3.5 Tensor Basis Covector Structure

Tensor Basis Covector Structure is the study of the individual members e^i of a dual basis as objects in their own right, distinct from the collective dual basis construction that produces them, covering the defining orthogonality-like relation each basis covector bears to the basis vectors, the way each basis covector singles out one coordinate direction of V, and how the full set of basis covectors, taken together, generates every other covector in V* through linear combination.

The Defining Relation of a Single Basis Covector

Orthogonality to All but One Basis Vector

A basis covector e^i is characterized entirely by its values on the basis vectors of V: e^i(e_i) = 1, and e^i(e_j) = 0 for every j ≠ i. In this sense, e^i is "orthogonal," in the pairing sense rather than any metric sense, to every basis vector except e_i, on which it takes the value 1; this is the covector analogue of a standard basis vector having a single nonzero coordinate equal to 1.

Uniqueness Given the Basis

For a fixed basis {e_1, ..., e_n} of V, there is exactly one covector satisfying this defining relation for each index i, by the uniqueness-from-basis-values property of linear functionals; this uniqueness is what makes it meaningful to speak of "the" basis covector e^i, rather than one of several candidates.

Each Basis Covector as a Coordinate-Direction Detector

Detecting Presence Along One Direction

Applying e^i to a general vector v = Σ_j v^j e_j isolates exactly the coefficient of e_i in that expansion, e^i(v) = v^i; conceptually, e^i measures "how much of v points in the e_i direction," in the coordinate sense determined by the basis, without reference to any notion of length or angle.

Basis Covectors Vanish on Complementary Spans

For a fixed index i, the basis covector e^i vanishes precisely on the span of the remaining basis vectors, span(e_j : j ≠ i), since every vector in this span is a linear combination of basis vectors other than e_i, each annihilated by e^i. This span is an (n-1)-dimensional subspace, and e^i is exactly the linear functional, up to the specific normalization e^i(e_i) = 1, whose kernel is this particular hyperplane, connecting basis covector structure directly to the kernel structure of linear functionals in general.

Basis Covectors as Generators of the Full Dual Space

Spanning V* by Linear Combination

Any covector ω ∈ V* is a linear combination ω = Σ_i ω_i e^i of the basis covectors, with coefficients ω_i = ω(e_i) recovered by evaluating ω on the corresponding basis vector; the collection {e^1, ..., e^n}, though each member is individually a narrowly specialized single-direction detector, together generates the entirety of V* by taking all possible linear combinations.

Independence Ensures No Redundancy

Because the basis covectors are linearly independent, no proper subset of them spans all of V*; each e^i contributes genuinely new information not obtainable as a combination of the others, matching the independence verification, Σ_i c_i e^i = 0 ⟹ all c_i = 0, obtained by evaluating the hypothetical dependency relation at each e_j in turn.

Basis Covectors and the Standard Coordinate Functions

Familiar Special Case

When V = F^n with the standard basis, the basis covectors e^i coincide exactly with the standard coordinate projection functions, e^i(x_1, ..., x_n) = x_i; this familiar special case is the concrete picture underlying the abstract basis covector construction for a general vector space and general basis, with the abstract construction reducing to ordinary coordinate projections whenever the standard basis is used.

Generalization Beyond the Standard Basis

For a non-standard basis, the basis covectors e^i are still coordinate-reading functionals, but relative to the coordinates induced by that particular basis rather than the ambient standard coordinates of F^n, as illustrated by the explicit computation e'^1(x, y) = x - y obtained in tensor dual basis construction for a specific non-standard basis of F^2; basis covector structure applies uniformly across every choice of basis, with the standard case being only the most immediately recognizable instance.