Linear Algebra 58 | Complex Vectors and Complex Matrices

TL;DR

Eigenvalue (spectrum) theory fits more naturally when eigenvalues are treated as complex numbers, supported by the Fundamental Theorem of Algebra.

Briefing Cornell Notes

Briefing

Complex linear algebra starts by extending vectors and matrices from real entries to complex entries, mainly to make eigenvalue (spectrum) theory work more naturally. Even when a matrix has only real numbers, its eigenvalues are best treated as a subset of the complex numbers because the Fundamental Theorem of Algebra guarantees that eigenvalue calculations fit cleanly into the complex setting. That shift motivates working with vectors in C^n and matrices with complex entries, written as C^{m×n}, so eigenvalue theory has the right “home” for its answers.

The definitions barely change in form: C^n is the set of column vectors with n components, except each component now comes from the complex numbers. Likewise, C^{m×n} is the set of m-by-n matrices whose entries are complex. The key operational upgrade is that scalar multiplication must allow scalars from C, not just from R. Vector addition still adds component-by-component, and scalar multiplication still scales component-by-component—only now the arithmetic inside each component uses complex addition and complex multiplication.

Once complex scalars are allowed, C^n becomes a complex vector space. The vector-space axioms remain the same in structure: addition is associative and commutative, there is a zero (neutral) vector, and every vector has an additive inverse. Scalar multiplication must be compatible with complex multiplication (so scaling by a product matches successive scaling), must leave vectors unchanged when the scalar is 1, and must distribute over vector addition and over scalar addition. Because these rules match the real case, the usual linear-algebra machinery—subspaces, spans, linear independence, bases, and dimension—carries over by replacing R^n with C^n.

A notable consequence is the dimension of C as a vector space over C: it equals 1. This can feel counterintuitive if complex numbers are pictured as points with two real coordinates, but the “degree of freedom” changes once scaling is permitted by complex numbers. In C viewed as a complex vector space, one basis element is enough: the canonical unit vector basis for C^n still spans the space, and complex scaling lets every vector be built from it.

The most important technical adjustment comes with inner products and norms. If the inner product were defined by simply multiplying corresponding components and summing, complex values could break the connection to lengths and angles. To ensure the norm is real and nonnegative, the inner product must include complex conjugation—specifically, conjugating the entries of the first vector. The standard norm is then defined as the square root of the inner product of a vector with itself, which effectively uses absolute values squared of complex components. A quick example with the vector (i, 1) yields a norm of √2 because |i|^2 = 1, avoiding the negative outcome that would occur from i^2 = −1. With these definitions in place, complex linear algebra proceeds with the same computational spirit as the real case, but with conjugation built into the geometry.

Cornell Notes

The shift from real to complex linear algebra is driven by eigenvalue theory: eigenvalues are naturally treated as complex numbers, even for real matrices. Vectors and matrices are extended by letting components and entries come from C, forming C^n and C^{m×n}. The vector-space axioms stay the same in structure, so most definitions (subspaces, span, linear independence, bases) transfer directly. The major change is geometric: the inner product on complex vectors must use complex conjugation to guarantee that the induced norm is real and nonnegative. As a result, the standard norm uses absolute values squared of components, and C has complex dimension 1 when viewed as a vector space over C.

Why does eigenvalue theory push linear algebra into the complex numbers?

Eigenvalues (the spectrum of a matrix) are best regarded as complex numbers because the Fundamental Theorem of Algebra ensures that polynomial equations—hence characteristic polynomials—have roots in C. That means eigenvalue computations fit cleanly in the complex setting, improving eigenvalue theory even when the original matrix has only real entries.

What changes when moving from R^n to C^n?

The definitions keep the same structure: C^n is still the set of column vectors with n components, but each component is complex. Operations also follow the same component-wise pattern: vector addition adds complex components, and scalar multiplication scales by complex numbers using complex multiplication. The vector-space axioms remain valid with scalars from C.

What makes C^n a complex vector space rather than just a real one?

The scalar multiplication operation allows scalars from C. The axioms require compatibility with complex multiplication (a scalar product matches successive scaling), the unit scalar 1 leaves vectors unchanged, and distributive laws connect scalar multiplication with both vector addition and scalar addition. These rules define a complex vector space.

Why must the complex inner product include complex conjugation?

Without conjugation, the quantity used to define length could become complex or negative, breaking the idea that lengths should be real and nonnegative. Conjugating the entries of the first vector ensures that ⟨u,u⟩ becomes a real nonnegative number, so the norm defined as √⟨u,u⟩ behaves like a length.

How does the norm of a complex vector work in practice?

The standard norm is defined as the square root of the inner product with itself, which effectively sums absolute-value squares of components. For example, for u = (i, 1), |i|^2 = 1, so ⟨u,u⟩ = 1 + 1 = 2 and ||u|| = √2. This avoids the negative result that would come from using i^2 = −1 without absolute values.

Why is the complex dimension of C equal to 1?

When C is treated as a vector space over C, complex scaling is allowed. That means one basis element is enough: any complex number can be obtained by multiplying that basis element by a complex scalar. So dim_C(C) = 1, even though C can be visualized as two real coordinates.

Review Questions

How do the vector-space axioms for C^n differ from those for R^n, and which part stays structurally the same?
What goes wrong if the complex inner product is defined without complex conjugation, and how does conjugation fix it?
Compute the standard norm of a complex vector (a, b) using the rule based on absolute values squared. What conditions ensure the result is real and nonnegative?

Key Points

1
Eigenvalue (spectrum) theory fits more naturally when eigenvalues are treated as complex numbers, supported by the Fundamental Theorem of Algebra.
2
Vectors and matrices extend from real entries to complex entries, forming C^n and C^{m×n} with the same component-wise definitions of addition and multiplication.
3
C^n becomes a complex vector space because scalar multiplication uses scalars from C and satisfies the same vector-space axioms with complex arithmetic.
4
Most linear-algebra concepts—subspaces, spans, linear independence, bases, and dimension—transfer from R^n to C^n by replacing the underlying field.
5
The inner product in complex vector spaces must use complex conjugation to ensure ⟨u,u⟩ is real and nonnegative.
6
The standard norm is defined as √⟨u,u⟩, which effectively sums absolute-value squares of complex components.
7
Viewing C as a vector space over C gives dim_C(C) = 1, reflecting that complex scaling collapses the “two-real-coordinate” intuition.

Highlights

Complex linear algebra is adopted largely to make eigenvalue theory work cleanly in the complex numbers.

The only structural change to vector-space definitions is allowing scalar multiplication by complex numbers; the axioms keep the same form.

Complex inner products require conjugation; without it, the norm would not reliably produce real, nonnegative lengths.

The standard norm uses absolute values squared—so |i|^2 = 1, not i^2 = −1.

Even though C can be visualized with two real coordinates, its complex dimension is 1 when scaling by complex numbers is allowed.

Topics

Complex Vectors
Complex Matrices
Eigenvalues
Complex Inner Product
Vector Space Axioms