Determinant computation and meaning — Hinglish notes

1.1.9 · AI-ML › Linear Algebra Essentials

WHAT hai ek determinant?

Teen defining properties (baaki sab inhi se derive hoti hain):

KAISE compute karein — 2×2 se shuruaat

2×2 first principles se

Maano $A = (a c b d)$ . Columns vectors $u = (a, c)$ aur $v = (b, d)$ hain. Parallelogram ka area jo yeh span karte hain woh base×height hai. 2D mein cross-product magnitude use karke:

$signed area = a d - b c$

Kyun? $u, v$ se spanned signed area $u_{x} v_{y} - u_{y} v_{x} = a d - c b$ hai. Yahi volume-scaling factor IS hai, kyunki unit square (area 1) is parallelogram pe map hota hai.

3×3 cofactor (Laplace) expansion se

Hum ek row ke along expand karte hain, har entry ko us matrix ke determinant se weight karte hain jo uski row & column delete karne ke baad bachi hai (uska minor), aur saath mein ek checkerboard sign hota hai.

Sign $(- 1)^{i + j}$ kyun? Yeh alternating axiom se aata hai: ek entry ko isolate karne ke liye rows/columns ko rearrange karne ki cost sign swaps ki hoti hai. Checkerboard bas unhi swaps ko track karta hai: $+ - + - + - + - +$

Figure — Determinant computation and meaning

KAISE (fast, AI-ML ke liye) — row reduction

Cofactor expansion $O (n!)$ hai — badi matrices ke liye bekar hai. Iske bajaye Gaussian elimination use karke upper-triangular $U$ banao, aur track karo ki har operation $det$ ko kaise change karta hai:

Triangular = diagonal ka product kyun? Pehle column ke neeche bar-bar cofactor-expand karo: har baar sirf top-left entry survive karti hai, ek diagonal factor har step pe peel off hota hai. Yeh $O (n^{3})$ hai — real software (LU decomposition) isi tarah determinants compute karta hai.

Key derived properties

Consequences:

$det (A^{- 1}) = 1/ det A$ (kyunki $det (A A^{- 1}) = det I = 1$ ).
$det (A^{T}) = det A$ (rows↔columns same volume span karte hain).
$det (k A) = k^{n} det A$ $n$ dimensions mein (harek $n$ rows scale hoti hain).
$det A = \prod λ_{i}$ (eigenvalues ka product) — volume factor = per-axis stretches ka product.

Worked examples

Recall Feynman: 12-saal ke bacche ko samjhao

Ek stamp imagine karo jo ek picture ko squish aur stretch karta hai. Determinant ek single number hai jo batata hai ki picture kitni badi ya chhoti ho jaati hai. Agar number 5 hai, toh sab kuch 5× zyada area cover karta hai. Agar negative hai, toh stamp picture ko mirror ki tarah flip bhi karta hai ( $- 5$ phir bhi 5× bada karta hai, bas mirrored). Aur agar number 0 hai, toh stamp poori picture ko ek line pe flat kar deta hai — use kabhi un-squash nahi kar sakte, toh matrix ka "koi undo button nahi" hota (not invertible).

Flashcards

Determinant geometrically kya measure karta hai?

Linear map

x \mapsto A x

ka signed volume (area) scaling factor; sign orientation flip encode karta hai.

Ek square matrix

det

ke terms mein kab invertible hota hai?

Bilkul tabhi jab

det A \neq = 0

ho.

2×2 determinant ka formula?

(a c b d)

ke liye

a d - b c

det = 0

ka matlab not invertible kyun hai?

Map space ko lower dimension mein collapse karta hai (zero volume); yeh one-to-one nahi hai, toh koi inverse exist nahi karta.

Do rows swap karne ka

det

pe effect?

- 1

se multiply hota hai.

Ek row ka multiple doosri row mein add karne ka effect?

det

unchanged rehta hai (shear se volume preserve hota hai).

Ek row ko

k

se scale karne ka effect?

det

k

se multiply hota hai.

Ek triangular matrix ka

det

Diagonal entries ka product.

det (A B)

batao.

det (A) det (B)

det (A^{- 1})

kya hai?

1/ det A

n \times n

ke liye

det (k A)

kya hai?

k^{n} det A

det

ko eigenvalues se relate karo.

det A = \prod_{i} λ_{i}

Cofactor

C_{ij}

kya hota hai?

(- 1)^{i + j} M_{ij}

jahaan

M_{ij}

minor hai (row

i

, col

j

delete karne ke baad det).

Bade

n

ke liye cofactor expansion ke badle row-reduction kyun prefer karte hain?

Cofactor

O (n!)

hai; row reduction

O (n^{3})

hai.

Kya

det (A + B) = det A + det B

hota hai?

Nahi — determinant poori matrix mein linear nahi hota.

Connections

Linear transformations — determinant = map ka volume-scaling.
Matrix inverse — invertible $⟺ det \neq = 0$ ; Cramer's rule cofactors use karta hai.
Eigenvalues and eigenvectors — $det A = \prod λ_{i}$ , $det (A - λ I) = 0$ characteristic equation hai.
LU decomposition — triangular product ke through practical determinant.
Cross product and area — 2×2/3×3 determinants signed areas/volumes HI hain.
Change of variables (Jacobian) — $∣ det J ∣$ integration volume elements ko scale karta hai.
Singular matrices and rank — rank-deficient ⇒ $det = 0$ .