A determinant is a single scalar value extracted from a square matrix. Think of it as a compression — you feed in an entire grid of numbers and get one number out that encodes critical geometric and algebraic information about the matrix.
Here is the most useful physical intuition: if you have two vectors in 2D, the absolute value of their determinant equals the area of the parallelogram they span. In 3D, a 3×3 determinant gives the signed volume of the parallelepiped. When that area or volume collapses to zero — meaning the vectors are coplanar or collinear — the determinant is zero. That matrix is called singular (অবিচ্ছেদ্য নয়, as it were), and it has no inverse.
For a 2×2 matrix:
This is the formula you already know. The 3×3 case extends this through cofactor expansion — you pick any row or column, multiply each element by its cofactor, and sum. The cofactor C_{ij} of element a_{ij} is (-1)^{i+j} times the determinant of the submatrix left after deleting row i and column j.
The sign pattern for a 3×3 is:
Why does any of this matter for NDA? Look at the PYQs: almost every determinant question is not asking you to grind through the full expansion. They are testing whether you recognize a property — a row sum of zero, a disguised linear dependence, a trigonometric identity that collapses a row — and apply it immediately. The candidate who computes from scratch runs out of time. The candidate who spots the structure answers in 30 seconds.
Property 1 — Row/Column Swap: Swapping any two rows (or columns) reverses the sign of the determinant. Swap twice — back to the original value.
Property 2 — Proportional Rows/Columns: If any two rows (or columns) are identical or proportional, the determinant is zero. This is the most-used property in NDA questions. When you see rows that are multiples of each other, write = 0 immediately.
Property 3 — Linear Combination of Rows: You can add any scalar multiple of one row to another row without changing the determinant. This is your primary manipulation tool. Formally: R_i → R_i + k·R_j leaves the value unchanged.
Property 4 — Scalar Multiplication: If every element of one row is multiplied by a scalar k, the determinant is multiplied by k. Critically — for an n×n matrix, |kA| = k^n |A|, NOT k·|A|.
Property 5 — Transpose: |A^T| = |A|. Determinant is unaffected by transposition. This means all row properties apply equally to columns.
Property 6 — Row Sum Zero Implies Linear Dependence: If there exist scalars (not all zero) such that a linear combination of rows gives the zero row, the rows are linearly dependent and |A| = 0. The most direct version: if R_1 + R_2 + R_3 = 0 (the zero row), the determinant is zero.
Always expand along the row or column with the most zeros. For the matrix:
Expand along the second row (two zeros):
That took 4 seconds instead of the full expansion's 25 seconds.
|A| = 0 means A is singular — no inverse exists. The inverse formula is:
The adjugate adj(A) is the transpose of the cofactor matrix. The denominator |A| being zero makes the inverse undefined.
For a system AX = B where A is n×n:
where A_i is the matrix formed by replacing column i of A with the column vector B. This works only when |A| ≠ 0. When |A| = 0, the system either has no solution or infinitely many — you cannot use Cramer's rule.
So |A^2| = |A|^2, |A^n| = |A|^n. If A is orthogonal (AA^T = I), then |A|^2 = 1, so |A| = ±1.
A large cluster of NDA questions pairs determinants with triangle geometry. The key tools:
a/sinA = b/sinB = c/sinC = 2RA + B + C = π, so tan(B+C) = -tanA, sin(B+C) = sinA, etc.tanA, sinB, cosC entries, always ask: can I substitute using B + C = π - A?The pattern in PYQs is: recognize that the triangle constraint forces a row sum to zero or makes two rows linearly dependent, then call the determinant zero immediately.
When you see a determinant where each row sums to the same value, factor that value out using C_1 → C_1 + C_2 + C_3, creating an all-same first column. If that sum is zero, the determinant is zero instantly.
Micro-example: |a-b, b-c, c-a; b-c, c-a, a-b; c-a, a-b, b-c| — add all three elements in any row: (a-b)+(b-c)+(c-a) = 0. Each row sums to zero, so R_1 + R_2 + R_3 = zero row → determinant = 0.
Standard expansion: ~90 seconds. Pattern recognition: 8 seconds.
Whenever a triangle problem puts tan(B+C) in a determinant, replace it immediately with -tanA. Since A + B + C = π, B + C = π - A, and tan(π - A) = -tanA.
This single substitution typically creates two identical terms that cancel, forcing the determinant to zero without any expansion needed.
Standard expansion + trig: ~3 minutes. Substitution first: 25 seconds.
The circulant determinant |a b c; b c a; c a b| has a known closed form:
Memorize this factored form. When you see the condition a + b + c = 0 anywhere in the problem, call the determinant zero immediately. The identity x^3 + y^3 + z^3 - 3xyz = (x+y+z)(x^2+y^2+z^2-xy-yz-zx) connects the sin³A + sin³B + sin³C = 3sinA sinB sinC condition to a + b + c = 0 via the sine rule.
Full expansion: ~4 minutes. Memorized form: 15 seconds.
Before starting any 3×3 expansion, scan all six rows+columns and count zeros. Expand along whichever has the most.
Each zero kills one 2×2 sub-determinant calculation. A row with two zeros reduces the problem to a single 2×2 determinant.
3-zero expansion: 1 term (trivial). 0-zero expansion: 3 terms, each needing a 2×2 calc — roughly 4× slower.
If A = B = C = 0 (or any condition that forces all angles or variables equal), write down the matrix explicitly and check for identical rows or columns before expanding.
Example: if A = B = C = 0, then cosA = cosB = cosC = 1, and the matrix |1 1 1; 1 1 1; 1 1 1| has all identical rows → determinant = 0. Calculation time: zero.
When a determinant question appears in the exam hall, run this decision tree:
Step 1 — Check for structural zeros (10 seconds).
Can you immediately see a zero row/column sum? Proportional rows? If yes, write = 0 and move on.
Step 2 — Check for a geometric/algebraic constraint.
Is there a triangle condition (A + B + C = π)? An algebraic identity (a + b + c = 0, A² + B² + C² = 0)? Apply the substitution and re-check Step 1.
Step 3 — Apply row/column operations.
Use R_i → R_i - R_j or C_i → C_i + C_j to manufacture zeros. Goal: get two zeros in one row or column.
Step 4 — Expand along the sparsest row/column. Only now start arithmetic. Expand along the row/column with the most zeros.
Step 5 — If answer options are numbers, try substitution.
Plug in a specific simple case (a = 1, b = 2, c = 3) to verify your value. This is faster than second-guessing algebra.
If you are past Step 3 and still stuck, move on and come back. Most NDA determinant questions resolve at Step 1 or 2.
Why this question: Tests whether you recognize that a row-sum-to-zero condition immediately kills the determinant, without any expansion.
Solving path: Add all three rows element by element in the left-hand determinant. Row 1 element 1: (a-b)+(b-c)+(c-a) = 0. The same cancellation happens for p,q,r and x,y,z. So adding R_1 + R_2 + R_3 gives the zero row — rows are linearly dependent — determinant = 0. Therefore k·|RHS| = 0, and since the right-hand determinant is not necessarily zero, k = 0.
Why this question: Tests cofactor expansion of a specific 3×3 determinant and connecting the sign of the result to an inequality between two expressions.
Solving path: Expand |x 1 1; 1 y 1; 1 1 z| along row 1:
= x(yz - 1) - 1(z - 1) + 1(1 - y)
= xyz - x - z + 1 + 1 - y
= xyz - (x + y + z) + 2
= q - p + 2.
For the determinant to be positive: q - p + 2 > 0, i.e., q + 2 > p. Answer: option C.
Why this question: Tests whether you can use the real-number constraint A² + B² + C² = 0 to force all variables to zero, then apply the identical-rows property.
Solving path: Over the reals, A² + B² + C² = 0 forces A = B = C = 0. Then cosA = cosB = cosC = 1. The matrix becomes |1 1 1; 1 1 1; 1 1 1|. All three rows are identical — determinant = 0 by the proportional-rows property.
Why this question: Classic triangle + determinant question testing the tan(B+C) = -tanA substitution.
Solving path: Since A + B + C = π, we have B + C = π - A, so tan(B+C) = tan(π - A) = -tanA. Substitute this into the matrix. Now expand along the first row:
cosC · [0·cosC - sinB·(-tanA)] - sinB · [tanA·cosC - sinB·0] + 0
= cosC · sinB · tanA - sinB · tanA · cosC
= 0.
The two surviving terms cancel exactly because tan(B+C) was replaced by -tanA.
Why this question: Connects the algebraic identity x³ + y³ + z³ = 3xyz to the cyclic determinant and the sine rule.
Solving path: The identity x³ + y³ + z³ - 3xyz = (x+y+z)(x²+y²+z²-xy-yz-zx) means sin³A + sin³B + sin³C = 3sinA sinB sinC implies (sinA + sinB + sinC)(sinA² + sinB² + sinC² - sinA sinB - sinB sinC - sinC sinA) = 0. Since all sines are positive in a triangle, the second factor must be zero (it equals ½[(sinA-sinB)² + (sinB-sinC)² + (sinC-sinA)²] = 0), so sinA = sinB = sinC, meaning a = b = c by the sine rule. Alternatively, using the sine rule, sinA + sinB + sinC = 0 is impossible for a valid triangle — instead, a + b + c = 0 cannot hold for positive sides. The correct track: the cyclic determinant |a b c; b c a; c a b| = -(a³+b³+c³-3abc) = -(a+b+c)(a²+b²+c²-ab-bc-ca). Since sin³A + sin³B + sin³C = 3sinA sinB sinC forces a = b = c (all sides equal), both factors of the product factored form could be nonzero — but for the NDA solution, the answer follows from the sine rule proportionality: when a = b = c, two rows of the cyclic determinant become identical, so = 0.
Applying |kA| = k|A| instead of |kA| = k^n|A|. For a 3×3 matrix, |2A| = 8|A|, not 2|A|. This trips up at least one question per mock exam.
Forgetting that adding a scalar multiple of one row to another does NOT change the determinant, but multiplying an entire row by k DOES. These are different operations. A row operation R_i → R_i + kR_j is free. A scaling R_i → kR_i multiplies the determinant by k.
Expanding along a dense row when a sparse one is available. Always scan all six rows/columns for zeros before starting. Expanding along the wrong row costs 2-3 minutes in a timed exam.
Missing the triangle identity tan(B+C) = -tanA. When A + B + C = π and you see tan(B+C) in the matrix, substitute immediately. Candidates who skip this spend 4+ minutes on what is a 30-second problem.
Applying Cramer's rule when |A| = 0. If the determinant of the coefficient matrix is zero, Cramer's rule is invalid. The system either has no solution or infinitely many — you need a different approach (row reduction).
Confusing cofactor sign: (-1)^{i+j}. The element at position (1,2) has sign (-1)^{1+2} = -1. Writing + instead of - for position (1,2) is the single most common arithmetic error in 3×3 expansions. Refer to the + - +; - + -; + - + pattern every time.