10.1.4 Tensor Transformation Matrix Scope

Explore how tensor transformation matrices operate within their defined scope, mapping tensor components across different coordinate systems.

Tensor Transformation Matrix Scope is the delineation of exactly which matrices are eligible to serve as the change-of-basis matrix in tensor transformation, specifying the structural properties such a matrix must possess and separating it from other matrices that may appear in adjacent calculations without playing this specific role.

Required Properties of the Transformation Matrix

Square and Invertible

The matrix (A) governing a change of basis must be square, with as many rows and columns as the dimension (n) of the vector space, and it must be invertible, since a valid basis change requires a well-defined inverse matrix to convert components back in the reverse direction.

A \in {GL}_{n} (ℝ)

Membership in the general linear group (GL_n) of invertible (n \times n) matrices is exactly the condition defining scope here: any matrix outside this group cannot serve as a legitimate basis transformation matrix, regardless of how it arises in a calculation.

Encoding a Genuine Basis Relationship

The scope requires that the entries of (A) arise specifically from expressing one basis in terms of another, column by column or row by row depending on convention, rather than being an arbitrary invertible matrix unconnected to any actual pair of bases.

e_{i}' = A_{i}^{j} e_{j}

Matrices Included Within Scope

Constant Matrices for Linear Basis Change

The most direct case within scope is a constant matrix relating two fixed bases of a finite-dimensional vector space, applicable uniformly at every point since there is no notion of position involved.

Jacobian Matrices for Coordinate Transformations

The scope extends to include Jacobian matrices built from the partial derivatives of one coordinate system with respect to another, provided this Jacobian is invertible at the point being considered; this generalizes the matrix's role from a single global object to a locally defined one that can vary from point to point.

A_{i}^{j} = \frac{\partial x^{j}'}{\partial x^{i}}

Matrices Excluded From This Scope

Non-Square or Singular Matrices

A matrix that is not square, or that is square but singular with zero determinant, cannot represent a change of basis, since it cannot map a valid basis to another valid basis of the same dimension; such matrices lie entirely outside the scope, regardless of any other role they might play in linear algebra.

\det (A) = 0 \Rightarrow excluded from scope

The Metric Tensor Matrix

Although the metric tensor is often represented as a matrix and used to raise and lower indices, it plays a fundamentally different role from a change-of-basis matrix: it encodes inner products between basis vectors rather than relating one basis to another, and so it lies outside this specific scope even though both objects are represented by square matrices.

Matrices Representing Linear Operators on the Space

A linear operator acting on vectors within a single fixed basis, such as a rotation applied physically to a vector, is represented by a matrix that looks identical in form to a change-of-basis matrix but serves an entirely different, active role; distinguishing this case from a passive change of basis is essential, and such operator matrices are excluded from the change-of-basis matrix scope even when numerically identical to one.

Visual Illustration

Why This Scope Must Be Drawn Carefully

Restricting the transformation matrix scope to square, invertible matrices that specifically encode a relationship between two bases prevents confusion with superficially similar matrices, such as the metric tensor or an active linear operator, that play entirely different roles within tensor algebra. Getting this scope right is what ensures the transformation rule for components is applied only where it is actually meaningful, avoiding the error of treating any convenient square matrix as though it were a legitimate change-of-basis matrix.