Quick Summary: Linear Algebra contributes roughly 8–10 marks in GATE DA — and it secretly powers many more, because PCA, SVD, regression and neural networks are all linear algebra in disguise. The official IIT Madras GATE 2027 syllabus covers vector spaces, special matrices, quadratic forms, systems of linear equations, Gaussian elimination, eigenvalues and eigenvectors, determinant-rank-nullity, projections, LU decomposition and the singular value decomposition. This guide explains every listed topic with diagrams, a solved eigenvalue example, and the exact GATE question patterns.
By Piyush Wairale — Instructor, BS Data Science program at IIT Madras · IIT Madras alumnus · 10,000+ GATE students mentored · Last updated: August 2026 · Verified against the official IIT Madras GATE 2027 DA syllabus
Key Takeaways
- The rank-nullity theorem — rank(A) + nullity(A) = number of columns — is the most-used identity in GATE linear algebra questions.
- Ax = b has a unique solution when rank(A) = rank([A|b]) = n, infinitely many when rank(A) = rank([A|b]) < n, and no solution when rank(A) < rank([A|b]).
- The trace of a matrix equals the sum of its eigenvalues; the determinant equals their product — two shortcuts that solve most 2×2 and 3×3 eigenvalue questions without the characteristic polynomial.
- A symmetric matrix always has real eigenvalues and orthogonal eigenvectors — which is exactly why PCA (eigen-analysis of a covariance matrix) works.
- SVD factorises any m×n matrix as A = UΣVᵀ; the singular values are the square roots of the eigenvalues of AᵀA, and the right singular vectors of a centred data matrix are its principal components.
On this page
Why linear algebra matters · Vector spaces & independence · Special matrices & quadratic forms · Systems of equations & Gaussian elimination · Determinant, rank & nullity · Eigenvalues (solved example) · LU & SVD · How GATE tests it · 4-week plan · FAQs
Linear algebra is the quiet workhorse of GATE DA. Its direct questions are short and concept-driven — a rank here, an eigenvalue there — but its ideas run underneath half the paper: multiple regression is a projection onto a column space, PCA is an eigendecomposition, ridge regression is a modified normal equation, and SVD sits behind dimensionality reduction. That double role makes it the recommended first subject in your preparation sequence. This guide covers every topic named in the official IIT Madras GATE 2027 syllabus, with the intuitions and the exam patterns — and pairs with the Linear Algebra for GATE DA Course & Test Series for lectures, PYQs and sectional tests.
Watch Free: Linear Algebra Lectures on YouTube
Piyush Wairale’s channel has free subject-wise GATE DA lectures — including the Linear Algebra series with PYQ-style problems solved step by step:
- All subject playlists — Linear Algebra, Probability & Statistics, DBMS, Python/DSA and more
- Machine Learning for GATE DA — full playlist (PCA and SVD live here too)
Subscribe to Piyush Wairale IITM on YouTube for new GATE DA lectures, PYQ solutions and strategy sessions.
Why Linear Algebra Is the Right First Subject
Three reasons this subject should open your preparation. First, it is a prerequisite multiplier: finish linear algebra and the mathematics inside PCA, SVD, regression and neural networks stops being mysterious — the ML section becomes learnable at full speed. Second, GATE tests it conceptually rather than computationally: you rarely grind through a 4×4 determinant; instead you reason about ranks, eigenvalue properties and solution existence, which rewards understanding over stamina. Third, its facts compress beautifully — a one-page sheet of identities (trace = sum of eigenvalues, det = product, rank-nullity, symmetric ⇒ real eigenvalues) answers a remarkable share of past questions. The syllabus is compact; four focused weeks close it.
Vector Spaces, Subspaces and Linear Independence
A vector space is a set of vectors closed under addition and scalar multiplication; a subspace is a subset that is itself a vector space — it must contain the zero vector and be closed under both operations (the standard MCQ trap: a line not through the origin is not a subspace). Vectors v₁,…,vk are linearly independent when c₁v₁ + … + ckvk = 0 forces every cᵢ = 0 — no vector is a combination of the others. The span of a set is all its linear combinations; a basis is an independent set that spans the space, and the dimension is the size of any basis. Quick tests GATE expects: k vectors in ℝⁿ with k > n are automatically dependent; n vectors in ℝⁿ are independent iff the matrix stacking them has non-zero determinant (equivalently full rank); any set containing the zero vector is dependent.
Special Matrices and Quadratic Forms
The syllabus names four special families, and each carries one or two exam-ready properties. A projection matrix P projects vectors onto a subspace; the orthogonal projection onto the column space of A is P = A(AᵀA)⁻¹Aᵀ — exactly the matrix behind least-squares regression. An orthogonal matrix Q satisfies QᵀQ = I: its columns are orthonormal, it preserves lengths and angles, its inverse is its transpose, and |det Q| = 1. An idempotent matrix satisfies A² = A; every projection matrix is idempotent, and the eigenvalues of an idempotent matrix can only be 0 or 1 — a favourite one-marker (proof in one line: Ax = λx ⇒ A²x = λ²x = λx ⇒ λ² = λ). A partition (block) matrix is a matrix written in sub-blocks; block multiplication follows the same rules as scalar multiplication provided dimensions match, and for block-diagonal matrices the determinant is the product of the blocks’ determinants.
A quadratic form is xᵀAx with A symmetric. Its character is decided by the eigenvalues of A: all positive ⇒ positive definite; all non-negative ⇒ positive semi-definite; mixed signs ⇒ indefinite. For a 2×2 symmetric matrix, positive definiteness is equivalent to a₁₁ > 0 and det A > 0 (leading principal minors) — the fastest route in an exam. Quadratic forms reappear in ML as loss surfaces: a positive-definite Hessian means a strictly convex bowl with a unique minimum.
Systems of Linear Equations and Gaussian Elimination
Gaussian elimination reduces [A|b] to row-echelon form with three row operations (swap, scale, add a multiple of another row), then back-substitutes. Its by-products are most of what GATE asks about: the number of non-zero rows is the rank, the pivots tell you which variables are determined, and comparing rank(A) with rank([A|b]) classifies the solution set completely:
For the homogeneous system Ax = 0, a non-trivial solution exists iff rank(A) < n — equivalently det A = 0 for square A. Parameter-hunting questions (“find k so that…”) reduce to forcing a determinant to zero or forcing a rank drop; practise both mechanics.
Determinant, Rank, Nullity and Projections
Determinant facts that pay: det(AB) = det A · det B; det(Aᵀ) = det A; det(kA) = kⁿ det A for an n×n matrix; a matrix is invertible iff det ≠ 0; row swaps flip the sign, and a matrix with a repeated row (or a row of zeros) has determinant zero. Rank is the dimension of the column space (= dimension of the row space — column rank equals row rank, a classic MCQ); nullity is the dimension of the null space {x : Ax = 0}. The rank-nullity theorem ties them: rank + nullity = n (number of columns). Also remember rank(AB) ≤ min(rank A, rank B) and that multiplying by an invertible matrix preserves rank. Projections close the loop with statistics: projecting b onto the column space of A gives the least-squares fitted values, with P = A(AᵀA)⁻¹Aᵀ idempotent and symmetric — GATE has asked for the projection of a vector onto a line span{a}, which is simply (aᵀb/aᵀa)·a.
Eigenvalues and Eigenvectors — With a Solved Example
An eigenvector of A is a non-zero vector whose direction A does not change: Av = λv, where λ is its eigenvalue. Eigenvalues solve the characteristic equation det(A − λI) = 0. The identity shortcuts win exams: trace(A) = Σλᵢ and det(A) = Πλᵢ; the eigenvalues of a triangular matrix are its diagonal entries; the eigenvalues of Aᵏ are λᵏ; A is singular iff 0 is an eigenvalue; and a symmetric matrix has real eigenvalues with orthogonal eigenvectors — the spectral theorem, and the reason PCA works.
Solved example (the standard 2-mark shape). Find the eigenvalues and eigenvectors of A = [[4, 1], [2, 3]]. Trace = 7, det = 4·3 − 1·2 = 10, so the eigenvalues satisfy λ² − 7λ + 10 = 0 → (λ−5)(λ−2) = 0 → λ = 5 and λ = 2 (check: 5+2 = trace ✓, 5×2 = det ✓ — the shortcut alone answers a “sum/product of eigenvalues” NAT instantly). For λ = 5: (A − 5I)v = 0 gives [−1, 1; 2, −2]v = 0, so v₁ = (1, 1). For λ = 2: [2, 1; 2, 1]v = 0 gives v₂ = (1, −2). Practise this pattern until it takes ninety seconds; then practise the reverse pattern (given eigenvalues and one entry, recover the matrix using trace and determinant).
LU Decomposition and the Singular Value Decomposition
LU decomposition writes a square matrix as A = LU, lower-triangular times upper-triangular — it is Gaussian elimination recorded as a factorisation. Existence requires elimination without row swaps (otherwise PA = LU with a permutation). Why it matters: solving Ax = b becomes two triangular solves (Ly = b, then Ux = y), so multiple right-hand sides reuse one factorisation; and det A is the product of U’s diagonal. GATE asks you to compute L and U for a 2×2 or 3×3, or to use the convention ℓᵢᵢ = 1 (Doolittle) to identify entries.
The singular value decomposition factorises any m×n matrix as A = UΣVᵀ: U (m×m) and V (n×n) orthogonal, Σ diagonal with non-negative singular values σ₁ ≥ σ₂ ≥ … ≥ 0. The computational route: the σᵢ are the square roots of the eigenvalues of AᵀA (equivalently AAᵀ); V’s columns are eigenvectors of AᵀA, U’s of AAᵀ. Exam-ready facts: the number of non-zero singular values equals rank(A); SVD exists for every matrix (unlike eigendecomposition); and for a centred data matrix the right singular vectors are precisely the principal components — the bridge to the PCA section of our Machine Learning guide. A typical NAT gives a small matrix and asks for its largest singular value: form AᵀA, find its largest eigenvalue, take the square root.
How GATE Actually Tests Linear Algebra
| Question pattern | What it looks like | Typical marks |
|---|---|---|
| Eigenvalue shortcuts | Sum/product of eigenvalues from trace and determinant; eigenvalues of Aᵏ or A⁻¹ | 1–2 |
| Solution classification | Value of a parameter for unique / infinite / no solutions | 2 |
| Rank & nullity | Rank of a given matrix; nullity via rank-nullity; rank of AB | 1–2 |
| Special-matrix properties | MSQ on orthogonal/idempotent/projection facts (e.g., eigenvalues 0 or 1) | 1–2 |
| SVD numerical | Largest singular value of a small matrix via AᵀA | 2 |
| Independence / basis checks | Which sets are linearly independent; dimension of a span or subspace | 1 |
| Quadratic-form definiteness | Classify xᵀAx via eigenvalue signs or principal minors | 1–2 |
The 4-Week Linear Algebra Study Plan
- Week 1 — Spaces and systems: vector spaces, independence, basis and dimension; Gaussian elimination and the solution-classification diagram; 25 problems including parameter-hunting questions.
- Week 2 — Matrices and rank: the four special matrix families and their one-line properties; determinant identities; rank-nullity drills; projections onto lines and column spaces.
- Week 3 — Eigen-everything: characteristic equations for 2×2 and 3×3; trace/determinant shortcuts; symmetric-matrix facts; quadratic-form classification.
- Week 4 — Decompositions and integration: LU by hand; SVD via AᵀA; connect to PCA in the ML guide; sectional test + GATE DA/CS PYQs with an error log.
Study Linear Algebra the Structured Way
Every topic above — with visual intuition, solved GATE PYQs, topic-wise tests and doubt support:
- Linear Algebra for GATE DA Course & Test Series — the complete syllabus, taught by IIT Madras alumnus Piyush Wairale.
- Complete GATE DA 2027 Course — all 7 subjects, 250+ hours, 4,000+ questions and full-length AI mock tests in one package.
FAQs on Linear Algebra for GATE DA
How many marks does Linear Algebra carry in GATE DA?
Typically 8–10 marks directly — plus the ML questions that ride on it (PCA, SVD, regression normal equations), which push its effective weight well beyond the headline number.
Do I need to compute SVD fully by hand?
For GATE, you need the recipe (singular values from eigenvalues of AᵀA) and the properties (existence for any matrix, rank = number of non-zero singular values, PCA connection). Full three-factor computations by hand are rare; largest-singular-value NATs are common.
What is the single most useful shortcut in this subject?
Trace = sum of eigenvalues and determinant = product of eigenvalues. Combined with “eigenvalues of a triangular matrix are its diagonal”, these settle a large fraction of GATE’s eigenvalue questions without solving the characteristic polynomial.
Which book should I follow?
Gilbert Strang’s “Introduction to Linear Algebra”, supported by his free MIT OCW lectures. See the best books for GATE DA guide for the full subject-wise list.
When should I study Linear Algebra in my preparation?
First — alongside Probability & Statistics, before Machine Learning. That ordering (detailed in the 6-month plan) means every later subject builds on ready foundations.
Linear algebra in GATE DA is a concept exam, not a computation exam. Internalise the solution-classification diagram, the eigenvalue shortcuts, the special-matrix property list and the SVD-to-PCA bridge, and its 8–10 marks become among the most reliable in your paper — while quietly making Machine Learning easier. Full roadmap: GATE DA Syllabus 2027.
Master GATE DA Linear Algebra
Vector spaces to SVD — lectures, solved PYQs, sectional tests and mentorship by Piyush Wairale (IIT Madras).
Join the Linear Algebra Course Get the Complete GATE DA CourseRecent Post
Programming, Data Structures and Algorithms carries ~10–14 marks in GATE DA — and it's the most practice-rewarding subject in the paper. This complete guide covers Python as GATE tests it, all five data structures, every named sorting and searching algorithm with a complexity table, graph traversals and shortest path, with diagrams, a worked output-prediction example, GATE patterns, a 5-week plan and FAQs, by IIT Madras instructor Piyush Wairale.
Probability and Statistics carries ~12–16 marks in GATE DA and underpins Machine Learning. This complete guide covers the full official syllabus — counting, Bayes' theorem, expectation, every named distribution, CLT, confidence intervals and the z/t/chi-squared tests — with diagrams, a worked Bayes example, GATE patterns, a study plan and FAQs, by IIT Madras instructor Piyush Wairale.
Machine Learning is the highest-weightage subject in GATE DA (~18–22 marks). This complete guide covers the full official syllabus — regression, classifiers, SVM, decision trees, bias-variance, cross-validation, neural networks, clustering and PCA — with diagrams, GATE question patterns, a 6-week plan and FAQs, by IIT Madras instructor Piyush Wairale.
Master reasoning under uncertainty for GATE DA 2027 — Bayesian networks, conditional independence, d-separation, variable elimination with a fully worked alarm-network example, and sampling-based approximate inference. With diagrams, GATE question patterns, a 10-day plan and FAQs, by IIT Madras instructor Piyush Wairale.
Complete GATE RA 2027 guide to servo motors: closed-loop principle, AC/DC/BLDC types, continuous vs intermittent torque-speed zones, nested PID control, sizing, diagrams and worked example.
Complete GATE RA 2027 guide to stepper motors: VR/PM/hybrid types, step angle, full/half/microstepping, pull-in and pull-out torque-speed characteristics, drives, diagrams and worked examples.
Learn Daily, Wherever You Are
Free lectures, exam updates, PYQ discussions, and job alerts — delivered through our YouTube channel and Telegram communities.
