10.23 Tensor Change of Basis Boundary

Tensor Change of Basis Boundary is the set of conditions and limiting cases that mark where a tensor change-of-basis operation remains mathematically valid and where it ceases to apply, encompassing the domain restrictions on the coordinate transformation, the points at which the transformation degenerates, and the interface between overlapping coordinate patches on a manifold.

Domain of Validity

Local Invertibility Requirement

A change of basis between two coordinate systems is only well defined on the subset of the space where the coordinate transformation is invertible, meaning the Jacobian determinant of the transformation is nonzero throughout that subset. Outside this subset, the transformation cannot be reversed and tensor components cannot be consistently converted back and forth.

Open Set Restriction

Because invertibility is typically only guaranteed on an open subset of the underlying space, the change of basis is described as being valid on that open set, and any claim about tensor transformation must be understood as implicitly restricted to this region rather than to the space as a whole.

Points of Degeneration

Singular Points of the Transformation

At points where the Jacobian determinant vanishes, the coordinate transformation degenerates, collapsing one or more directions of the tangent space, and the corresponding Jacobian factors either become undefined or fail to form an invertible matrix. Such points are excluded from the domain over which the change of basis is applied.

Coordinate Singularities Versus Geometric Singularities

A boundary of this kind may reflect a mere artifact of the chosen coordinates, called a coordinate singularity, which can be removed by switching to a different coordinate system, or it may reflect a genuine geometric degeneracy of the underlying space that persists no matter which coordinates are chosen. Distinguishing between these two cases is essential before concluding that a tensorial quantity itself is singular.

Chart Boundaries on a Manifold

Overlapping Coordinate Patches

On a general manifold, a single coordinate system rarely covers the entire space, so the manifold is described using a collection of overlapping coordinate patches, called charts. The boundary of a chart is the edge of the region on which that particular coordinate system remains valid, and change of basis is performed only within the overlap of two charts.

Transition Functions

The change of basis formula relating the coordinates of two overlapping charts is called a transition function, and this function is only defined, and only required to be smooth, on the overlap region. Outside the overlap, the notion of transforming components between the two charts has no meaning, since one of the two coordinate systems simply does not extend there.

Boundaries Introduced by the Space Itself

Edges and Boundary Manifolds

When the underlying space has a genuine topological boundary, such as a manifold with boundary, a coordinate chart near that boundary is typically restricted to a half-space, and tensor components defined there must respect this restriction. A change of basis performed near such an edge must map boundary points to boundary points, since interior points and boundary points cannot be mixed by a valid coordinate transformation.

Behavior Under Approach to the Boundary

As a point of evaluation approaches a boundary of validity, whether it is a singular point of the transformation or the edge of a chart, the individual Jacobian factors may grow without bound, oscillate, or otherwise fail to converge. Whether the tensor components themselves remain well behaved in this limit depends on the specific transformation and cannot be assumed automatically.

Practical Consequences for Working With Tensors

Restricting Claims to the Valid Region

Any statement about how a tensor's components transform is only as reliable as the domain over which the Jacobian factors used in that statement are well defined. Extending a transformation formula past its boundary of validity without justification is a common source of erroneous conclusions about tensor behavior.

Patching Together Global Results

Global statements about tensors on an entire manifold are typically built by verifying that the local change-of-basis formulas agree consistently on every overlap between charts, so that boundaries between individual coordinate patches do not introduce contradictions when the local pieces are assembled into a single global description.