Chapter 3 — Linear Algebra
Vector spaces and linear maps, the matrices that represent them and the rank that decides every linear system; eigenvalues and the search for the simplest matrix of an operator; and the geometry of inner products, special matrices and quadratic forms — the language in which almost every other chapter of GATE MA is written.
A search engine ranks billions of web pages by finding one vector — the eigenvector of a huge matrix of links. An engineer checks that a bridge will not resonate with the wind by computing the eigenvalues of its stiffness matrix. A phone compresses a photograph by keeping only the largest few "singular values" of the matrix of pixels, and a statistician finds the main directions of variation in a cloud of data from the eigenvectors of a symmetric matrix. Behind all of these is one subject. Linear algebra studies spaces in which vectors can be added and scaled, maps that respect those operations, and the numbers — dimension, rank, eigenvalues — that survive every change of coordinates.
This chapter develops the subject in three steps. Section 3.1 sets up the language: vector spaces, bases and dimension, linear maps and the rank–nullity theorem, the matrix of a map in a basis and how it changes when the basis changes, and the rank conditions that decide whether a linear system has no solution, one, or infinitely many. Section 3.2 looks for the basis in which a given operator is simplest: eigenvalues and eigenvectors, the characteristic and minimal polynomials, the Cayley–Hamilton theorem, the exact tests for diagonalisability and, when diagonalisation fails, the Jordan canonical form read off from ranks. Section 3.3 adds lengths and angles: inner products and the Gram–Schmidt process, the symmetric, Hermitian, normal, orthogonal and unitary matrices and their eigenvalues, unitary diagonalisation, and quadratic forms with their signature and definiteness. The results are used everywhere else in this course — the Hessian test of §1.2, linear systems of differential equations in Chapter 6, Hilbert spaces in Chapter 8, numerical linear algebra in Chapter 9 and the simplex method in Chapter 12.
- decide whether a set is a subspace, compute the dimension of spaces of vectors, matrices and polynomials cut out by conditions, and count vectors, bases and subspaces over a finite field;
- find the kernel, image, rank and nullity of a linear map, use rank–nullity to prove or disprove the existence of maps, and write the matrix of a map in any basis and change it by $P^{-1}AP$;
- decide by ranks whether a system $Ax=b$ (with parameters) has no solution, a unique solution or infinitely many, and use the rank inequalities in "which of these is true" questions;
- compute eigenvalues, eigenspaces and characteristic polynomials quickly (traces, minors, rank-one matrices, spectral mapping), and apply the Cayley–Hamilton theorem to inverses and powers;
- find minimal polynomials, decide diagonalisability with the exact tests, find the Jordan form from rank data and count similarity classes;
- orthonormalise with Gram–Schmidt, use the eigenvalue facts for symmetric, skew-symmetric, Hermitian, normal, orthogonal and unitary matrices, diagonalise normal matrices unitarily, and classify quadratic forms by signature and definiteness.
How this is examined
Linear algebra is one of the most heavily weighted parts of GATE MA, and it appears in all three formats. MCQs are short computations: the dimension of a subspace of matrices or polynomials, the rank of a structured matrix, the values of a parameter for which a system has infinitely many solutions, a determinant or trace computed from eigenvalues, a minimal polynomial, the signature of a quadratic form. MSQs list statements about ranks, eigenvalues, diagonalisability, Jordan forms or special matrices and ask which are always true — and the false ones are exactly the standard counterexamples collected in the ⚠️ boxes (a $2\times2$ Jordan block, a rotation, a pair of non-commuting matrices). NATs ask for a number: a nullity, a count of subspaces over a finite field, a trace of an inverse, the number of similarity classes with a given characteristic and minimal polynomial, or a value of a quadratic form at an optimum. The same material is examined in IIT JAM and, with more proofs, in CSIR NET Part C.
Sections in this chapter
Vector Spaces, Linear Maps and Linear Systems
Subspaces, bases and dimension, counting over finite fields, linear maps and rank–nullity, matrices and change of basis, row reduction, rank rules and the Rouché–Capelli theorem.
Eigenvalues, Diagonalisation, Minimal Polynomial and Jordan Form
Characteristic and minimal polynomials, Cayley–Hamilton, the tests for diagonalisability, the Jordan form from ranks, counting similarity classes.
Inner Products, Special Matrices and Quadratic Forms
Inner products and Gram–Schmidt, symmetric, Hermitian, normal, orthogonal and unitary matrices, unitary diagonalisation, bilinear and quadratic forms, signature and definiteness.
You need matrix algebra at BSc level — matrix multiplication, transposes, inverses of $2\times2$ matrices — together with $2\times2$ and $3\times3$ determinants (cofactor expansion, the effect of row operations), solving small linear systems by elimination, complex numbers and their conjugates, and polynomials and their roots. Nothing from Chapters 1 and 2 is required, although §1.2 already used the definiteness of a symmetric matrix, which §3.3 explains in full (see §1.2). The results of this chapter are used in the linear systems of §6.4, the Hilbert spaces of Chapter 8, the direct and iterative methods of §9.1 and the simplex method of §12.1. If this is your first course of this kind, read the welcome page for how the sections are organised and how the 🎯 GATE corner works.