GATE DA linear algebra has a reputation problem. Most aspirants treat it as "the same LA from GATE CS" and copy over their old notes — but the GATE DA syllabus is meaningfully broader. Projection matrices, orthogonal matrices, idempotent matrices, partition matrices, quadratic forms, and SVD are all explicitly listed. CS notes do not cover them. The result: a subject where a well-prepared candidate picks up marks that an under-prepared one did not even know were available.
Below is the full topic map, the recommended book (Gilbert Strang), the "matrix zoo" you must be fluent on, and a study sequence that moves from foundations to eigenvalues to SVD.
In brief
- GATE DA linear algebra is broader than GATE CS — projection, orthogonal, idempotent, partition matrices and quadratic forms are explicitly in the syllabus.
- SVD is in the official syllabus and shows up in past papers. Do not skip it.
- Use Gilbert Strang — Introduction to Linear Algebra paired with the free MIT 18.06 lectures.
- LA is a prerequisite for machine learning (PCA, regression, SVM) — strong LA pays off twice.
Verify first: Confirm the current syllabus version on the official GATE 2027 syllabus page before finalising your plan.
Full Syllabus — Straight from the Official Document
The GATE DA 2027 linear algebra section (IIT Madras — unchanged from 2026) covers:
- Vector space, subspaces, linear dependence and independence of vectors
- Matrices, projection matrix, orthogonal matrix, idempotent matrix, partition matrix and their properties
- Quadratic forms
- Systems of linear equations and solutions; Gaussian elimination
- Eigenvalues and eigenvectors
- Determinant, rank, nullity, projections
- LU decomposition
- Singular Value Decomposition (SVD)
Note: Topics like basis, dimension, inner products, orthogonality, and diagonalisation are standard prerequisites that Strang covers and that feed into the listed topics — they are worth studying even though not explicitly named in the syllabus document.
Where DA Differs from CS
If you are coming from GATE CS material, four additions matter:
- Special matrix types — projection, orthogonal, idempotent, partition matrices — named in the syllabus.
- Quadratic forms — definiteness, signature, connection to convexity.
- SVD — CS notes typically stop at eigen-decomposition.
- Inner products and orthogonality — supporting concepts for the listed projections and orthogonal matrices.
Cover the explicitly listed topics and the foundations needed to understand them. This is a pedagogical comparison, not a claim about relative marks in DA and CS.
Syllabus Overview Table
Scope and weightage: The official syllabus is the checklist above. The table below also includes supporting concepts such as basis, inverse, trace and diagonalisation. Its ML relevance and the study sequence are pedagogical guidance, not mark forecasts: the official pattern specifies only 15 marks for General Aptitude and 85 for DA technical subjects combined, not a linear algebra allocation.
| Topic | Relevance for ML | Skip from Strang? |
|---|---|---|
| Vector spaces, subspaces, linear independence, basis, dimension | Foundational | No |
| Matrices — rank, inverse, determinant, trace | Foundational | No |
| Systems of linear equations, Gaussian elimination | Medium | No |
| LU decomposition | Medium | No |
| Projection matrices, orthogonal matrices | High (PCA, regression) | No |
| Idempotent and partition matrices | Medium | No |
| Quadratic forms | Medium | No |
| Eigenvalues, eigenvectors, diagonalisation | High | No |
| Singular Value Decomposition (SVD) | High | No |
| Numerical LA (iterative solvers) | Out of syllabus | Yes |
| Jordan normal form | Out of syllabus | Yes |
Book and Resources
| Resource | Role | Use for | Skip |
|---|---|---|---|
| Gilbert Strang — Introduction to Linear Algebra | Primary book | Concepts, geometric intuition, all syllabus topics | Specialist iterative solvers beyond the listed LU/SVD and matrix topics |
| MIT 18.06 video lectures (Gilbert Strang) | Free supplement | SVD, eigenvalues, projections — visual intuition | Engineering-application lectures outside syllabus |
| Official GATE DA PYQs (2024 onwards) | Practice | Pattern recognition, question style | — |
| The ML Hub mentor notes | Bridge material | Matrix-zoo summary, SVD intuition for GATE-style problems | — |
Full book list across all subjects: GATE DA books and resources guide.
Deep Dive by Topic
Vector Spaces, Subspaces, Basis, Dimension
Start with Strang's geometric framing — "the four fundamental subspaces" — which is more intuitive than the abstract algebraic approach. Solve relevant chapter exercises; these concepts support linear systems, rank, projections and decompositions.
The Matrix Zoo — Projection, Orthogonal, Idempotent, Partition
Build a reference card during your first pass. Definitions, properties, and how each type appears in past questions:
- Projection matrix — P² = P, so eigenvalues are 0 or 1. An orthogonal projection is also symmetric; a general oblique projection need not be. Orthogonal projections connect to least-squares regression and PCA.
- Orthogonal matrix — QTQ = I, columns are orthonormal, preserves length and angle.
- Idempotent matrix — A² = A, eigenvalues 0 or 1. It represents a projection onto its range; symmetry is additionally required for an orthogonal projection.
- Partition (block) matrix — operations using block structure; inverse and determinant formulas for block matrices.
Most GATE-style questions ask you to identify the type from properties or compute something using a defining property.
Linear Systems, Determinant, Rank, Nullity
Gaussian elimination is the workhorse. Be able to identify whether a system has no solution, a unique solution, or infinitely many — using rank of the coefficient matrix vs rank of the augmented matrix. Rank-nullity connects the listed rank and nullity topics.
LU Decomposition
Practise computing L and U from a given matrix; L stores the row-operation multipliers from Gaussian elimination.
Quadratic Forms
Write a quadratic form as xTAx, classify it as positive definite / positive semi-definite / negative definite / indefinite using eigenvalues, and link definiteness to convexity in optimisation.
Eigenvalues, Eigenvectors, Diagonalisation
Eigenvalues and eigenvectors are explicitly listed. Useful supporting tools include: characteristic polynomial, computation of eigenvalues and eigenvectors, algebraic vs geometric multiplicity, diagonalisability conditions, trace/determinant connection.
Singular Value Decomposition
The topic students most often skip — and the one the syllabus explicitly names. Watch the MIT 18.06 SVD lectures twice. Key facts: any matrix A = UΣVT; U and V are orthogonal; Σ is diagonal with singular values; singular values are square roots of eigenvalues of ATA; SVD always exists, even for non-square matrices.
SVD bridges directly into machine learning — PCA is essentially SVD applied to centred data.
Inner Products and Orthogonality
Inner product, orthogonal complement, Gram-Schmidt process, orthogonal projection onto a subspace. These tie projections, least squares, and SVD together.
Practice Patterns for Official Topics
Use these exercise categories alongside official PYQs; this is a practice checklist, not a frequency analysis:
- Eigenvalue / eigenvector computation for a specific matrix
- Identify the matrix type from given properties (matrix zoo)
- Rank / nullity / solvability of linear systems
- SVD-related conceptual MSQs (relationship between singular values and eigenvalues, properties of U and V)
Solve every LA question from the official GATE DA 2024 and 2025 papers untimed, then re-solve timed. Tag each by sub-topic.
Free benchmark
The linear algebra module in our free GATE DA demo course includes a topic-test — no payment required.
What Is Not in the Syllabus
LA has a lot of beautiful material that falls outside the GATE DA boundary:
- Numerical LA — iterative solvers (Jacobi, Gauss-Seidel, conjugate gradient), QR algorithm
- Jordan normal form, Schur decomposition
- Pseudoinverse beyond its SVD definition
- Specialist theory of Hermitian matrices and complex vector spaces beyond what is needed for the listed matrix topics
- Tensor algebra, multilinear algebra
Use the official topics to set depth, not a literal chapter-title match. Basis, dimension, inner products, Gram-Schmidt and diagonalisation can be useful supporting methods for vector spaces, projections and SVD without becoming separate official syllabus headings.
Study Sequence
Weeks 1–3: Foundations
- Vector spaces, subspaces, basis, dimension — Strang chapters 1–3.
- Matrix algebra, systems of linear equations, Gaussian elimination, LU decomposition.
- Rank, nullity, determinant, the four fundamental subspaces.
Weeks 4–6: Eigenvalues, SVD, and the Matrix Zoo
- Eigenvalues, eigenvectors, diagonalisation.
- Projection, orthogonal, idempotent and partition matrices — build the matrix-zoo card.
- Inner products, orthogonality, Gram-Schmidt.
- SVD — book + MIT 18.06 lectures.
- Quadratic forms and definiteness.
Weeks 7–8: PYQs and Revision
- Every linear algebra PYQ from GATE DA 2024 and 2025.
- 2–3 topic-wise tests under exam conditions from The ML Hub GATE DA test series.
- Finalise the matrix-zoo card and SVD-properties sheet.
Traps to Avoid
- Reusing GATE CS LA notes as-is. They miss projection / orthogonal / idempotent / partition matrices, quadratic forms, and SVD.
- Confusing eigenvalues and singular values. Singular values of A are square roots of eigenvalues of ATA; for symmetric PSD matrices they are equal.
- Skipping SVD. It is in the official syllabus. The MIT 18.06 lectures make it tractable.
- Mixing up algebraic and geometric multiplicity. Diagonalisability depends on geometric multiplicity matching algebraic.
- Over-reading numerical methods. Iterative solvers and Jordan normal form are out of scope.
The ML Hub's LA Module
The linear algebra block in The ML Hub's GATE DA course covers the official GATE DA 2027 syllabus with separate sessions on the matrix zoo and SVD intuition for GATE-style problems. Mentor-led notes bridge Strang's textbook to the specific question patterns seen in DA 2024 and 2025. Topic-wise tests let you benchmark sub-topics, and ranker journeys show how AIR 9 and AIR 6 candidates used this material.
Get the matrix zoo and SVD right
Linear algebra is a useful prerequisite: the same matrix machinery reappears in ML and data analysis. This is a reason for the study order, not a claim about marks.
- Mentor-led lectures on every LA topic, including the matrix zoo and SVD
- Topic-wise eigenvalue, SVD, and matrix-type tests in the test series
- Matrix-zoo reference card from GATE DA rankers
FAQs
Is GATE DA linear algebra the same as GATE CS?
No — DA is broader. It explicitly adds projection, orthogonal, idempotent and partition matrices, quadratic forms, and SVD. CS notes will have gaps.
Which book for linear algebra GATE DA?
Gilbert Strang's Introduction to Linear Algebra, paired with the MIT 18.06 video lectures. The book maps cleanly to the GATE DA syllabus; the lectures make SVD and eigenvalues intuitive.
What is SVD in GATE DA?
The factorisation A = UΣVT where U and V are orthogonal and Σ is diagonal with singular values. It is explicitly in the GATE DA 2027 syllabus and appears in past papers as conceptual MSQs.
Is LU decomposition in the GATE DA syllabus?
Yes. Practise computing L and U and understand how LU relates to Gaussian elimination.
Related Guides
LA connects to several other subjects: Probability & Statistics (joint distributions, covariance), Calculus & Optimization, and Machine Learning (PCA, SVM, neural networks all sit on linear algebra). For the full subject list, see GATE DA books and resources and the GATE DA syllabus 2027 guide. For a structured month-by-month plan, see How to Prepare for GATE DA in 8 Months.