GATE Electrical: Linear Algebra (Engineering Mathematics), previous-year questions solved

Every linear algebra question from the last five GATE Electrical papers, 15 in all: skew-symmetric and orthogonal matrices, rotations and normal vectors, invertibility and rank, consistency of linear systems, eigenvalues, eigenvectors, the matrix exponential and Cayley-Hamilton. Each solution matches the official answer key, uses the exam shortcut, names the trap, and shows where the same matrix idea is used in power systems, machines or control.

Questions from official GATE EE papers (IIT Kharagpur 2022, IIT Kanpur 2023, IISc Bengaluru 2024, IIT Roorkee 2025, IIT Guwahati 2026); solutions by Skill-Lync. · Updated 16 Nov 2026 · 15 questions

GATE 2022 · EE · Q.20MCQ · 1 markMatrix properties and special matrices

3 × 3 matrix with elements (i − j)³: symmetric, skew-symmetric, unitary or null?

Consider a matrix whose -th element, . Then the matrix will be
  1. (A)symmetric.
  2. (B)skew-symmetric.
  3. (C)unitary.
  4. (D)null.

Answer: (B) Official key: B

  1. Swap the indices: , because an odd power keeps the sign. So : option (B).
  2. Written out, . The diagonal is zero, as it must be for any skew-symmetric matrix.
  3. It is not null (off-diagonal entries are non-zero) and not unitary: the first row has length .

Trap: with , or any even power, the sign disappears and the matrix is symmetric. Check the parity of the exponent before choosing.

Where this shows up in industry

Skew-symmetric matrices appear whenever something rotates. Differentiating the Park transformation matrix with respect to time gives times a skew-symmetric matrix, (sign depending on the convention), and that is exactly where the speed-voltage terms and in the dq equations of a synchronous or induction machine come from.

GATE 2026 · EE · Q.35NAT · 1 markMatrix properties and special matrices

A is real skew-symmetric and xᵀx = 1: what is xᵀAx?

is an skew-symmetric matrix with real-valued entries, and is an -dimensional column vector with real-valued entries such that . The quantity evaluates to _________ (Answer in integer)

Answer: 0 Official key: 0 to 0

  1. is a matrix, a plain number, so it equals its own transpose: .
  2. Skew-symmetric means , so , which forces .
  3. The condition only fixes the length of ; the answer is 0 for every .

Term by term: , and each pair cancels because , while the diagonal is zero. We checked 300 random skew-symmetric matrices of size 2 to 6 with random vectors in exact fractions: every result was exactly 0.

Where this shows up in industry

An ideal gyrator has the impedance matrix , which is skew-symmetric. The power it absorbs is for any pair of port currents, so it is lossless: it only moves power from one port to the other. The same argument shows that any skew-symmetric part of a network's impedance matrix can never dissipate power.

GATE 2026 · EE · Q.55MSQ · 2 marksMatrix properties and special matrices

Orthogonal matrix A: determinant, eigenvalues and preserved lengths and dot products

Consider an orthogonal matrix with real entries and each column having unit Euclidean norm. Which of the following statements is/are correct?
  1. (A)The value of the determinant of is either or
  2. (B)The eigenvalues of have modulus 1
  3. (C), for all , where denotes the Euclidean norm of , and , for all distinct
  4. (D), for all , where denotes the Euclidean norm of , and , for all

Answer: (A), (B) and (D) Official key: A;B;D

  1. Orthogonal means . Take determinants: , so : (A) is correct.
  2. for every , and with this gives : (D) is correct.
  3. If (allowing complex and ), lengths are preserved, so and : (B) is correct. A rotation by has eigenvalues , complex but of modulus 1.
  4. (C) contradicts the identity in step 2: dot products are preserved for every pair, distinct or not. Answer (A), (B) and (D).

We drew 200 random orthogonal matrices (rotations and reflections, so determinants of both and came up) and checked every claim; the eigenvalue moduli came from the roots of each characteristic polynomial.

Where this shows up in industry

The power-invariant Clarke and Park transforms (scaled by ) are orthogonal matrices, so is the same in abc and in dq0 coordinates: power computed in the rotating frame is the real three-phase power. The amplitude-invariant version (scaled by ) is not orthogonal, which is why its power formula carries the extra factor , as in .

GATE 2025 · EE · Q.14MCQ · 1 markMatrix properties and special matrices

P = [[2, 1, 0], [−1, 0, 0], [0, 0, 1]]: what is P² equal to?

Let and let be the identity matrix. Then is equal to
  1. (A)
  2. (B)
  3. (C)
  4. (D)

Answer: (A) Official key: A

  1. Multiply directly: .
  2. , the same matrix: option (A).

Shortcut: the top-left block has trace 2 and determinant 1, so by Cayley-Hamilton it satisfies , that is ; the bottom-right 1 satisfies as well. The corner entry alone rules out (B) and (C): , while and have 2 and 1 there. (D) matches that corner but fails at entry , where has 1 and has .

Where this shows up in industry

All three eigenvalues of equal 1, yet while , so has fewer independent eigenvectors than its size. As a discrete-time state matrix, , this matters: grows linearly with even though every eigenvalue sits on the unit circle. A repeated pole on the stability boundary (like a double pole at , a digital double integrator) is unstable, which is why controller design checks the eigenvector structure, not only the eigenvalue magnitudes.

GATE 2023 · EE · Q.11MCQ · 1 markVectors, linear independence and transformations

w = [1 2 3]ᵀ: which vector is normal to the plane wᵀx = 1?

For a given vector , the vector normal to the plane defined by is
  1. (A)
  2. (B)
  3. (C)
  4. (D)

Answer: (D) Official key: D

  1. Take any two points on the plane. Subtracting and gives : is perpendicular to every direction lying in the plane.
  2. So the normal is itself, : option (D). The right-hand side (here 1) only shifts the plane; it does not tilt it.

Traps: options (A) and (B) satisfy ( and ), so they lie in the plane, the opposite of normal. Option (C) is written backwards; and it is not parallel to .

Where this shows up in industry

Balanced three-phase quantities satisfy , that is : they live in a plane whose normal is . The Clarke transform splits any set of phase values into the two in-plane components and the zero-sequence component along that normal, . Ground faults and unbalanced loads are what push a system off the plane.

GATE 2023 · EE · Q.24MCQ · 1 markVectors, linear independence and transformations

Matrix A maps u = (4, 3) to v = (5, 0): which A is correct?

In the figure, the vectors and are related as: by a transformation matrix . The correct choice of is
Vector u from the origin (0, 0) to the point (4, 3); vector v from the origin along the x axis to (5, 0)(0, 0)(4, 3)u(5, 0)v
  1. (A)
  2. (B)
  3. (C)
  4. (D)

Answer: (A) Official key: A

  1. Just apply each option to and look for .
  2. (A): . Option (A).
  3. (B) gives , (C) gives (right first entry, but off the axis) and (D) gives : (D) lands on the right axis but at the wrong length.

Why (A): , so only has to rotate. sits at above the axis, so is a clockwise rotation by : with , . Option (B) is the same rotation anticlockwise, the classic sign trap: it turns further away from the axis, to , which sits at .

Where this shows up in industry

This is the Park transformation in miniature: a rotation matrix with and that turns a rotating space vector so it lies along one axis. In field-oriented control of a motor, or the PLL of a grid-tied inverter, is chosen so the flux or grid-voltage vector sits on the d axis, exactly as puts on the axis; the other component is then zero and the controller works with DC quantities.

GATE 2026 · EE · Q.49MCQ · 2 marksVectors, linear independence and transformations

p column vectors with n real entries: which statement is ALWAYS correct?

Which one of the following statements is ALWAYS correct about a collection of column vectors, each having real-valued entries?
  1. (A)If , then the column vectors must be linearly dependent
  2. (B)If , then the column vectors must be linearly independent
  3. (C)If , then the column vectors must be orthogonal
  4. (D)If , then the column vectors must be linearly independent

Answer: (A) Official key: A

  1. Put the vectors side by side as an matrix. Its rank is at most the number of rows, . If , the rank is below , so some combination of the columns with not-all-zero weights gives the zero vector: the vectors are dependent. Option (A), and (B) is the opposite claim.
  2. (C) fails: and are two vectors in that are independent but not orthogonal (dot product 1). Orthogonality is much stronger than independence.
  3. (D) fails: and are two vectors in and one is twice the other. Few vectors can be independent; nothing forces them to be.

We also built 300 random sets with (integer entries, from 1 to 4): every one had rank at most , so all were dependent.

Where this shows up in industry

Power-system state estimation leans on this. With unknown bus voltages and angles, any set of more than measurements is linearly dependent, and that redundancy is the point: the estimator uses it to detect and reject a bad meter reading. The opposite case is (D): having meters does not guarantee observability, because poorly placed meters can still give dependent rows in the measurement matrix.

GATE 2024 · EE · Q.11MCQ · 1 markInverse, rank and systems of linear equations

Four 3 × 3 matrices: which one has an inverse?

Which one of the following matrices has an inverse?
  1. (A)
  2. (B)
  3. (C)
  4. (D)

Answer: (C) Official key: C

  1. A matrix is invertible exactly when its rows are independent (determinant non-zero). Look for a row that is a multiple of another before expanding anything.
  2. (A): row 3 row 1. (B): row 2 row 1. (D): row 3 row 1. All three are singular.
  3. (C): . Option (C).

Notice that (A), (C) and (D) share the first two rows; only the third row decides. The exam is testing whether you spot proportional rows in a few seconds instead of expanding four determinants.

Where this shows up in industry

The DC load-flow matrix built from line susceptances alone has every row summing to zero (each row is a bus, and the entries balance), so, like options (A), (B) and (D), it is singular. Load-flow programs remove the slack bus row and column, which makes the reduced matrix invertible and fixes the angle reference; without that step the solver reports a singular matrix.

GATE 2025 · EE · Q.13MCQ · 1 markInverse, rank and systems of linear equations

A = [[1, 1, 1], [−1, −1, −1], [0, 1, −1]], b = [1/3, −1/3, 0]: how many solutions?

Let , and . Then, the system of linear equations has
  1. (A)a unique solution.
  2. (B)infinitely many solutions.
  3. (C)a finite number of solutions.
  4. (D)no solution.

Answer: (B) Official key: B

  1. Row 2 is row 1, and too. So equation 2 is equation 1 multiplied by : it adds nothing, and it does not contradict anything.
  2. That leaves two independent equations, and , in three unknowns: .
  3. One unknown is free: , for any real . Option (B).

Trap: seeing and jumping to “no solution”. A singular matrix gives either no solution or infinitely many; which one depends on . Had been anything other than , the ranks would differ and (D) would be right. (C) is never possible for a linear system.

Where this shows up in industry

Writing Kirchhoff's current law at every node of a circuit gives exactly this kind of system: the node equations add up to zero, so one of them is always a combination of the others, just as row 2 here is minus row 1. That is why nodal analysis picks a reference (ground) node and drops its equation; the remaining equations then have a unique solution.

GATE 2026 · EE · Q.56MSQ · 2 marksInverse, rank and systems of linear equations

Ax = b has a unique solution: what follows about A⁻¹, Aᵐ and the ranks?

Consider the system of linear equations: , where is an matrix, and and are -dimensional column vectors. Suppose this system of equations has a unique solution. Which of the following statements is/are correct?
  1. (A) exists
  2. (B)The system of equations also has a unique solution for
  3. (C), for
  4. (D), where denotes the augmented matrix.

Answer: (A), (B) and (C) Official key: A;B;C

  1. A square system has a unique solution only when (otherwise there is either no solution or infinitely many). So exists: (A).
  2. , so is invertible too and has the unique solution : (B).
  3. Both and are invertible matrices, so both have rank : (C).
  4. (D) describes an inconsistent system (no solution at all). Here . Answer (A), (B) and (C).

We checked 200 random non-singular integer matrices of size 2 to 4 with exact fractions: had full rank for to 4 and the augmented matrix never had a larger rank.

Where this shows up in industry

Each Newton-Raphson load-flow iteration solves , and needs a unique solution, that is a non-singular Jacobian . As a system is loaded towards its maximum transfer limit, the Jacobian approaches singularity at the nose of the PV curve, which is the voltage-collapse point; continuation power-flow tools track its smallest eigenvalue as a stability margin.

GATE 2024 · EE · Q.32NAT · 1 markEigenvalues, eigenvectors and Cayley-Hamilton

Sum of the eigenvalues of [[1, 2], [3, 4]]²

The sum of the eigenvalues of the matrix is ______ (rounded off to the nearest integer).

Answer: 29 Official key: 29 to 29

  1. The sum of the eigenvalues equals the trace. So square the matrix and add the diagonal: , trace .
  2. Check with eigenvalues: has , that is and . Squaring a matrix squares its eigenvalues, and .

Answer 29. Trap: squaring the trace of the original matrix gives ; the square of a sum is not the sum of the squares.

Where this shows up in industry

In a discrete-time model (a digital controller, or a power-system model sampled at a fixed step), two steps are , and each mode is multiplied by . That is why the eigenvalues of must lie inside the unit circle for stability: their powers then shrink at every step. A trace check like this one is a quick sanity test on the eigenvalues a small-signal stability program reports.

GATE 2025 · EE · Q.12MCQ · 1 markEigenvalues, eigenvectors and Cayley-Hamilton

Eigenvectors for distinct eigenvalues of a real symmetric 3 × 3 matrix: which statement is true?

Let and be the two eigenvectors corresponding to distinct eigenvalues of a real symmetric matrix. Which one of the following statements is true?
  1. (A)
  2. (B)
  3. (C)
  4. (D)

Answer: (B) Official key: B

  1. Let and with . Then .
  2. So , and since , : the eigenvectors are orthogonal. Option (B).
  3. (C) and (D) would make , the same direction, which can't belong to two different eigenvalues.

Example: has eigenvectors () and (), dot product 0. Trap: this needs symmetry. has eigenvectors and , which are not orthogonal. We also checked 300 random symmetric integer matrices with distinct eigenvalues: all eigenvector pairs had dot product below .

Where this shows up in industry

A fully transposed three-phase line has a symmetric impedance matrix with equal self terms and equal mutual terms . Its eigenvectors, , and , are orthogonal and are exactly the rows of the Clarke transform. Transforming to these modal (zero, alpha, beta) components decouples the three coupled phases into independent circuits with impedances and , which is how EMTP-type programs model transposed lines.

GATE 2026 · EE · Q.30MSQ · 1 markEigenvalues, eigenvectors and Cayley-Hamilton

A and B share eigenvalue 2 with the same eigenvector: which determinants are zero?

Two matrices and have a common eigenvalue 2, and the same corresponding nonzero eigenvector. Which of the following options is/are correct? (Note: is the identity matrix.)
  1. (A)
  2. (B)
  3. (C)
  4. (D)

Answer: (A), (B) and (D) Official key: A;B;D

  1. Let the shared eigenvector be : and .
  2. with , so is singular: (A). Likewise (B).
  3. Add the two equations: , so is an eigenvector of with eigenvalue 4, and : (D).
  4. (C) would need 2 to be an eigenvalue of , which nothing guarantees. Answer (A), (B) and (D).

We built 120 random pairs of matrices sharing the eigenpair , with exact fractions: (A), (B) and (D) were zero every time, and was non-zero in 114 of them. The shared eigenvector is essential: two matrices that merely share the eigenvalue 2, with different eigenvectors, need not make (D) zero.

Where this shows up in industry

Every transposed three-phase element (a line, a transformer, a generator) has a symmetric circulant impedance matrix, and all of them share the same eigenvectors, the symmetrical-component vectors. That is why sequence impedances of elements in series simply add: the zero-sequence impedance of a line plus a transformer is , just as the eigenvalue of here is . Fault studies would not decouple into separate positive, negative and zero-sequence networks without this.

GATE 2022 · EE · Q.40MCQ · 2 marksEigenvalues, eigenvectors and Cayley-Hamilton

λ, v are an eigenpair of A: is e^λ an eigenvalue and v an eigenvector of e^A?

denotes the exponential of a square matrix A. Suppose is an eigenvalue and is the corresponding eigen-vector of matrix A. Consider the following two statements:
Statement 1: is an eigenvalue of .
Statement 2: is an eigen-vector of .
Which one of the following options is correct?
  1. (A)Statement 1 is true and statement 2 is false.
  2. (B)Statement 1 is false and statement 2 is true.
  3. (C)Both the statements are correct.
  4. (D)Both the statements are false.

Answer: (C) Official key: C

  1. , and gives for every .
  2. So .
  3. One line proves both statements: is an eigenvector of and its eigenvalue is . Option (C).

The same argument works for any power series in : , (eigenvalue ), (eigenvalue ) all keep the eigenvectors. We checked numerically for 7 eigenpairs of four matrices, including the defective that has only one eigenvector; the series was summed to 60 terms.

Where this shows up in industry

The state-transition matrix of is , and each eigenvalue of becomes a mode along its eigenvector. When a controller or a power-system model is discretised with step , the discrete matrix is , so every continuous pole maps to with the same eigenvector, exactly the result of this question. MATLAB's c2d with the zero-order hold computes this matrix exponential.

GATE 2022 · EE · Q.42MCQ · 2 marksEigenvalues, eigenvectors and Cayley-Hamilton

6A⁻¹ = A² + cA + dI for a 3 × 3 matrix: find c + d

Consider a matrix .
The matrix satisfies the equation , where and are scalars and is the identity matrix.
Then is equal to
  1. (A)5
  2. (B)17
  3. (C)
  4. (D)11

Answer: (A) Official key: A

  1. is block diagonal: the 1, and the block with trace 5 and determinant 6, so eigenvalues 2 and 3. The eigenvalues of are 1, 2, 3.
  2. Characteristic polynomial: .
  3. Cayley-Hamilton: . Multiply by : , so .
  4. , , : option (A).

Checked in exact fractions: has rows , , , and equals entry by entry. Traps: the other options are the pieces, (C) and (D), or (B) after dropping the minus sign on .

Where this shows up in industry

Cayley-Hamilton is why the controllability matrix of an -state system stops at : every higher power of is a combination of , so it adds no new directions. It is also how is computed by hand for small state-space models in control and power-electronics courses: , with the 's found from the eigenvalues.

From GATE prep to a power, controls or EV career

Talk to a counsellor.

Talk to a career counsellor

Free 20-minute call. We'll map the right course to where you want to work.

By submitting you agree to our Terms and Privacy Policy.

Talk to a career counsellor

Talk to a career counsellor

Free 20-minute call. We'll map the right course to where you want to work.

By submitting you agree to our Terms and Privacy Policy.