Visual walkthrough — Types of matrices — row, column, square, diagonal, identity, zero, symmetric, skew-symmetric

2.6.2 · D2 · Maths › Matrices & Determinants — Introduction › Types of matrices — row, column, square, diagonal, identity,

Parent note ne humein characters ka cast diya tha — row, column, square, diagonal, identity, zero, symmetric, skew-symmetric. Yeh page unhe do ko jodne wala sabse sundar fact prove karta hai:

Koi bhi square matrix ek symmetric part aur ek skew-symmetric part mein toodi ja sakti hai — aur sirf ek hi tarike se.

Hum ise dum se shuru karte hain. Agar aapne kabhi matrix, transpose, ya "symmetric" shabd nahi dekha, Step 1 se shuru karo aur kuch bhi bahar se mat dekho.

---

Step 1 — Matrix kya hoti hai, aur ise "flip" karne ka matlab kya hai

KYA. Ek matrix bas numbers ka ek rectangular grid hota hai. Hum har number ko kaun si row aur kaun sa column mein hai, is hisaab se label karte hain. Row $i$, column $j$ mein jo number hota hai, use $a_{ij}$ likhte hain.

KYUN. Grids ko flip ya add karne se pehle, humein har cell ka naam chahiye. Address $a_{ij}$ wahi naam hai — pehla index row hai (upar se neeche ginne par), doosra column hai (ek taraf se doosri taraf ginne par).

PICTURE. Neeche diye gaye $2\times 2$ board ko dekho. Row 1, column 2 mein white cell $a_{12}$ hai; diagonal ke paar uska mirror (row 2, column 1) $a_{21}$ hai. Ye do "partner" cells hi aage ki saari cheez ka dil hain.

---

Step 2 — Transpose: grid ko uski diagonal ke across reflect karna

KYA. $A$ ka transpose, jo $A^T$ likha jata hai, woh grid hai jo aapko har cell $a_{ij}$ ko $a_{ji}$ se swap karne par milta hai. Seedhe shabdon mein: rows columns ban jaati hain.

$(A^T{)}_{ij}=\underset{\text{cell from the swapped address}}{\underbrace{a_{ji}}}$

Equation ko term by term padho: address $(i,j)$ par jo cell $A^T$ ke andar hai, woh original $A$ ke andar address $(j,i)$ se li gayi hai. Dono indices apni jagah badal lete hain.

KYUN. Humein ek aisa machine chahiye jo matrix ko "mirror" kare taaki hum pooch sakein "kya mirror original ke barabar hai?" Wahi ek sawaal symmetry define karta hai. Mirror line principal diagonal hai — diagonal cells ($i=j$) kabhi nahi hilte, kyunki $a_{ii}\to a_{ii}$.

PICTURE. Blue arrow dikhata hai $a_{12}$ diagonal ke paar jaake naya $a_{21}$ ban raha hai, aur pink arrow ulta. Diagonal cells apni jagah rahte hain (yellow).

---

Step 3 — Do special mirrors: symmetric aur skew-symmetric

KYA. Mirror ka sawaal poochho toh bilkul do "perfect" jawab hote hain:

• $A^T=A$ → mirror identical hai. $A$ ko symmetric kaho. Cell-wise: $a_{ij}=a_{ji}$.
• $A^T=-A$ → mirror bilkul ulta negative hai. $A$ ko skew-symmetric kaho. Cell-wise: $a_{ij}=-a_{ji}$.

KYUN. Ye do building blocks hain jinmein hum har matrix ko split karenge. Symmetric = partners barabar hain; skew = partners opposite hain.

Skew ke liye ek forced consequence hai. $i=j$ rakho (ek diagonal cell): $a_{ii}=-a_{ii}⟹2a_{ii}=0⟹\underset{\text{every diagonal cell is zero}}{\underbrace{a_{ii}=0}}$ Term by term: ek diagonal cell apne aap ke negative ke barabar hona chahiye; aur sirf $0$ hi aisa number hai jo apne aap ke negative ke barabar ho. Isliye skew-symmetric matrix ka diagonal hamesha zero hota hai.

PICTURE. Left board: barabar partners (dono $+3$). Right board: opposite partners ($+3$ aur $-3$) aur forced-zero diagonal.

---

Step 4 — Trick: kisi bhi matrix se ek symmetric matrix banana

KYA. Koi bhi square $A$ lo. Banao: $S=\frac{1}{2}(\underset{\text{original}}{\underbrace{A}}+\underset{\text{its mirror}}{\underbrace{A^T}}).$

Claim: $S$ symmetric hai. Use transpose karke check karo, do facts use karte hue — transpose linear hai ($(X+Y{)}^T=X^T+Y^T$) aur self-undoing hai ($(A^T{)}^T=A$): $S^T=\frac{1}{2}(A^T+(A^T{)}^T)=\frac{1}{2}(A^T+A)=S.$ Har symbol: humne sum ko transpose kiya, use split kiya, doosre term par double-transpose undo kiya, aur wapas $S$ par aa gaye. Kyunki $S^T=S$, yeh symmetric hai. ✓

KYUN. Ek matrix ko uske mirror se add karna force karta hai ki partners barabar ho jaayein — har off-diagonal pair dono taraf $\frac{1}{2}(a_{ij}+a_{ji})$ ban jaati hai. $\frac{1}{2}$ sirf bookkeeping hai taaki kuch double na ho.

PICTURE. Partners $a_{12}$ aur $a_{21}$ ko dekho jo diagonal ke dono taraf same value mein average ho rahe hain.

---

Step 5 — Usi $A$ se ek skew-symmetric matrix banana

KYA. Ab mirror ko subtract karo: $K=\frac{1}{2}(\underset{\text{original}}{\underbrace{A}}-\underset{\text{its mirror}}{\underbrace{A^T}}).$

Claim: $K$ skew-symmetric hai. Use transpose karo: $K^T=\frac{1}{2}(A^T-(A^T{)}^T)=\frac{1}{2}(A^T-A)=-\frac{1}{2}(A-A^T)=-K.$ Term by term: difference ko transpose karo, double-transpose undo karo, minus sign bahar nikalo, andar $K$ pehchano. Kyunki $K^T=-K$, yeh skew hai. ✓ Aur iska diagonal automatically zero hai, kyunki $\frac{1}{2}(a_{ii}-a_{ii})=0$.

KYUN. Ek matrix ko uske mirror se subtract karna force karta hai ki partners opposite ho jaayein: har pair ek taraf $\frac{1}{2}(a_{ij}-a_{ji})$ ban jaati hai aur doosri taraf $\frac{1}{2}(a_{ji}-a_{ij})$ (uska negative).

PICTURE. Partners $+d$ aur $-d$ mein split hote hain; diagonal $0$ par collapse ho jaata hai.

---

Step 6 — Dono pieces wapas $A$ dete hain (decomposition)

KYA. Dono pieces add karo: $S+K=\frac{1}{2}(A+A^T)+\frac{1}{2}(A-A^T)=\frac{1}{2}(A+A^T+A-A^T)=\frac{2A}{2}=A.$ Har symbol apni jagah earn karta hai: $S$ ka $+A^T$ aur $K$ ka $-A^T$ cancel ho jaate hain, bacha $A+A$, jise halves ek single $A$ mein badalte hain.

KYUN. Yahi to poori baat hai: koi bhi square matrix ek symmetric part aur ek skew-symmetric part ka sum hai. $\boxed{A=\underset{\text{symmetric }S}{\underbrace{\frac{1}{2}(A+A^T)}}+\underset{\text{skew }K}{\underbrace{\frac{1}{2}(A-A^T)}}}$

PICTURE. $S$ board aur $K$ board stack hokar original $A$ board banaate hain.

---

Step 7 — Edge cases: kya yeh kabhi toot sakta hai?

Hum koi bhi scenario dikhaye bina kabhi nahi chhod sakte.

• $A$ pehle se symmetric ($A^T=A$): toh $K=\frac{1}{2}(A-A)=O$, zero matrix. Skew part gayab ho jaata hai — jaisa hona chahiye, fix karne ke liye kuch hai hi nahi.
• $A$ pehle se skew ($A^T=-A$): toh $S=\frac{1}{2}(A+(-A))=O$. Symmetric part gayab ho jaata hai.
• $A$ diagonal (ek special symmetric matrix): phir se $K=O$; diagonal matrices apna khud ka symmetric part hoti hain.
• $A=O$ (zero matrix): dono $S=O$ aur $K=O$. Trivially sach.
• $A$ square nahi hai: $A^T$ ka alag shape hai, isliye $A+A^T$ ban hi nahi sakta. Decomposition ke liye square matrix chahiye — isliye Step 1 ne square par zor diya tha.

PICTURE. Char collapse-cases side by side, aur "shape mismatch" jo non-square inputs ko forbid karta hai.

---

Ek-picture summary

Ek diagram, poori derivation: $A$ ko mirror karo $A^T$ pane ke liye; unhe average karo symmetric half $S$ ke liye; half difference lo skew half $K$ ke liye; dono halves add hokar $A$ dete hain.

Recall Feynman retelling — seedhe shabdon mein bolo

Ek numbers ka grid aur uski mirror image socho jo top-left-to-bottom-right diagonal ke across reflect ho. Agar aap grid ko uske mirror ke saath average karo, matching partners barabar ho jaate hain — woh hai symmetric grid $S$. Agar aap instead half the difference lo, partners exact opposites ban jaate hain aur diagonal force hokar zero ho jaata hai — woh hai skew-symmetric grid $K$. Clever baat yeh hai: jab aap $S$ aur $K$ add karte ho toh mirror-copies cancel ho jaate hain aur aapko original grid milta hai, bilkul theek se. Toh har square grid secretly ek symmetric part aur ek skew part hai, aur ise karne ka sirf ek hi tarika hai. Agar grid pehle se symmetric tha, toh skew part sab zeros hai; agar pehle se skew tha, toh symmetric part sab zeros hai; aur agar grid square nahi hai, toh aap use uske mirror ke saath align hi nahi kar sakte, isliye trick apply nahi hoti.

Recall

$A$ ke symmetric part ka formula ::: $S=\frac{1}{2}(A+A^T)$ Skew-symmetric part ka formula ::: $K=\frac{1}{2}(A-A^T)$ Skew matrix ka diagonal hamesha zero kyun hota hai ::: $a_{ii}=-a_{ii}\Rightarrow a_{ii}=0$ Transpose cell $a_{ij}$ ke saath kya karta hai ::: use address $(j,i)$ par bhejta hai, yaani $(A^T{)}_{ij}=a_{ji}$ Skew part kab gayab ho jaata hai ::: jab $A$ pehle se symmetric ho ($A^T=A$)