6.22.5 Tensor Classification Consistency

Tensor Classification Consistency is the requirement that every way of describing a tensor's order-type status — its index notation, its membership in a specific tensor product space, its transformation law under a change of basis, and its behavior under contraction and tensor product — must agree with every other description, so that no tensor can be simultaneously classified as, say, type (1,2) by one criterion and type (2,1) by another. This consistency is not an extra property imposed on tensors from outside; it follows automatically from the way the tensor product construction and the order-type relation are defined, but it is worth isolating explicitly because it is what makes tensor classification a reliable tool rather than a set of independent, potentially conflicting labels.

The Criteria That Must Agree

Index Notation Criterion

Reading the number of superscripts and subscripts on T^{i₁...i_p}_{j₁...j_q} gives a candidate type (p, q). This criterion is purely syntactic and can be checked by inspection of the written symbol alone.

Vector Space Membership Criterion

Identifying which tensor product space T was defined to live in, ⊗ᵖV ⊗ ⊗ᑫV*, gives a second candidate type, read off from the construction of the space itself rather than from any particular notation used to write elements of it.

Transformation Law Criterion

Observing how the components of T actually transform under a test change of basis — how many index positions pick up a factor of the transition matrix A versus its inverse — gives a third, operationally verifiable candidate type.

The Consistency Requirement

and the analogous equality for q. Tensor classification consistency is the assertion that these three numbers, computed by three structurally different means, always coincide for any correctly constructed and correctly notated tensor.

Why the Consistency Holds by Construction

The Definitions Are Not Independent

The three criteria are not independent axioms that happen to agree; the transformation law is derived from the vector space membership (an element of V must transform contravariantly because that is what distinguishes V from V*), and the index notation is a convention adopted specifically to encode that same transformation behavior. Consistency therefore holds because the later criteria are built to track the earlier ones, not because of a separate theorem requiring proof.

What a Violation Would Indicate

If the three criteria ever disagreed — for instance, if an object notated with two upper indices were found to transform with one factor of A and one factor of A⁻¹ — this would indicate a notational error or a miscomputed transformation, not a new kind of tensor. Classification consistency therefore functions as a diagnostic: disagreement among the criteria is treated as a signal to locate and correct the error, rather than as evidence of some exotic mixed-type object.

Diagram of the Three Criteria Converging

Consequences for Practical Tensor Work

Safe Interchangeability of Descriptions

Because the criteria are guaranteed consistent, a working mathematician or student can freely switch between describing a tensor by its index pattern, by its home space, or by its transformation behavior, using whichever is most convenient for the task at hand, without risk of the description silently changing the object under discussion.

Foundation for Automated and Symbolic Verification

Classification consistency is also what allows symbolic computation systems to check tensor equations automatically: verifying that an equation is well-typed reduces to checking index balance (the notation criterion), which is guaranteed to reflect the true transformation behavior (the transformation criterion) precisely because consistency holds, sparing the need to re-derive the transformation law from scratch for every expression checked.