3.11 Tensor Dual Space Dimension Structure

Tensor Dual Space Dimension Structure is the study of how the dimension of a dual space V* relates to the dimension of the original space V, established through the explicit construction of a dual basis in the finite-dimensional case and contrasted sharply with the very different, strictly larger dimensional behavior that occurs when V is infinite-dimensional. In finite dimensions, V* has exactly the same dimension as V, a fact that underlies the existence of a non-canonical isomorphism between the two spaces, while in infinite dimensions this equality breaks down entirely.

The Finite-Dimensional Case

Equality of Dimension

For a finite-dimensional vector space V of dimension n, the dual space V* also has dimension n:

Proof via the Dual Basis

Given any basis e_1, ..., e_n of V, the dual basis e^1, ..., e^n, defined by e^i(e_j) = δ^i_j, is a basis of V*. It spans V* because every linear functional f decomposes as f = f_i e^i with f_i = f(e_i), and it is linearly independent because if c_i e^i = 0 as a functional, evaluating at each e_j forces every c_j = 0. Having exhibited a basis of V* with exactly n elements, the dimension of V* is confirmed to equal n.

The Non-Canonical Isomorphism V ≅ V*

Existence of an Isomorphism

Since V and V* share the same finite dimension n over the same field, they are isomorphic as vector spaces: there exists a linear bijection between them. One such isomorphism sends each basis vector e_i to the corresponding dual basis covector e^i, extended linearly.

Why This Isomorphism Is Not Canonical

This isomorphism depends entirely on the choice of basis used to construct it; a different basis of V generally produces a different isomorphism V -> V*, with no way to select one particular isomorphism as more natural or preferred than another without additional structure, such as a fixed inner product. This is in sharp contrast to the double-dual identification V ≅ V**, which requires no choice of basis at all and is therefore canonical.

The Infinite-Dimensional Case

Dimension of the Dual Space Strictly Exceeds the Original

When V is infinite-dimensional, the algebraic dual space V*, consisting of all linear functionals on V, has strictly greater dimension than V itself. This striking asymmetry arises because a basis of an infinite-dimensional space, while it exists by the axiom of choice, has a cardinality that is strictly smaller than the cardinality of all possible functions from that basis to the field, and V* can be shown to have dimension equal to this larger cardinality.

Illustrative Example

If V has a countably infinite basis, indexed by the natural numbers, V* can be shown to have dimension equal to the cardinality of the continuum, since a linear functional can assign an arbitrary field value to each basis vector independently, and there are uncountably many such independent assignments, whereas V itself, built from finite linear combinations of basis vectors, remains countably infinite-dimensional.

Failure of a Natural Dual Basis Construction

The finite-dimensional construction of a dual basis from e^i(e_j) = δ^i_j does not produce a basis of the full algebraic dual V* in the infinite-dimensional setting; it produces only a proper subspace of V*, since a general linear functional need not vanish on all but finitely many basis vectors, while the elements built from this delta relation only ever act nontrivially on finitely many basis vectors at a time.

Consequences for Working with Dual Spaces

Finite-Dimensional Intuitions Do Not Transfer Directly

Many of the convenient facts used throughout finite-dimensional tensor algebra, such as the equality of dimension between V and V*, the existence of a dual basis spanning all of V*, and the canonical identification of V with V**, rely essentially on finiteness and must be handled with additional care, or replaced with topological duals restricted to continuous functionals, when working in infinite-dimensional settings such as function spaces.

Why Tensor Algebra Typically Assumes Finite Dimension

Most introductory treatments of tensor algebra restrict attention to finite-dimensional vector spaces precisely so that the dimension structure of V* behaves simply and predictably, matching V exactly, which keeps constructions such as the dual basis, the double-dual identification, and general tensor spaces T^p_q(V) well-behaved and finite-dimensional themselves.

Diagrammatic Summary

The diagram contrasts the equal dimensions of V and V* in the finite case with the strictly larger dimension of V* that occurs once V becomes infinite-dimensional.