Foundations — Subspaces — four fundamental subspaces of a matrix

4.5.44 · D1 · Maths › Linear Algebra (Full) › Subspaces — four fundamental subspaces of a matrix

Yeh page kuch bhi assume nahi karta. Hum har woh symbol build karte hain jis par parent note depend karta hai, ek aisi order mein jahan har nayi idea sirf pehle se defined ideas use kare.

1. ${\mathbb{R}}^n$ — woh space jahan vectors rehte hain

Seedhi baat: chota $\mathbb{R}$ ka matlab hai "real numbers" (jaise $-2$, $0.5$, $\pi$). Superscript $n$ count karta hai ki har vector mein kitne slots hain.

Picture yeh hai: ${\mathbb{R}}^2$ woh flat page hai jis par tum draw karte ho — har point $(x,y)$ origin $(0,0)$ se us point tak ek arrow hai. ${\mathbb{R}}^3$ woh room hai jismein tum baithe ho. $n$ jyada hone par hum draw nahi kar sakte, lekin rules bilkul same rehte hain.

Topic ko iska kyun zaroori hai: parent kehta hai do subspaces "${\mathbb{R}}^n$ mein rehte hain" (inputs) aur do "${\mathbb{R}}^m$ mein rehte hain" (outputs). Yeh sentence tab tak meaningless hai jab tak tum yeh na jaano ki ${\mathbb{R}}^n$ sirf "$n$-slot vectors ka space" hai.

2. Ek vector, aur origin $0$

Picture yeh hai: origin woh single anchor point $(0,0)$ hai jahan se har arrow shuru hota hai. Bold $0$ woh anchor khud hai, ek (degenerate) arrow ki tarah dekha gaya.

Topic ko iska kyun zaroori hai: subspace ka pehla rule hai "$0$ ko contain karo." Ek subspace ek "flat cheez origin ke through" hai. $0$ ke bina tum rule 1 bhi state nahi kar sakte.

3. Do vector operations: add aur scale

Picture yeh hai: addition "tip-to-tail" hai — pehle arrow ke saath chalo, phir doosre ke saath; jahan tum khatam hote ho woh sum hai. $c$ se scale karna arrow ko stretch karta hai (agar $|c|>1$), shrink karta hai (agar $|c|<1$), ya origin ke around flip karta hai (agar $c<0$).

Topic ko iska kyun zaroori hai: subspace rules 2 aur 3 exactly yahi do operations hain. "Addition/scaling ke under closed" ka matlab hai: yeh operations karo aur tum kabhi set se bahar nahi jaate. Har subspace sirf adding aur scaling se bana hai.

4. Linear combination aur $\text{span}$

Picture yeh hai: ek nonzero vector ka span us se guzarne wali line hai. Do non-parallel vectors ka span woh poora plane hai jo woh sweep karte hain. Span matlab hai "har jagah jahan tum mixing se pahunch sakte ho."

Topic ko iska kyun zaroori hai: column space defined hai columns ke $\text{span}$ ke roop mein, aur row space rows ke $\text{span}$ ke roop mein. Span woh word hai jo "kuch vectors" ko "ek poora subspace" mein badalta hai.

5. Dot product $u\cdot v$ aur "perpendicular"

Yeh tool kyun, aur koi nahi? Topic ka sabse gehra claim yeh hai ki subspaces perpendicular pairs mein aate hain. "Perpendicular" numbers mein kehne ke liye tumhe ek aisa gadget chahiye jo exactly right angle par $0$ return kare — dot product woh gadget hai. Koi angle-measuring nahi, koi protractor nahi: ek single sum batata hai ki kya do arrows $90^{\circ }$ par milte hain.

Picture yeh hai: ek arrow ko slide karo taaki dono origin se shuru hon. Agar woh right angle banaate hain, unka dot product $0$ hai. Agar woh same direction mein jhukein toh positive hai; opposite direction mein, negative.

Topic ko iska kyun zaroori hai: "Section 4: Orthogonality" poora dot-product language mein likha hai. "$x$ har row ke orthogonal hai" ka matlab hai har row ko $x$ se dot karo aur $0$ milta hai.

6. Matrix $A$ aur product $Ax$

Seedhi baat: list $x$ feed karo; $x$ ki har entry kehti hai "us column ka kitna use karna hai." Scaled columns add karo aur tumhe output milta hai.

Picture yeh hai: $A$ ko ek machine ki tarah socho. Andar jaata hai ek $n$-slot vector, bahar aata hai ek $m$-slot vector. $A$ ke columns machine ke "building blocks" hain; $x$ yeh recipe hai ki kitna kitna mix karna hai.

Topic ko iska kyun zaroori hai: sab kuch $Ax$ hai.

• Column space = saare outputs $Ax$ (reachable juices).
• Null space = inputs jahan $Ax=0$ (recipes jo kuch nahi banaate). Jab ek baar tum maano ki "$Ax$ = columns ka mix hai," definitions khud padhi jaati hain.

7. Transpose $A^T$

Picture yeh hai: matrix ko ek side par tilt karo. Row 1 column 1 ban jaata hai.

Topic ko iska kyun zaroori hai: row space defined hai $C(A^T)$ ke roop mein — "flipped matrix ka column space" — kyunki $A$ ki rows hain hi $A^T$ ke columns. Similarly left null space $N(A^T)$ hai. Transpose woh trick hai jo humein rows ke liye "column space" aur "null space" language reuse karne deta hai.

8. Rank $r$, pivots, aur dimension

Picture yeh hai: rank count karta hai ki kitne arrows actually nayi directions mein point karte hain. Agar column 2 sirf column 1 ka double hai, toh woh koi nayi direction add nahi karta — woh rank nahi badhata.

Topic ko iska kyun zaroori hai: single number $r$ charon dimensions control karta hai: $\dim C(A)=r,\dim C(A^T)=r,\dim N(A)=n-r,\dim N(A^T)=m-r.$ Machinery ke liye dekho Rank of a Matrix aur Rank–Nullity Theorem; yahan hum sirf note karte hain ki "rank" ke bina parent ke Section 3 ka koi vocabulary nahi hai.

9. Direct sum $\oplus$ aur orthogonal complement

Picture yeh hai: ${\mathbb{R}}^3$ mein, origin ke through ek plane aur origin ke through us plane ke perpendicular line orthogonal complements hain — $0$ ke alawa koi shared arrow nahi, aur koi bhi vector = (plane par uski shadow) + (line ke along uska part).

Topic ko iska kyun zaroori hai: Section 4 ka punchline hai $N(A)\oplus C(A^T)={\mathbb{R}}^n$ with $N(A)\perp C(A^T)$. Woh single symbol $\oplus$ package karta hai "woh koi overlap aur koi gap ke bina ek saath fit hote hain."

Prerequisite map

Equipment checklist

Apne aap ko test karo — har line prompt ::: answer hai.

Connections

• Subspaces — four fundamental subspaces of a matrix — woh parent jiske liye yeh page tumhe prepare karta hai.
• Rank of a Matrix — Section 8 mein built number $r$.
• Rank–Nullity Theorem — jahan se $\dim N(A)=n-r$ aata hai.
• Orthogonal Complements — Section 9 ko gehraata hai.
• RREF and Pivots — pivots (Section 8) kaise dhunde jaate hain.
• Solving Ax=b — Section 6 se $Ax$ use karta hai.
• Least Squares & Projections — dot product aur $\oplus$ use karta hai.