Chapter 9 — Eigenvalues and Canonical Forms
Change the basis and a matrix changes; its eigenvalues do not. This chapter finds the directions an operator only stretches, the polynomial identities it obeys, and the simplest matrix it can be given — diagonal when possible, triangular always over $\mathbb{C}$, and the Jordan form in general.
Multiply a vector by the same matrix again and again — the Fibonacci recursion, a Markov chain of market shares, an age-structured population, the power method that ranks web pages — and something remarkable happens: whatever the starting vector, the direction settles down, and each step multiplies the length by nearly the same number. The limiting direction is an eigenvector and the factor is the largest eigenvalue. Eigenvalues also decide whether solutions of $x'=Ax$ decay or explode, give the natural frequencies of a bridge or a molecule, and, in quantum mechanics, are the values a measurement can return. They are the part of a linear operator that no change of basis can hide.
Chapter 8 showed that one operator has many matrices, all similar to one another. This chapter asks for the best one. The characteristic polynomial collects the eigenvalues; the Cayley–Hamilton theorem says every matrix satisfies it; the minimal polynomial records exactly which polynomial identities hold. Its factorisation decides everything that follows: distinct linear factors mean the matrix is diagonalisable; linear factors with repetitions still give a triangular form (even a unitary one, by Schur's theorem) and, refining it, the Jordan canonical form, which classifies matrices up to similarity completely. For CSIR NET this is the most examined block of linear algebra.
- compute eigenvalues, eigenspaces, algebraic and geometric multiplicities, and the characteristic polynomial from traces and principal minors;
- prove the Cayley–Hamilton theorem and use it to compute inverses, powers and polynomial expressions in a matrix;
- find the minimal polynomial and use it, together with spectral mapping and the $AB$/$BA$ theorem, to answer questions about $\det$, $\operatorname{tr}$ and eigenvalues of matrices built from others;
- decide whether a matrix is diagonalisable over $\mathbb{R}$, $\mathbb{C}$ or a finite field, and diagonalise it;
- use invariant subspaces, the primary decomposition and Schur's theorem, and diagonalise or triangularise commuting matrices simultaneously;
- compute the Jordan form from ranks of powers of $A-\lambda I$ and count similarity classes with given characteristic and minimal polynomials.
How this is examined
Every CSIR NET paper has several questions from this chapter. In Part B the typical question gives a matrix with a pattern — all entries equal, a rank-one perturbation of the identity, a triangular or block matrix, a matrix satisfying $A^2=A$ or $A^3=A$, the matrix of a map such as $X\mapsto X^T$ or $p\mapsto p'$ — and asks for its eigenvalues, its trace or determinant, the determinant of a polynomial in it, its minimal polynomial, or the rank of $A-\lambda I$; data such as $\operatorname{tr}A$, $\operatorname{tr}A^2$ and $\det A$ often have to be turned into the characteristic polynomial first. Part C, where several options may be correct, prefers statements to be judged: "$AB$ and $BA$ have the same minimal polynomial", "if $A^k=I$ then $A$ is diagonalisable over $\mathbb{R}$", "two matrices with the same characteristic and minimal polynomials are similar", "commuting matrices have a common eigenvector", and the counting of similarity classes or Jordan forms with prescribed invariants. Every false option is one of the counterexamples collected in the ⚠️ boxes of the three sections. GATE MA asks the same material in single-answer and multiple-select form.
Sections in this chapter
Eigenvalues, the Minimal Polynomial and Cayley–Hamilton
Eigenvalues and eigenspaces, the characteristic polynomial and principal minors, algebraic and geometric multiplicity, Cayley–Hamilton with proof, the minimal polynomial, $f(A)$, $AB$ versus $BA$, companion matrices.
Diagonal and Triangular Forms
Diagonalisability criteria, the minimal-polynomial test, invariant subspaces, the primary decomposition, triangularisation and Schur's theorem, simultaneous diagonalisation of commuting matrices.
The Jordan Canonical Form
Generalised eigenvectors, Jordan blocks, existence and uniqueness, computing the Jordan form from ranks of powers, nilpotent matrices, counting similarity classes, the rational canonical form.
This chapter builds directly on Chapter 8. You need bases, dimension, direct sums and quotient spaces (§8.1); linear maps, kernels and images, rank–nullity, matrices of linear maps, change of basis and similarity, projections and nilpotent maps (§8.2); and determinants with the product rule, the adjugate and the rank of a matrix (§8.3). From algebra you need the division algorithm for polynomials, greatest common divisors and Bézout's identity $\gcd(f,g)=uf+vg$, and the fact that every complex polynomial is a product of linear factors (the fundamental theorem of algebra, proved in complex analysis — CSIR Unit 2). Unitary matrices and Gram–Schmidt are used once, in Schur's theorem, and are developed fully in Chapter 10.