5.10.3 Tensor Product Finite Dimension Case

Tensor Product Finite Dimension Case is the setting in which every factor space participating in a tensor product has finite dimension, and the collection of simplifications this setting affords: a single, unambiguous construction of the tensor product with no completion step required, a concrete multidimensional array representation, and computational tractability for the standard operations of tensor algebra.

Defining Features of the Finite-Dimensional Setting

When V1, V2, …, Vn are finite-dimensional vector spaces over a field F, with dimensions d1, …, dn, the tensor product V1 ⊗ V2 ⊗ ⋯ ⊗ Vn is itself finite-dimensional, with dimension given exactly by the product

This finiteness is what allows every construction of the tensor product — the free-vector-space quotient, the universal property characterization, and any explicit basis-driven construction — to agree exactly, with no distinction between algebraic and topological completions, since a finite-dimensional vector space carries a unique natural topology compatible with its linear structure.

Absence of Completion Issues

Unlike the infinite-dimensional case, the finite-dimensional tensor product requires no analytic machinery to be fully and unambiguously defined.

No Norm Ambiguity

In infinite dimensions, different reasonable choices of norm on the tensor product (projective, injective, and others) can produce genuinely different completed spaces. In the finite-dimensional case, all norms on a finite-dimensional vector space are topologically equivalent, so this ambiguity simply does not arise: there is exactly one finite-dimensional tensor product space, regardless of which norm, if any, one chooses to place on it afterward.

Elements Are Always Finite Sums

Every element of a finite-dimensional tensor product is, trivially, a finite sum of decomposable tensors — indeed a sum of at most dim(V1 ⊗ ⋯ ⊗ Vn) such terms, using the induced basis itself as a (typically inefficient) decomposition — so the finite-dimensional case never requires the notion of a convergent infinite sum that is essential to defining Hilbert or Banach space tensor products.

Concrete Array Representation

The finite-dimensional case is the setting in which tensors are most directly identified with ordinary multidimensional arrays of numbers, since a finite basis for every factor produces a finite, indexable coordinate system for the whole tensor product.

Coordinates as an n-Way Array

Once bases are chosen for each Vi, every tensor corresponds to a unique array of shape d1 × d2 × ⋯ × dn, and the abstract operations of the tensor product — addition, scalar multiplication, tensoring, and contraction — correspond to concrete, directly programmable operations on these arrays.

Matrix Case as the Prototype

When n = 2, this array representation reduces exactly to an ordinary matrix, and the finite-dimensional two-factor tensor product recovers the entire classical theory of matrices — rank, singular value decomposition, outer products — as a special case, providing the intuition and the base examples that motivate the more general finite n-factor theory.

Computational Tractability

Because every relevant object in the finite-dimensional case is a finite array of numbers, standard operations on tensors are directly computable by finite algorithms, in sharp contrast to the analytic subtleties of the infinite-dimensional theory.

Decidable Recognition and Bounded Rank

Recognizing whether a given finite-dimensional tensor is decomposable reduces to a finite, polynomial-time check on the ranks of its flattenings, and the tensor rank of any element in a finite-dimensional tensor product is itself always a finite, well-defined nonnegative integer bounded above by the ambient dimension.

Finite Basis for Linear Maps and Bilinear Forms

The finite-dimensional case also guarantees that spaces of linear or multilinear maps built from the factor spaces — dual spaces, spaces of bilinear forms, spaces of linear operators — are themselves finite-dimensional, with dimensions computable directly from the factor dimensions via the same multiplicative and related formulas, closing the theory off from any need to invoke infinite-dimensional functional-analytic tools.

Relation to the General Theory

The finite-dimensional case is not a separate theory but the specialization of the general tensor product construction to a setting where every simplifying assumption (finite bases, no completion, unique norm topology) happens to hold simultaneously; every general theorem about tensor products applies to the finite-dimensional case without modification, while the reverse — extending finite-dimensional intuition and techniques to the infinite-dimensional setting — requires the additional analytic care described in the treatment of the infinite-dimensional case.

Illustrative Diagram

The bounded, fully drawn grid represents a finite-dimensional tensor product's coordinate array in its entirety, with no unbounded or completed region left implicit, unlike the infinite-dimensional case.