Dual Spaces Intro: Concepts, Formulas and Examples

The dual space V* of a vector space V is the collection of all linear functionals on V: linear maps that take a vector and return a single number. If vectors are the objects you measure, dual vectors are the measuring instruments, each one extracting one linear quantity from any vector.

Gradients, row vectors, and evaluation maps are all dual vectors in disguise. Each linear equation is a functional set to zero, and quantum bra vectors live in dual spaces.

Key Concepts

• Linear functional. f: V → F with f(au + bv) = af(u) + bf(v); e.g. f(x, y) = 3x − y on R².
• Dual space. V* = {all linear functionals on V}, a vector space itself.
• Dual basis. fi(ej) = δij (1 if i = j, 0 otherwise).
• Dimension. For finite-dimensional V, dim V* = dim V.
• Double dual. V** ≅ V canonically via evaluation, f ↦ f(v).

Formulas

• Dual basis condition: fi(ej) = δij determines each fi completely.
• Coordinate extraction: if v = Σ vi ei then fi(v) = vi.
• Transpose map: T: V → W induces T*: W* → V* by (T*g)(v) = g(Tv).
• Assumptions and limits: needs finite dimension; for infinite-dimensional V the dual is strictly larger.

Worked Example

Find the dual basis for the basis e1 = (1, 1), e2 = (1, −1) of R².

1. Write the unknown functionals as f1(x, y) = ax + by and f2(x, y) = cx + dy.
2. Apply the dual condition to f1: f1(e1) = a + b = 1 and f1(e2) = a − b = 0. Adding gives 2a = 1, so a = 1/2, then b = 1/2.
3. Apply it to f2: f2(e1) = c + d = 0 and f2(e2) = c − d = 1. Adding gives 2c = 1, so c = 1/2, then d = −1/2.
4. Result: f1(x, y) = (x + y)/2 and f2(x, y) = (x − y)/2.
5. Verify on a test vector, say v = (3, 1) = 2e1 + e2: f1(v) = (3 + 1)/2 = 2 and f2(v) = (3 − 1)/2 = 1, exactly the coordinates of v in the (e1, e2) basis. Correct.

Common Misconceptions

• Treating dual vectors as ordinary vectors. Without an inner product there is no natural way to view a functional as a vector of V.
• Assuming the dual basis is trivial. fi(ej) = δij must be solved; only orthonormal bases are self-dual.
• Thinking V and V* are canonically the same. The isomorphism needs a basis or inner product; only V ≅ V** is natural.

Real-World Applications

• Quantum mechanics: bra vectors ⟨ψ| are dual vectors pairing with ket vectors |φ⟩ to give probability amplitudes ⟨ψ|φ⟩.
• Optimization: Lagrange multipliers are dual variables; each constraint is a functional, and the multipliers live in the dual space.
• Differential geometry: differential 1-forms like dx are dual to tangent vectors, and integration pairs forms with curves.

The dual space V* collects all linear scalar-valued measurements on V, with a dual basis fi defined by fi(ej) = δij that reads off coordinates. Gradients, row vectors, bras, and Lagrange multipliers are all dual vectors, making V* one of the most quietly useful constructions in linear algebra.