6.20 Tensor Covector First Order Classification

Tensor Covector First Order Classification is the categorization of a linear functional as a tensor of type (0, 1), meaning it carries no contravariant index and exactly one covariant index, placing it alongside the type (1,0) vector as one of the two possible first-order tensors and establishing the covariant transformation law as the feature that distinguishes it from a vector. This classification formalizes the notion of a covector, an element of the dual space V*, as a special case of the general tensor concept, mirroring the vector classification but with the opposite transformation behavior under a change of basis.

Defining Features of the Covector Classification

Index Structure

A type (0,1) tensor is written with a single subscript, φ_i, reflecting that p = 0 and q = 1 in the general notation. Coordinate-free, a covector φ is an element of V*, the space of linear functionals on V, since the tensor product defining type (0,1) objects, zero factors of V and one factor of V*, reduces to V* alone.

A Single Free Index and n Components

Since a type (0,1) tensor has exactly one free index ranging over the dimension n of V, it has n independent components, φ_1 through φ_n, matching the ordinary notion of a covector's coordinates relative to a chosen dual basis.

The Covariant Transformation Law

Behavior Matching the Basis Vectors

Under a change of basis with transition matrix A, where e'_i = A^k_i e_k, the components of a covector transform using the same forward matrix A:

This matching behavior is precisely why the index is termed "covariant," meaning it varies together, or co-, with the way the basis vectors themselves vary, in contrast to the inverse behavior exhibited by contravariant vector components.

Why This Law Preserves the Pairing with Vectors

The covariant transformation law is derived, and its correctness confirmed, by requiring the scalar pairing φ(v) = φ_i v^i to remain invariant under any change of basis: substituting the transformation laws for both φ_i and v^i and using A B = I shows that the value of the pairing is unaffected by the choice of basis, exactly as required for a well-defined scalar quantity.

The Covector as a Linear Functional on Vectors

Direct Functional Action

The defining functional role of a type (0,1) tensor is direct: a covector φ maps each vector v in V to a scalar via φ(v) = φ_i v^i, satisfying linearity, φ(av + bw) = aφ(v) + bφ(w), for all scalars a, b and vectors v, w. This is the fundamental definition from which the dual space V* is constructed as the set of all linear functionals on V.

The Dual Basis Origin of Covector Components

Given a basis {e_i} of V, the dual basis {e^i} of V* is defined by the requirement e^i(e_j) = δ^i_j, and every covector φ expands as φ = φ_i e^i, with the components φ_i recoverable by direct evaluation, φ_i = φ(e_i).

Diagram of the Covector Classification

Distinguishing Covectors from Related First-Order Tensors

Versus Vectors of Type One Zero

A vector, or type (1,0) tensor, has one upper index instead, transforming with the inverse matrix B rather than the forward matrix A; vectors and covectors both have total order one and both have n components, but they belong to different underlying spaces, V and V* respectively, and their opposite transformation behaviors are precisely what makes their pairing φ_i v^i produce a coordinate-invariant scalar.

Building Blocks for Higher-Order Tensors

Every higher-order tensor is built from tensor products of type (1,0) vectors and type (0,1) covectors, making the covector classification, alongside the vector classification, one of the two fundamental generating building blocks of the entire tensor algebra, with the scalar classification serving as the base case beneath both.

Practical Role in Applications

The covector classification underlies the representation of gradients and linear measurement functionals in applications, and its covariant transformation law is what guarantees that a physical rate of change measured by a gradient remains the same physical quantity regardless of the coordinate grid used, differing only in the numerical values of its components from one coordinate system to another.