1.6 Tensor Invariance Foundations

Tensor Invariance Foundations is the body of concepts explaining why tensors, despite being represented by arrays of numbers that change from basis to basis, describe objects and relations that do not themselves depend on any choice of basis or coordinate system. It establishes the conceptual bridge between the component tensor representation, which is basis-dependent, and the abstract multilinear object that representation stands for, and it identifies the precise conditions an array of numbers must satisfy to qualify as invariant tensor content rather than an artifact of a particular frame.

What Invariance Means for a Tensor

The Object Versus Its Description

A tensor, understood abstractly, is a multilinear map defined without reference to any basis. Its component representation is merely a convenient numerical encoding obtained once a basis has been chosen. Invariance is the property that guarantees these two levels, the abstract object and its numerical encoding, always refer to the same underlying entity, no matter which basis produced the encoding.

Invariance Is Not Constancy

An invariant tensor is not required to have the same components in every basis; its components generally do change. What is invariant is the tensor itself, together with any basis-independent quantity, such as a scalar formed from it, that can be extracted from those components by an operation compatible with the transformation law.

The Transformation Law as the Guarantor of Invariance

The General Law

The transformation law for a type (p, q) tensor specifies exactly how its components must change when the basis changes, using p factors of the change-of-basis matrix A for its upper indices and q factors of the inverse matrix A^{-1} for its lower indices.

Why This Particular Law

This law is not an arbitrary convention; it is precisely the condition that makes the sum reconstructing the abstract tensor from its components, T = T^{...}_{...} e_{i_1} ⊗ ... ⊗ e^{j_1} ⊗ ..., evaluate to the same abstract object in every basis. An array obeying a different rule would fail to represent a single fixed object across changes of basis.

Invariants Constructed from Tensors

Full Contraction as Invariant Extraction

Whenever every upper index of a tensor is paired with a lower index and summed, factors of A and factors of A^{-1} appear together and cancel, since A and A^{-1} are inverse matrices. The result of such a full contraction is a number that is the same in every basis, an invariant.

Examples of Invariants

The trace of a type (1, 1) tensor, the norm of a vector computed through a metric tensor, and the determinant of a linear map are all classical examples of scalar invariants obtained from tensors of higher rank by contraction, each one independent of the basis used in its computation even though the intermediate component arrays are not.

Invariance Under Restricted Versus General Transformations

Full Invariance Under All Bases

A tensor's status as a tensor requires its components to obey the transformation law under every admissible change of basis of the vector space in question, with no restriction on which changes are allowed.

Invariance Under a Restricted Group

In many settings, attention is restricted to a subgroup of basis changes, such as orthogonal transformations or transformations preserving orientation, and a quantity may be invariant under that restricted group without being invariant under the full group of basis changes. Such quantities are still meaningfully invariant, but only relative to the structure the restricted group preserves.

Invariance and Physical Law

Coordinate-Independent Formulation

Physical and geometric laws expressed in terms of tensors are automatically valid in every coordinate system once verified in one, because the tensor transformation law guarantees that an equation between tensors of matching type, if it holds in one basis, holds in every basis obtained from it by a change of coordinates.

Distinguishing Genuine Tensor Equations from Coordinate Artifacts

An equation written using arrays of numbers that do not obey the tensor transformation law may hold in one basis and fail in another; tensor invariance foundations supply the criterion for recognizing when an equation expresses a real, basis-independent relationship versus when it is merely a coincidence of a particular coordinate system.

Diagrammatic Summary

The diagram represents the central claim of tensor invariance foundations: a single abstract tensor T gives rise to different component arrays in different bases, connected to one another by the transformation law, while T itself and every scalar invariant extracted from it remain fixed.