Skip to main content

Matrix Operations Cheat Sheet

Quick reference for every matrix operation — formulas, rules, and worked examples.

📖 6 min read · Updated May 2026 · Reviewed by our math editorial team

Matrix Addition & Subtraction

Rule

Add/subtract entry-by-entry. Matrices must be the same size (m×n).

(A + B)[i][j] = A[i][j] + B[i][j]

[1 2] + [5 6] = [6  8]
[3 4]   [7 8]   [10 12]

Scalar Multiplication

(kA)[i][j] = k · A[i][j] — multiply every entry by k.

Properties: (k+l)A = kA + lA, k(A+B) = kA + kB

3 · [1 2] = [3  6]
    [3 4]   [9 12]

Matrix Multiplication → Calculator

C = AB: C[i][j] = Σₖ A[i][k]·B[k][j] (dot product of row i of A with column j of B)

Requires: cols(A) = rows(B). Result: (rows(A)) × (cols(B)).

⚠ Not commutative: AB ≠ BA in general

Associative: A(BC) = (AB)C | (AB)ᵀ = BᵀAᵀ | (AB)⁻¹ = B⁻¹A⁻¹

[1 2] · [5 6] = [1·5+2·7  1·6+2·8]
[3 4]   [7 8]   [3·5+4·7  3·6+4·8]
              = [19 22]
                [43 50]

Transpose → Calculator

Aᵀ[i][j] = A[j][i] — swap rows and columns. m×n becomes n×m.

Properties: (Aᵀ)ᵀ = A | (AB)ᵀ = BᵀAᵀ | (A+B)ᵀ = Aᵀ+Bᵀ | det(Aᵀ)=det(A)

Symmetric: A = Aᵀ | Skew-symmetric: Aᵀ = −A | Orthogonal: Aᵀ = A⁻¹

A  = [1 2 3]   Aᵀ = [1 4]
     [4 5 6]        [2 5]
                    [3 6]

Determinant → Calculator

2×2: det([[a,b],[c,d]]) = ad − bc

3×3: Cofactor expansion along row 1: a(ei−fh) − b(di−fg) + c(dh−eg)

Row ops: swap rows → ×(−1) | scale row → ×c | add multiple → no change

det(AB) = det(A)·det(B) | det(A⁻¹) = 1/det(A) | det(kA) = kⁿ·det(A)

det([3 -2]) = 3·4−(−2)·1 = 12+2 = 14
    [1  4]

Triangular matrix: det = product of diagonal

Matrix Inverse → Calculator

A · A⁻¹ = A⁻¹ · A = I. Exists iff det(A) ≠ 0.

2×2 formula: A⁻¹ = (1/det) · [[d,−b],[−c,a]]

General: Gauss-Jordan on [A|I] → [I|A⁻¹]

Properties: (AB)⁻¹=B⁻¹A⁻¹ | (Aᵀ)⁻¹=(A⁻¹)ᵀ | (A⁻¹)⁻¹=A

A = [4 7]   det = 4·6−7·2 = 10
    [2 6]

A⁻¹ = (1/10)[6 -7] = [3/5  -7/10]
             [-2 4]   [-1/5  2/5 ]

RREF and Row Operations → Calculator

Elementary row operations: (1) swap rows R_i ↔ R_j, (2) scale row R_i → c·R_i (c≠0), (3) replace R_i → R_i + c·R_j

RREF conditions: (1) zero rows at bottom, (2) leading entries = 1 (pivots), (3) staircase pattern, (4) zeros above AND below each pivot

Rank: number of pivots in RREF. Rank(A) = rank(Aᵀ). For m×n: rank ≤ min(m,n).

Null space / Nullity: nullity = n − rank (rank-nullity theorem). Free variables correspond to null space dimensions.

Eigenvalues & Eigenvectors

Definition: Av = λv, v ≠ 0. λ is the eigenvalue; v is the eigenvector.

Finding eigenvalues: Solve characteristic equation det(A − λI) = 0. This is a degree-n polynomial in λ.

Finding eigenvectors: For each eigenvalue λ, solve (A − λI)v = 0 (null space of A − λI).

Trace = sum of eigenvalues. Det = product of eigenvalues.

Diagonalizable: A = PDP⁻¹ where D = diagonal matrix of eigenvalues, P = matrix of eigenvectors (columns).

Dot & Cross Products

Dot Product

u · v = Σᵢ uᵢvᵢ = |u||v|cos(θ)

u · v = 0 ↔ orthogonal

u · u = |u|² (squared length)

Cross Product (3D only)

u × v = det([i j k; u; v])

|u × v| = |u||v|sin(θ) = area of parallelogram

u × v = −(v × u) (anti-commutative)

Special Matrices

Identity I

Diagonal entries = 1, all others 0. AI = IA = A.

Zero matrix 0

All entries = 0. A + 0 = A, A·0 = 0.

Diagonal

Non-zero entries only on main diagonal. Easy to invert: d⁻¹ on diagonal.

Symmetric

A = Aᵀ. Real eigenvalues, orthogonal eigenvectors (spectral theorem).

Orthogonal Q

Qᵀ = Q⁻¹, Qᵀ Q = I. Preserves lengths and angles (rotations, reflections).

Upper/Lower triangular

All entries below/above main diagonal = 0. det = product of diagonal.

Non-Commutativity: A Worked Counter-Example

AB ≠ BA in general. Here is a concrete pair of matrices where both products exist but are different:

AB:

A = [1 2]   B = [0 1]
    [3 4]       [1 0]

AB = [1·0+2·1  1·1+2·0] = [2 1]
     [3·0+4·1  3·1+4·0]   [4 3]

BA:

BA = [0·1+1·3  0·2+1·4] = [3 4]
     [1·1+0·3  1·2+0·4]   [1 2]

AB ≠ BA — the products have different entries. This is the general rule: matrix multiplication is non-commutative unless A and B have special structure (e.g., both are diagonal, or one is a scalar multiple of I). When solving Ax = b, never "divide by A" from the right — always left-multiply by A⁻¹.

Key Properties Summary Table

PropertyHolds?Exception / Note
A + B = B + A✓ AlwaysAddition is commutative
AB = BA✗ Usually notOnly if both diagonal, or special cases
A(BC) = (AB)C✓ AlwaysMultiplication is associative
(AB)ᵀ = BᵀAᵀ✓ AlwaysOrder reverses on transpose
(AB)⁻¹ = B⁻¹A⁻¹✓ If both invertibleOrder reverses on inverse
det(AB) = det(A)det(B)✓ AlwaysMultiplicativity of determinant
det(A + B) = det(A) + det(B)✗ FalseDet is not additive
rank(AB) ≤ min(rank A, rank B)✓ AlwaysRank can only decrease under multiplication
(Aᵀ)⁻¹ = (A⁻¹)ᵀ✓ If A invertibleInverse and transpose commute

Connection to RREF → Calculator

RREF ties together all major matrix operations. Here is how each operation connects to row reduction:

  • Solving Ax = b: Form augmented matrix [A|b] and compute RREF. Read solution from last column.
  • Finding A⁻¹: Augment [A|I] and compute RREF. If left half becomes I, right half is A⁻¹.
  • Computing det(A): Row reduce to upper triangular form, tracking determinant changes: swaps multiply det by −1, scalings multiply det by c.
  • Finding rank: Compute RREF and count pivot columns. This is the most reliable method.
  • Finding null space: Compute RREF of A, identify free variables, express basic variables in terms of free variables.
  • Finding column space basis: Compute RREF, identify pivot columns, return the corresponding original columns of A (not RREF columns).
  • Testing linear independence: Form matrix with vectors as columns, compute RREF. Linearly independent ↔ every column is a pivot column.

Row reduction is the universal algorithm. Every operation above reduces to RREF with the right setup. See the linear algebra basics guide for the conceptual foundations.