Matrices: Operations, Inverses, Determinants

A matrix is a rectangular array of numbers; matrices can be added, scaled, and multiplied, and square matrices may possess an inverse (an "undo" matrix) and a single number called the determinant that detects invertibility. Key result: a square matrix is invertible if and only if its determinant is nonzero, and the inverse exists uniquely when it does. Used everywhere systems of linear equations appear — graphics, physics, statistics, economics. Formalized by Cayley, Sylvester, and Cramer between the 18th and 19th centuries.

A matrix is a rectangular array of numbers arranged in rows and columns, written with brackets, and serving as a compact bookkeeping device for linear relationships between quantities. Matrices matter because nearly every system of linear equations, geometric transformation, or coupled set of variables can be encoded as a matrix and manipulated with a small algebra of operations. We can add matrices of the same shape, multiply a matrix by a scalar, and — crucially — multiply two matrices when their inner dimensions match. For square matrices, the determinant is a single number computed from the entries, and the inverse is the matrix that reverses the original's action. The surprising fact: matrix multiplication is not commutative — $AB$ and $BA$ are usually different matrices, and can even have different sizes, which makes matrices behave unlike ordinary numbers.

The Core Insight

A matrix is best understood not as a static grid of numbers but as a transformation — a rule that turns input vectors into output vectors. Multiplication composes transformations, the determinant measures how much the transformation scales area or volume (and whether it flips orientation), and the inverse is the transformation that undoes the original. A zero determinant means the transformation collapses space (squashing a region to zero area), which is exactly why no inverse can exist — you cannot un-squash what has been flattened.

Formal Statement

Matrix

An $m\times n$ matrix $A$ is a rectangular array of numbers with $m$ rows and $n$ columns. The entry in row $i$, column $j$ is written $a_{ij}$:

The matrix is square if $m=n$. The main diagonal consists of the entries $a_{ii}$.

Matrix Addition and Scalar Multiplication

For two $m\times n$ matrices $A$ and $B$, and a scalar $c$:

These operations make matrices of a fixed shape behave like vectors: associative, commutative addition, with a zero matrix and additive inverses.

Matrix Multiplication

If $A$ is $m\times n$ and $B$ is $n\times p$ (inner dimensions match), the product $AB$ is the $m\times p$ matrix with entries:

The entry $(AB{)}_{ij}$ is row $i$ of $A$ "marched along" column $j$ of $B$, multiplying paired entries and summing.

Identity Matrix

The $n\times n$ identity matrix $I_n$ has 1's on the main diagonal and 0's elsewhere:

Inverse Matrix

A square $n\times n$ matrix $A$ is invertible (or nonsingular) if there exists a matrix $A^{-1}$ with:

If no such $A^{-1}$ exists, $A$ is called singular.

Determinant — 2×2 and 3×3

For a $2\times 2$ matrix:

For a $3\times 3$ matrix, expand along the first row (cofactor expansion):

Inverse of a 2×2 Matrix

Let $A=\left[\begin{array}{cc}a& b\\ c& d\end{array}\right]$. Then $A$ is invertible if and only if $\det A=ad-bc\ne 0$, and:

Invertibility Criterion

A square matrix $A$ is invertible if and only if $\det A\ne 0$. Moreover, the determinant of a product satisfies:

Why It Holds / Derivation

The 2×2 Inverse Formula

We verify that the claimed inverse [8] works.

Determinant Detects Singularity (sketch)

Arthur Cayley | 1821–1895

The English mathematician Arthur Cayley founded matrix algebra as a subject in its own right. In his 1858 Memoir on the Theory of Matrices, he treated matrices as objects that could be added and multiplied, introduced the inverse, and stated what became the Cayley–Hamilton theorem (every square matrix satisfies its own characteristic equation). Cayley collaborated closely with James Joseph Sylvester, who coined the very word "matrix" (Latin for "womb") in 1850, picturing it as the array from which determinants are born.

Building Intuition

Think of a $2\times 2$ matrix as instructions for transforming a sheet of graph paper. The two columns tell you where the basis arrows $\hat{i}=(1,0)$ and $\hat{j}=(0,1)$ land after the transformation. Once you know where those two arrows go, every point's destination is determined.

The determinant is the area of the parallelogram formed by the two transformed arrows. If that area is 1, the transformation preserves size; if it's 3, areas triple; if it's negative, the plane gets flipped over (orientation reversed); if it's 0, the two arrows became parallel and the whole plane has been crushed onto a single line.

A 2×2 matrix transforms the unit square; the determinant is the area of the resulting parallelogram.

Worked Examples

Multiplying Two Matrices

Determinant, Inverse, and Solving a System

Common Misconceptions

Connections

Matrices are the computational engine behind Systems of Linear Equations — solving $A\mathbf{x}=\mathbf{b}$ is exactly inverting or row-reducing $A$. They generalize the idea of a Vector and provide the language for Linear Transformations, where each matrix is a transformation. The determinant connects to Area and Volume Scaling in geometry and to the Cross Product in three dimensions (which is itself a determinant). At a deeper level, matrices lead into Eigenvalues and Eigenvectors, where one seeks the special directions a matrix merely stretches without rotating.

Real-World Applications

• Computer graphics & video games: Every rotation, scaling, and translation of 3D models uses $4\times 4$ transformation matrices; the GPU multiplies millions per second.
• Google's PageRank: The original web-ranking algorithm finds the dominant eigenvector of a giant matrix encoding hyperlinks between pages.
• Cryptography (Hill cipher): Text is encrypted by multiplying letter-vectors by a matrix; decryption uses the inverse matrix modulo 26.
• Economics (Leontief input–output models): A matrix encodes how each industry's output feeds into others; the inverse $(I-A{)}^{-1}$ predicts total production needed for given demand.
• Engineering (structural analysis): Stiffness matrices relate forces to displacements in bridges and trusses; solving the linear system gives stresses.
• Computer vision & machine learning: Image transformations, neural-network layers, and dimensionality reduction (PCA) are all matrix operations.
• 3D physics & robotics: Rotation matrices and their inverses orient robot arms and spacecraft, converting between coordinate frames.

Historical Development

• c. 200 BCE — China: The Nine Chapters on the Mathematical Art solves simultaneous equations using a grid of coefficients and column operations — essentially Gaussian elimination, two millennia early.
• 1683 — Japan & 1693 — Europe: Seki Takakazu and, independently, Gottfried Leibniz introduce determinants to study systems of equations.
• 1750 — Gabriel Cramer: Publishes Cramer's Rule, expressing solutions of linear systems as ratios of determinants.
• 1812 — Cauchy: Augustin-Louis Cauchy gives the modern definition of determinant and proves the product rule $\det (AB)=\det A\det B$.
• 1850 — James Joseph Sylvester: Coins the term "matrix" for the array generating minors and determinants.
• 1858 — Arthur Cayley: Memoir on the Theory of Matrices establishes matrix algebra — addition, multiplication, inverses — as an independent theory.
• 1879 — Frobenius: Georg Frobenius develops the theory of matrix rank and canonical forms.
• 1925 — Heisenberg & Born: Matrix mechanics formulates quantum theory entirely in terms of (infinite) matrices, giving the abstract subject deep physical meaning.
• 1947–present: With electronic computers, matrix algorithms (LU decomposition, QR, SVD) become the backbone of scientific computing.

Subtleties & Limits

Although the $2\times 2$ and $3\times 3$ determinant and inverse formulas are clean, they explode in size for large matrices: the general "expansion by minors" requires roughly $n!$ operations, hopeless even for $n=20$. Practical computation instead uses Gaussian elimination ($\sim n^3$ operations), but rounding errors in floating-point arithmetic can make a matrix that is mathematically invertible behave as singular — this is the study of numerical conditioning. The threshold between "invertible" and "effectively singular" is genuinely blurry on real computers. A live research frontier asks: what is the fastest possible algorithm for matrix multiplication? The naïve method is $n^3$; Strassen (1969) achieved $\sim n^{2.81}$, and the current record sits near $n^{2.37}$, but whether one can reach the conjectured ideal of essentially $n^2$ remains open.

See Also