karta hai.

Edge cases

Kya hota hai ek BVP ke saath jab do boundary points merge hone lagte hain,

b \to a

M

ki dono rows identical ho jaati hain (same point), toh

det M \to 0

— problem degenerate ho jaati hai. Limit mein tumhe

a

par

y

aur

y^{'}

chahiye hogi, matlab yeh ek IVP structure ki taraf collapse hoti hai.

Ek "BVP" jisme tumne accidentally dono conditions same point par rakhi — woh actually kya hai?

Yeh actually ek IVP hai (ya ek over/under-determined single-point problem). Ek point par genuinely distinct data ke saath yeh IVP ki tarah behave karta hai; agar wahi condition do baar di toh yeh underdetermined hai.

y^{''} = 0

[0, L]

with

y (0) = T_{0}

y (L) = T_{1}

: kya kuch special hota hai jab

L \to 0

det M = L \to 0

, toh uniqueness limit mein kho jaati hai — tum essentially ek hi jagah par do temperatures fix karne ki koshish kar rahe ho, jo contradictory hai jab tak

T_{0} = T_{1}

na ho. Nonzero

L

ise well-posed rakhta hai.

y^{''} + 4 y = 0

y (0) = 0

y (L) = 0

ke liye: kaun se

L

nontrivial solutions dete hain aur unke beech mein kya hota hai?

y = c_{1} cos 2 x + c_{2} sin 2 x

likhne par,

y (0) = c_{1} = 0

y = c_{2} sin 2 x

bacha; nontrivial solutions tab exist karte hain jab

sin 2 L = 0

, matlab

2 L = nπ

, toh

L = \frac{nπ}{2}

. Kisi bhi aur

L

ke liye,

det M \neq = 0

aur sirf trivial

y \equiv 0

solution hai.

Ek nonzero forcing term

f (x)

wala BVP lekin singular

M

— kya yeh hopeless hai?

Automatically nahi. Yeh solvable hai agar target vector

b

singular system ke saath compatible ho (ek solvability/orthogonality condition, Green's Functions se related); warna koi solution nahi. Singular

M

sirf guarantee hata deta hai.

Trivial solution

y \equiv 0

: kis type ke problems ke liye yeh hamesha available hai?

Kisi bhi homogeneous linear BVP ke liye zero boundary data ke saath (jaise

y (0) = y (L) = 0

y \equiv 0

hamesha satisfy karta hai. Interesting sawaal yeh hai ki kya koi nonzero solution bhi exist karta hai — tab hota hai jab

det M = 0

Zero-order "equation" (koi derivatives nahi), sirf

y = g (x)

ek boundary condition ke saath — IVP hai ya BVP?

Usual sense mein koi bhi nahi: fix karne ke liye koi arbitrary constants

c_{1}, c_{2}

nahi hain, toh side conditions redundant ya contradictory hain. IVP/BVP distinction tabhi bite karta hai jab general solution arbitrary constants carry kare.

Recall Upar ke har trap ka ek-sentence summary

Verdict (IVP vs BVP) yeh set karta hai ki side conditions kahaan rehti hain; solvability phir $M c = b$ mein coefficient matrix ke determinant se set hoti hai — IVPs ke liye hamesha-nonzero Wronskian, BVPs ke liye possibly-singular $M$ — aur singularity ka matlab non-uniqueness hai, jo "koi solution nahi" ya "infinitely many" mein split hoti hai target $b$ ke hisaab se.

Connections

Parent: IVP vs BVP
Wronskian and Linear Independence — IVP determinant.
Sturm-Liouville Theory — jahaan $det M = 0$ eigenvalues ban jaata hai.
Picard-Lindelöf Existence and Uniqueness Theorem — kyun IVPs reliably unique hote hain.
Green's Functions — forced singular BVPs ki solvability.