A matrix is a rectangular arrangement of numbers (or functions) organized into rows and columns, enclosed in brackets. That is the formal definition — but here is the intuition that actually matters for exam thinking.
Think of a matrix as a transformation machine. When you multiply a matrix by a column vector, you are transforming that vector — rotating it, scaling it, or shearing it in some geometric space. This is why matrices are central to linear algebra, and why every entrance exam at the 12th-pass level tests them heavily. The NDA paper is no exception.
A matrix with m rows and n columns is called an m × n matrix. Its size is called the order. The element sitting at the intersection of the ith row and jth column is denoted aᵢⱼ. So if someone says "the (2,3) element is 7", they mean row 2, column 3 contains 7.
Here is the analogy that works in the classroom: treat a matrix like a grid of spreadsheet cells. Addition combines two same-size spreadsheets cell by cell. Multiplication is more like a lookup-and-combine operation — far more powerful, and also far less obvious.
Two things that make matrices different from ordinary numbers:
AB ≠ BA in general. This surprises students who are used to 2 × 3 = 3 × 2.For NDA specifically, matrices questions almost always reduce to one of three things: computing |AB| or |Aᵀ| using determinant product rules, performing a direct 2 × 2 or 3 × 3 multiplication, or identifying the type of matrix from its structure. The concept section below covers the types. The Deep Dive covers the operations and properties that translate directly into marks.
| Name | Defining Property |
|---|---|
| Row matrix | Only 1 row (1 × n) |
| Column matrix | Only 1 column (m × 1) |
| Square matrix | m = n |
| Zero (Null) matrix | All entries are 0 |
| Identity matrix (I) | Square; aᵢᵢ = 1, all other aᵢⱼ = 0 |
| Diagonal matrix | Square; aᵢⱼ = 0 whenever i ≠ j |
| Scalar matrix | Diagonal with all diagonal entries equal |
| Upper triangular | aᵢⱼ = 0 for all i > j |
| Lower triangular | aᵢⱼ = 0 for all i < j |
| Symmetric | A = Aᵀ, i.e., aᵢⱼ = aⱼᵢ |
| Skew-symmetric | A = -Aᵀ, i.e., aᵢⱼ = -aⱼᵢ (diagonal entries must be 0) |
| Orthogonal | AAᵀ = I, equivalently Aᵀ = A⁻¹ |
| Idempotent | A² = A |
| Involutory | A² = I |
| Nilpotent | Aⁿ = O for some positive integer n |
NDA questions frequently drop one of these terms in an MCQ stem and ask for a conclusion. Recognise the definition on sight — no derivation needed.
Addition and Subtraction: Only defined when matrices have the same order. Add element by element. (A + B)ᵢⱼ = aᵢⱼ + bᵢⱼ. Commutative and associative.
Scalar Multiplication: (kA)ᵢⱼ = k · aᵢⱼ. Every element gets multiplied.
Matrix Multiplication: AB is defined only when the number of columns of A equals the number of rows of B. If A is m × p and B is p × n, then AB is m × n.
The formula: (AB)ᵢⱼ = Σₖ aᵢₖ · bₖⱼ
For 2 × 2 matrices:
Repeat the row-into-column dot product mentally. With practice, a 2 × 2 product takes under 30 seconds.
Aᵀ is obtained by swapping rows and columns: (Aᵀ)ᵢⱼ = aⱼᵢ. A 3 × 2 matrix becomes a 2 × 3 after transposing.
Key properties:
(Aᵀ)ᵀ = A(A + B)ᵀ = Aᵀ + Bᵀ(kA)ᵀ = kAᵀ(AB)ᵀ = BᵀAᵀ — note the reversal of order. This is a classic trap.|Aᵀ| = |A| — the determinant does not change under transposition.These four properties are the backbone of every NDA "find the value of |...|" question:
|AB| = |A| · |B||Aᵀ| = |A||kA| = kⁿ|A| for an n × n matrix (scalar comes out raised to the order, not just once)|A⁻¹| = 1/|A|Look — these four lines answer the majority of NDA determinant-of-product questions in under 20 seconds each. Memorise them as a block.
Cofactor Cᵢⱼ of element aᵢⱼ: delete row i and column j, compute the determinant of the remaining matrix (the minor Mᵢⱼ), then apply sign: Cᵢⱼ = (-1)^(i+j) · Mᵢⱼ.
Adjoint (adj A): Transpose of the cofactor matrix. That is, (adj A)ᵢⱼ = Cⱼᵢ.
Inverse: A⁻¹ = (adj A) / |A|, valid only when |A| ≠ 0 (i.e., A is non-singular).
Key results:
A · (adj A) = (adj A) · A = |A| · I|adj A| = |A|^(n-1) for an n × n matrixadj(AB) = (adj B)(adj A) — again, order reversesFor a 2 × 2 matrix, the adjoint has a shortcut:
Swap the diagonal, negate the off-diagonal. This is the fastest hand-calculation path to the inverse of a 2 × 2.
Elementary row operations (EROs): swap two rows, multiply a row by a non-zero scalar, add a scalar multiple of one row to another. These do not change the rank of a matrix. The rank is the number of non-zero rows in row-echelon form — but for NDA purposes, rank questions are less common than the product and inverse questions above.
Whenever you see a formula involving a product of matrices — transpose, adjoint, or inverse — the order reverses:
(AB)ᵀ = BᵀAᵀ, (AB)⁻¹ = B⁻¹A⁻¹, adj(AB) = (adj B)(adj A).
One mental hook: think of putting on shoes and socks — you take them off in reverse order. Socks go on first (A first), shoes go on second (B second). To undo: remove shoes (B) first, then socks (A). Standard method of deriving this each time: 60s. Pattern recall: 3s.
Students routinely write |kA| = k|A|. Wrong. For an n × n matrix, each row gets multiplied by k, so |kA| = kⁿ|A|.
Micro-example: A is 3 × 3, k = 2. |2A| = 2³|A| = 8|A|. If |A| = 5, then |2A| = 40, not 10.
Standard error rate on this in mock tests is high. The fix: always write n as a superscript before substituting. Step count: 1 step vs. the common 1-step-but-wrong approach. Gets the answer right 100% vs. roughly 40% without the rule.
For A = [[a,b],[c,d]]:
|A| = ad - bc.adj A: swap a and d, negate b and c.adj A by |A|.No cofactor expansion, no 3×3 determinant. Standard cofactor method for 2×2: 6-8 steps. This method: 4 steps. Time: standard 50s vs. shortcut 20s. Works for any invertible 2×2.
|Aᵀ| = |A| always. So |AAᵀ| = |A| · |Aᵀ| = |A| · |A| = |A|².
When the question gives you |A| and asks for |AAᵀ|, square the given value immediately. If |A| = -2, then |AAᵀ| = (-2)² = 4. This is a one-second read once you see the pattern. Standard expansion approach: 30-40s. Pattern recall: 5s.
Check these in order and stop as soon as one fits:
aᵢⱼ = aⱼᵢ → Symmetric.aᵢⱼ = -aⱼᵢ (so diagonal = 0) → Skew-symmetric.This elimination chain resolves 90% of "what type is this matrix" MCQs in under 10 seconds without computing anything. Standard approach of checking all definitions: 30-40s.
When you see a matrix question in the exam hall, run this decision tree:
Step 1 — What is the question asking?
|something|" → go to determinant property rules (|AB| = |A||B|, |Aᵀ| = |A|, |kA| = kⁿ|A|). Do not expand unless forced.A⁻¹" → check if it is 2×2. If yes, use the swap-and-negate adjoint shortcut. If 3×3, use cofactors.AB" → check dimensions first, then row-into-column multiplication.Step 2 — Is there a given condition you haven't used?
If the question states A is symmetric, A is orthogonal, or A² = I, that condition almost always unlocks a one-line solution. Don't ignore it.
Step 3 — Does the answer need to be a number or a matrix?
If it is a number (like |AB|), verify sign. Determinants of real matrices can be negative; |A|² is always non-negative.
Step 4 — Check order consistency.
Before multiplying, write (m × p)(p × n) = m × n. If sizes don't match, the product is undefined — and that itself might be the answer.
Why this question: This is a high-frequency NDA pattern — |AAᵀ| combines the transpose determinant property with the product rule. If you know the two-line rule, this is a guaranteed mark.
Solving path: |Aᵀ| = |A| = -2. Apply |AAᵀ| = |A| · |Aᵀ| = (-2)(-2) = 4. Notice the answer is positive — squaring removes the negative. Do not write -4 just because |A| is negative.
Why this question: The product rule |AB| = |A||B| is tested directly. Students who haven't internalized this rule waste time computing the full product matrix and then its determinant — an approach that is both slower and error-prone for larger matrices.
Solving path: |AB| = |A| × |B| = 3 × 2 = 6. No matrix computation needed. If this took you more than 5 seconds, drill the four determinant product rules until they are automatic.
Why this question: Direct 2 × 2 matrix multiplication. Tests whether you can read the product rule without errors under time pressure. The diagonal structure of A makes this slightly faster than a general 2 × 2 case.
Solving path: A = [[2,0],[0,3]], B = [[1,1],[1,1]].
Row 1 of AB: (2×1 + 0×1, 2×1 + 0×1) = (2, 2).
Row 2 of AB: (0×1 + 3×1, 0×1 + 3×1) = (3, 3).
Result: [[2,2],[3,3]].
Shortcut observation: since A is diagonal, multiplying A by B simply scales row 1 of B by 2 and row 2 of B by 3. A diagonal matrix diag(d₁, d₂) scales rows — this halves your mental calculation time whenever you spot a diagonal matrix.
Commutativity assumption. Writing AB = BA is wrong in general. Before assuming commutativity in any step, check whether the matrices are both diagonal, or both scalar, or one is the identity — only then does AB = BA hold reliably.
|kA| = k|A| for any size matrix. This is the single most common error in NDA mock tests. The correct formula is |kA| = kⁿ|A| for n × n. A 3 × 3 matrix with k = 2 multiplies the determinant by 8, not 2.
(AB)ᵀ = AᵀBᵀ instead of BᵀAᵀ. The order reversal is mandatory. A quick check: if A is 2 × 3 and B is 3 × 4, then AB is 2 × 4, and (AB)ᵀ is 4 × 2. Now Bᵀ is 4 × 3 and Aᵀ is 3 × 2, so BᵀAᵀ is 4 × 2. Correct. The other order AᵀBᵀ is 3 × 2 times 4 × 3 — dimensions don't even match.
Adjoint = Cofactor matrix (not its transpose). The cofactor matrix and the adjoint are different. adj A = (cofactor matrix)ᵀ. If you skip the transpose step, every element of your inverse will be wrong.
Inverse of a product: (AB)⁻¹ = A⁻¹B⁻¹. Wrong. It is B⁻¹A⁻¹. Same reverse-order rule. This comes up in theoretical MCQs and in proving properties of orthogonal matrices.
Confusing |adj A| = |A|^(n-1) with |adj A| = |A|. For n = 2, |adj A| = |A|¹ = |A| — that coincidence makes students think the formula is just |A| for all sizes. For n = 3, |adj A| = |A|². Test yourself on n = 3 specifically.