Matrix multiplication — conditions, process, non-commutativity: Foundations

2.6.5 · D1 · Maths › Matrices & Determinants — Introduction › Matrix multiplication — conditions, process, non-commutativi

Isse pehle ki aap parent page Matrix Multiplication par bharosa kar sako, aapko uska har squiggle bina rukke padhna aana chahiye. Ye page har ek cheez ko zero se build karta hai, ek aise order mein jahan har idea sirf usse pehle waale ideas par tikaa ho.

1. Ek number (scalar) — sabse chhota brick

Picture: number line par ek dot, ya ek "kitna hai" ka label. Abhi tak koi direction nahi hai — scalar sirf kitna batata hai, kidhar nahi.

Topic ko ye kyun chahiye: matrix ke andar har entry ek scalar hai, aur jo final answers hum compute karte hain (jaise $58$ ) wo bhi scalars hain. Agar aap plain numbers ko add aur multiply kar sakte ho, toh aapke paas pehle se hi neeche ki har cheez ke atoms hain.

2. Ek vector — components waala ek arrow

Figure — Matrix multiplication — conditions, process, non-commutativity

Figure dekho. Blue arrow $x$ hai. Horizontal axis par uski shadow ki length $3$ hai (pehla component); vertical axis par uski shadow ki length $2$ hai (doosra component). Arrow hi number ka pair hai — na zyada, na kam.

Topic ko ye kyun chahiye: matrices vectors par act karti hain. Poori story " $B$ transforms $x$ into $B x$ " tab tak kuch matlab nahi rakhti jab tak aap $x$ ko ek arrow ke roop mein picture nahi kar sakte.

3. Dot product — multiplication ki heartbeat

Wo single computation jo parent page baar baar repeat karta hai wo hai dot product. Ise master kar lo aur matrix multiplication copy-paste ban jaata hai.

"Position by position, phir add" kyun? Kyunki har pair $a_{k} b_{k}$ ek chhota sa sawaal ka jawaab deta hai — "ingredient $k$ kitna hai, price $k$ par?" — aur sum har ingredient ka contribution ek single total mein collect karta hai. Ye exactly wo smoothie story hai jo parent ke Feynman box mein hai.

Figure dekho. Top row (orange) aur side column (blue) line up hain. Har matched pair multiply hota hai (green links), aur saare products ek green box mein flow karte hain — answer. Notice karo ki row aur column same length ke hone chahiye, warna kuch entries ka koi partner nahi hoga. Ye length-matching rule §6 mein dimension rule ka seed hai.

4. Ek matrix — scalars ki ek grid, rows aur columns ke roop mein padhna

Ise picture karne ke do tarike, dono useful:

Rows ki stack ke roop mein — har row ek row vector hai, dot product ki left side banने के लिए ready.
Columns ki line ke roop mein — har column ek column vector hai, dot product ki right side banने के लिए ready.

Parent page $A$ ki rows aur $B$ ke columns use karta hai, toh dono pictures live rakho.

5. Subscripts aur dimensions — address system

Upar wale $A$ ke liye: $A_{12} = 2$ (row 1, column 2) aur $A_{23} = 6$ (row 2, column 3).

Topic ko ye kyun chahiye: parent ka core formula $C_{ij} = \sum_{k} A_{ik} B_{k j}$ (jahan $C$ §4 ka product hai) poori tarah subscript navigation hai. Agar $A_{ij}$ second nature ban jaaye, toh wo formula ek sentence ki tarah padhta hai.

6. Compatibility condition — $A$ aur $B$ multiply bhi kar sakte hain kya?

§3 ka dot-product length rule seedha us rule mein grow karta hai jo poori matrices ko govern karta hai.

Exactly ye rule kyun? Har entry $C_{ij}$ dot product hai $A$ ki ek row (length $n$ , kyunki $A$ mein $n$ columns hain) ka $B$ ke ek column (length $n$ , kyunki $B$ mein $n$ rows hain) ke saath. §3 se, dot product ko equal lengths chahiye — toh row length $n$ aur column length same $n$ hona chahiye. Bas yehi condition hai, dot product ke partner-matching se inherited.

7. Sigma notation $\sum$ — "ek pattern ko jodo"

Parent ka compact formula $\sum$ use karta hai. Ye log ko sirf isliye darata hai kyunki ye unfamiliar hai; iska matlab kuch bhi nahi sirf "inhe jodo."

Concretely, $n = 3$ ke liye: $\sum_{k = 1}^{3} A_{ik} B_{k j} = A_{i 1} B_{1 j} + A_{i 2} B_{2 j} + A_{i 3} B_{3 j}$

Isko §3 se compare karo: ye literally dot product hai $A$ ki row $i$ aur $B$ ke column $j$ ka, aur ye §4 ke product ki entry $C_{ij}$ produce karta hai. Toh $\sum$ koi naya machinery nahi hai — ye us adding-up step ka shorthand hai jo aapne pehle se haath se kiya hua hai.

8. Composition idea — ye sab exist kyun karta hai

Parent ka sabse gehra claim ye hai ki $A B$ ka matlab hai "pehle $B$ karo, phir $A$ ." Ye composition ka concept hai, Linear transformations se liya gaya.

Figure dekho. Ek arrow box $B$ mein enter karta hai, badla hua bahar aata hai, phir box $A$ mein enter karta hai aur phir badla hua bahar aata hai. $A B$ label wala dashed shortcut box dono changes ek saath karta hai. Boxes ko right to left padho — isliye $B$ (jo right par likha hai) pehle act karta hai.

Prerequisite map

Neeche di gayi picture dikhati hai kaunsa idea kise feed karta hai. Agar aapka reader ka viewer diagrams render nahi karta, toh yehi baat words mein: scalar atom hai; ye vector aur matrix dono build karta hai. Vector aur matrix mil kar dot product dete hain; matrix subscripts & dimensions bhi deta hai, jo compatibility condition mein grow karte hain aur, dot product ke saath, sigma notation mein. Alag se, vector aur matrix composition dete hain. Phir sigma notation aur composition dono target topic, Matrix Multiplication, ko feed karte hain.

Do streams notice karo jo topic ko feed karte hain: mechanical stream (dot product → compatibility → sigma) batata hai compute kaise karo, aur conceptual stream (composition) batata hai answer ka matlab kya hai.

Equipment checklist

Self-test: right side cover karo aur zyaano se jawab do.

Main

x = [32]

ko ek arrow ke roop mein padh sakta hoon

Haan — 3 right jao, 2 upar; components uske axis-shadows hain.

Row vector aur column vector mein kya fark hai

Row flat likhi jaati hai

[32]

; column vertically stack hoti hai; parent

A

ki rows ko

B

ke columns se multiply karta hai.

Dot product kaise lena hai

Do lists ko position by position multiply karo, phir saare products ko ek scalar mein jodo; lists same length ki honi chahiye.

A_{ij}

kahaan point karta hai

Row

i

, column

j

ka scalar — row pehle, column doosre, dono 1 se start hote hue.

A_{2 \times 3}

mujhe kya batata hai

A

mein 2 rows aur 3 columns hain.

C = A B

mein

A

B

, aur

C

kaun hain

A

left factor hai (

m \times n

B

right factor (

n \times p

) jo pehle act karta hai, aur

C

product (

m \times p

) jo ek entry ek time par build hota hai.

A

kab

B

ko multiply kar sakta hai

Jab

A

m \times n

ho aur

B

n \times p

ho — inner

n

's match karte hain; product

m \times p

hai; warna undefined.

\sum_{k = 1}^{n} A_{ik} B_{k j}

ko kaise unpack karo

k

1

n

tak chalao, har product

A_{ik} B_{k j}

banao, aur jodo — ye exactly dot product hai jo

C_{ij}

deta hai.

A B

ka matlab "

B

pehle, phir

A

" kyun hai

Kyunki

A B

composition hai —

A (B x)

— aur hum machines ko right to left padhte hain.

A B \neq = B A

generally kyun

Order swap karna woh transformation swap karta hai jo pehle chalta hai; alag pipeline alag result deti hai.

Connections

Matrix Multiplication — wo parent jiske liye ye page aapko prepare karta hai
Matrix addition and scalar multiplication — wo gentler operations jo grid picture share karte hain
Linear transformations — "matrix as machine" aur composition ideas ka source
Identity matrix — wo do-nothing machine jo aap aage miloge
System of linear equations — jahan matrix-times-vector pehli baar pay off hota hai