4.11.5 Tensor Multilinear Evaluation Result

Tensor Multilinear Evaluation Result computes the outcome of applying multilinear maps to tensors, revealing structural properties through coordinate-wise operations.

Tensor Multilinear Evaluation Result is the scalar value produced once every slot of a tensor has been filled by the tensor multilinear evaluation operation, together with the properties that this value necessarily satisfies as a consequence of the tensor's multilinearity and basis independence. It is the final output of evaluation, distinct from the tensor itself and from the arguments supplied to it, and its behavior under changes to the arguments, the tensor, or the underlying basis is what characterizes it as a genuine invariant of the algebraic structure rather than an incidental number.

Identity and Membership of the Result

The Result as an Element of the Base Field

For a type (p, q) tensor T on a vector space V over a field F, evaluating T on p covectors and q vectors always produces an element of F itself, never an element of V, V*, or any other space:

T (α^{1}, \dots, v_{q}) \in F

This membership is guaranteed by the definition of T as a map into F, and it holds regardless of the dimension of V, the rank p + q of T, or the particular arguments chosen, since the codomain of the map is fixed once and for all as part of the tensor's type.

Uniqueness of the Result for Given Arguments

For a fixed tensor T and a fixed input tuple, the evaluation result is a single, uniquely determined element of F; there is no ambiguity or choice involved in the output, since T is a well-defined function and functions assign exactly one output to each input.

Behavior Under Changes to the Inputs

Linear Response to Each Argument

Because T is multilinear, the result responds linearly to changes in any single argument while the others are held fixed: scaling one argument by a constant scales the result by the same constant, and replacing one argument by a sum of two vectors or covectors replaces the result by the sum of the two results obtained separately.

Response to Simultaneous Scaling

If every argument is scaled at once, by factors λ_1, ..., λ_{p+q} respectively, the result scales by the product of all the factors:

T (λ_{1} α^{1}, \dots, λ_{p + q} v_{q}) = (\prod_{r = 1}^{p + q}, λ_{r}) T (α^{1}, \dots, v_{q})

a direct consequence of applying linearity in each slot successively across all p + q arguments.

Basis Independence of the Result

The Result Does Not Depend on Coordinate Choices

Although the result can be computed via the component evaluation formula, which explicitly references a chosen basis, the numerical value obtained is exactly the same no matter which basis was used to carry out the computation; this is because a change of basis simultaneously transforms the components of T and the coordinates of every argument in mutually compensating ways, leaving their fully contracted product unchanged.

Consistency Across Equivalent Descriptions of the Tensor

If the same tensor T is described by two different, but mathematically equivalent, tensor product representations, evaluating either representation on the same input tuple must yield the identical result, since both representations describe the same underlying multilinear map and the evaluation operation depends only on that map, not on how it happens to be represented.

The Result in Relation to Rank Reduction

Results from Partial Evaluation

When only some slots of T are filled via slot substitution, the outcome is not yet a scalar evaluation result but a lower-rank tensor; the full evaluation result is obtained only once every slot has received an argument, at which point the type of the remaining object has been reduced all the way to (0, 0), identifying it with a single scalar.

The Result as the Terminal Case of Contraction

The evaluation result can be understood as the terminal case of tensor contraction applied repeatedly: each contraction removes one pair of matched indices, and once no indices remain to be contracted, what is left is precisely the evaluation result, a number with no further tensorial structure.

Interpreting the Result

The Result as a Measurement

In applications where T encodes a geometric or physical quantity, such as a metric tensor or a strain tensor, the evaluation result obtained from specific vector arguments is often interpreted as a measurement, such as a length, an angle, or an energy, extracted from the abstract tensor by supplying concrete directions or displacements as arguments.

The Result as a Coefficient in an Expansion

When T is expanded in a basis, each individual component of T is itself an evaluation result, obtained by evaluating T on the appropriate combination of basis vectors and basis covectors, so the entire component array of a tensor can be understood as a table of evaluation results indexed by every possible basis input tuple.

Diagrammatic Summary

The diagram shows the tensor and its input tuple combining through evaluation to produce a single element of the base field, the evaluation result, which carries no further indices or tensorial structure of its own.