3.10.4 Tensor Covector Component Dual Basis Dependence

Tensor Covector Component Dual Basis Dependence is the specific requirement that a covector's component description is only meaningful, and only reconstructs the covector correctly, when the dual basis used alongside it is exactly the biorthogonal partner of the primal basis of V, rather than any arbitrary basis of V*. This is a sharper and more structural claim than ordinary basis dependence: it is not just that changing the basis of V changes the numbers, but that pairing a fixed set of components with the wrong basis of V* produces an entirely incorrect covector, even if the numbers themselves stay the same.

The Two-Sided Dependency

Extraction Uses the Primal Basis Alone

The extraction step, f_i = f(e_i), requires only the basis e_1, ..., e_n of V; it does not reference the dual basis at all. Any choice of e_1, ..., e_n produces a corresponding, well-defined list of numbers f_1, ..., f_n, regardless of whether a dual basis has even been constructed yet.

Reconstruction Requires the Correct Dual Basis

The reconstruction step, f = f_i e^i, only recovers the original covector f when e^1, ..., e^n is precisely the dual basis satisfying e^i(e_j) = δ^i_j relative to the same e_1, ..., e_n used in extraction. Substituting components extracted with one primal basis into a dual basis built from a different or unrelated basis produces a different, generally incorrect covector.

Illustrating the Failure Mode

A Mismatched Reconstruction

Let V = R^2 with standard basis e_1 = (1, 0), e_2 = (0, 1) and dual basis e^1(x, y) = x, e^2(x, y) = y. Take f(x, y) = 2x + 3y, so f_1 = 2, f_2 = 3, and correctly, f = 2e^1 + 3e^2. Suppose instead these same numbers 2, 3 are mistakenly combined with a different basis of V*, say g^1(x, y) = x + y, g^2(x, y) = y, which is not dual to e_1, e_2. The resulting combination 2g^1 + 3g^2 evaluates to 2(x+y) + 3y = 2x + 5y, which does not equal the original f.

Why the Mismatch Occurs

The failure arises precisely because g^1, g^2 do not satisfy the biorthogonality relation with e_1, e_2: g^1(e_2) = 1, not 0, so the Kronecker-delta cancellation used to derive the reconstruction formula no longer applies, and the reconstructed sum picks up unintended cross terms.

The Biorthogonality Condition as the Structural Requirement

Stating the Condition Precisely

The dual basis dependence of components is fully captured by the single biorthogonality condition e^i(e_j) = δ^i_j. Whenever this condition holds between a chosen basis of V and a chosen basis of V*, extraction and reconstruction are guaranteed to be mutually inverse. Whenever it fails, neither the extraction formula nor the reconstruction formula can be trusted to interoperate correctly.

Uniqueness of the Compatible Dual Basis

For a fixed basis of V, there is exactly one basis of V* satisfying biorthogonality with it; this uniqueness is what justifies referring to "the" dual basis rather than "a" dual basis, and it is why component descriptions are typically presented alongside an explicit or implicit statement of which primal basis, and hence which associated dual basis, is in use.

Practical Implications

Always Pairing Consistent Bases

When components of several different covectors are combined in a calculation, care must be taken that all of them were extracted relative to the same primal basis, and that any reconstruction step uses the dual basis associated with that same primal basis; mixing components extracted under different bases without applying the appropriate transformation first reproduces the same kind of error illustrated above.

Detecting an Inconsistent Setup

If a reconstructed covector, computed as f_i e^i, fails to reproduce known values of f at the original basis vectors, the most likely cause is that the dual basis used in the reconstruction does not actually satisfy biorthogonality with the primal basis used during extraction, and the dual basis should be recomputed directly from the biorthogonality condition before proceeding.

Relation to General Coordinate Expressions

Recovering the General Pairing Matrix

If a covector basis that is not dual to the chosen vector basis must be used for some reason, the correct procedure is not to apply the simple reconstruction formula, but to use the general pairing matrix P^a_i = g^a(e_i) and the corresponding coordinate formula f_a v^i P^a_i, which properly accounts for the mismatch between the two bases rather than assuming a biorthogonality that does not hold.

Why the Simple Formula Is the Default

The overwhelming preference for using the dual basis in practice is precisely because it eliminates the need to track a separate pairing matrix, reducing the general formula to the simplest possible case, P = I, and this simplification is only valid under the dual basis dependence described here.

Diagrammatic Summary

The diagram contrasts reconstructing with the correct, biorthogonal dual basis against reconstructing with an unrelated basis of V*, showing that only the former recovers the original covector.