Course Summary

Linear algebra starts from vectors and linear combinations and climbs towards the structure of a vector space itself. The course first builds the geometric intuition that a matrix is a linear transformation, then works through determinants, matrix operations, the rank of a matrix, inverse matrices, the structure of solution spaces, and systems of linear equations; it then describes the skeleton of a transformation with inner products, eigenvalues and eigenvectors, and finally rises to change of basis, abstract vector spaces, and quadratic forms. Looking back over the whole course, one thread runs through every chapter: start from concrete vectors and transformations, abstract to the structure of a linear space, and let eigenvalues and quadratic forms answer what the transformation stretches and rotates the space into.

1. The Knowledge System of the Course

TopicCore contentKey methods
Vectors and Linear CombinationsDefinition of a vector, addition and scalar multiplication, linear combinations, span, linear dependence and independence, bases and dimensionDraw the geometry to judge whether a set of vectors can span the whole space
Linear Transformations and MatricesA matrix as the table of a transformation, the same transformation represented in a new basis, similar matricesTrack where the basis vectors go; convert basis to standard basis and back with BB, AA, B−1B^{-1} to obtain A′=B−1ABA' = B^{-1}AB
DeterminantsDefinition of the determinant, the six basic properties, expansion by a row or a column, the geometric meaning as a volume-scaling factorRow operations to simplify, the expansion method, the property method
Matrices and Their OperationsAddition, scalar multiplication, multiplication, transpose, elementary matrices, block matrices, the rank of a matrixElementary row transformations to row echelon form, then read the rank off the number of nonzero rows
Inverse Matrices and Solution SpacesThe inverse as the reverse transformation, the adjugate matrix, the dimension of the solution space, the rank conditions for one solution, infinitely many solutions, or no solutionA−1=1∣A∣A∗A^{-1} = \frac{1}{\lvert A\rvert}A^*; the solution space of Ax⃗=0⃗A\vec{x} = \vec{0} has dimension n−r(A)n - r(A)
Systems of Linear EquationsHomogeneous and non-homogeneous systems, Gaussian and Gauss-Jordan elimination, the structure of the solution set, Cramer’s ruleElementary row transformations; general solution = one particular solution + the general homogeneous solution
Inner Products and OrthogonalisationThe inner product, the length and angle it defines, orthogonal vectors, Gram-Schmidt orthogonalisation, orthogonal matricesUse the inner product to test orthogonality; run Gram-Schmidt to build an orthogonal basis
Eigenvalues and EigenvectorsAx⃗=λx⃗A\vec{x} = \lambda\vec{x}, the characteristic equation, the trace and determinant relations, real symmetric matricesSolve ∣A−λI∣=0\lvert A - \lambda I\rvert = 0; check the conditions for diagonalisability
Change of Basis and Abstract Vector SpacesThe change-of-basis matrix BB, coordinates relative to a basis, abstract vector spaces, the spaces of polynomials and of matricesThe columns of BB are the new basis vectors in standard coordinates; [v]B=B−1v[\mathbf{v}]_\mathcal{B} = B^{-1}\mathbf{v}
Quadratic FormsThe matrix of a quadratic form, standard forms, congruence transformations, the law of inertia, positive definitenessCompleting the square or a congruence transformation to reach a standard form; the leading principal minors decide positive definiteness

2. Core Conclusions on Determinants, Eigenvalues, and Quadratic Forms

These three topics are where linear algebra computation and proof problems land most often, so the conditions below have to be remembered exactly:

TopicCore resultHow to use it
Determinant propertiesA zero row or column gives D=0D = 0; two proportional rows or columns give D=0D = 0; adding kk times one row or column to another leaves DD unchangedD≠0D \neq 0 is the condition for AA to be invertible and for r(A)=nr(A) = n
The rank of a matrixr(A)r(A) is the order of the highest-order nonzero minor, and also the number of nonzero rows in row echelon formr(A)=n  ⟺  Ar(A) = n \iff A is invertible   ⟺  \iff the rows and columns of AA are linearly independent
The characteristic equationEigenvalues come from ∣A−λI∣=0\lvert A - \lambda I\rvert = 0; substitute each eigenvalue back into (A−λI)x⃗=0(A - \lambda I)\vec{x} = 0 for the eigenvectors∑λi=tr⁡(A)\sum \lambda_i = \operatorname{tr}(A) and ∏λi=∣A∣\prod \lambda_i = \lvert A\rvert make a quick check
Similarity diagonalisationAA is diagonalisable   ⟺  \iff it has nn linearly independent eigenvectors; nn distinct eigenvalues guarantee diagonalisabilityA real symmetric matrix is always orthogonally diagonalisable, and all of its eigenvalues are real
Testing positive definitenessA quadratic form is positive definite   ⟺  \iff all leading principal minors are positive, and equivalently   ⟺  \iff all eigenvalues are positiveThe positive index of inertia is then nn and the negative index is 00; the law of inertia fixes these indices under congruence

Key takeaway: whether the determinant DD is zero threads “invertible / full rank / a unique solution to the system” into one line; an eigenvalue is the scaling factor along a direction the transformation leaves unchanged, and similarity diagonalisation is the search for a basis in which the transformation only stretches and never rotates; the quickest test for positive definiteness is the leading principal minors, with the all-positive-eigenvalue criterion as an equivalent.

3. How to Solve Problems

When a linear algebra problem turns up, choose the method along the following path:

  1. Classify the problem first: is it a determinant computation, a rank or inverse computation, a system of linear equations, an eigenvalue problem, or a quadratic form? Name the object first, then pick the tool.
  2. Start with row operations for determinants and rank: reduce to row echelon form with elementary row transformations, record the determinant multiplier at every step, and read the rank off the number of nonzero rows instead of expanding a high-order determinant by hand.
  3. Keep the structure of the solution set in mind: for a homogeneous system find a fundamental system of solutions; for a non-homogeneous system find one particular solution first and write the general solution as that particular solution plus the general homogeneous solution.
  4. Follow the fixed steps for eigenvalues and diagonalisation: compute the characteristic equation to get λ\lambda, then find the eigenvectors one by one, build PP and DD, and check A=PDP−1A = PDP^{-1}; a real symmetric matrix can always be orthogonally diagonalised.
  5. Write the matrix of a quadratic form before judging it: the diagonal entries of that matrix are the coefficients of the squared terms and the off-diagonal entries are half the coefficients of the mixed terms; reduce to a standard form by completing the square or by a congruence transformation, and decide positive definiteness from the leading principal minors.

Course Roadmap

  1. 1

    Exploring Functions in Advanced Mathematics

    Prerequisite

    Functions are a core idea of advanced mathematics. This course walks through foundational concepts, key properties, and classic constants so you can read, reason, and compute with confidence.

    Go to course
  2. 2

    Vector Algebra and Analytic Geometry of Space

    Prerequisite

    Master vector operations and the equations of points, lines, and planes in space, together with their mutual relationships.

    Go to course
  3. 3

    Linear Algebra

    Current course

    Master determinants, matrices, vectors, systems of linear equations, eigenvalues, and quadratic forms, and understand the abstract structure of linear spaces.

    Go to course

More courses

Recommended courses

Found an issue with this content?

Tell us what could be improved.