7.23.4 Tensor Component Transformation Boundary

Tensor Component Transformation Boundary is the limit of the domain over which the tensor transformation law — the rule relating a tensor's components in one basis or coordinate system to its components in another — remains applicable, invertible, and well-defined. It marks where the Jacobian matrix connecting two coordinate systems fails to be invertible or fails to exist, so that the standard component transformation formula can no longer be used to pass from one representation to the other.

The Transformation Law and Its Requirements

The Law Itself

For a change of coordinates from x^{i} to x^{i′}, tensor components transform according to a rule built from the Jacobian matrix ∂x^{i′}/∂x^{i} and, for a fully mixed tensor, its inverse ∂x^{i}/∂x^{i′} applied to the subscript slots. A single mixed tensor component transforms as

with implicit summation over the repeated indices i and j.

Where the Law Requires Invertibility

For this law to make sense, the Jacobian matrix must be invertible at the point in question, since the subscript indices are transformed using its inverse. The transformation boundary is precisely the set of points at which this invertibility fails — where the Jacobian's determinant vanishes, or where one or both coordinate systems are simply not defined.

Sources of a Transformation Boundary

Vanishing Jacobian Determinant

At a point where the Jacobian determinant is zero, the coordinate change is not locally invertible there, so no well-defined inverse Jacobian exists to transform subscript indices; the transformation law breaks down at exactly these points, independent of whether either coordinate system individually remains numerically defined.

Non-Overlapping Chart Domains

The transformation law between two coordinate charts is meaningful only on the overlap of their two domains; outside this overlap, at least one of the two coordinate systems assigns no value to the point in question, so there is nothing for the transformation law to relate, and this non-overlap region is likewise a boundary of the transformation's applicability.

Diagram of a Transformation Boundary

Consequences of Crossing the Transformation Boundary

No Well-Defined Component Values on the Far Side

If a point lies beyond the transformation boundary between two specific charts, there is no meaningful way to convert a component value computed in one chart into a component value in the other at that point; attempting to apply the transformation formula there, using a Jacobian that is singular or simply undefined, produces a result with no valid interpretation.

Distinguishing a Transformation Boundary from a True Tensor Singularity

As with other forms of tensor component boundary, a breakdown of the transformation law at a point does not by itself imply that the tensor is singular there; it may simply mean that the particular pair of coordinate systems being related is poorly suited to that point. Evaluating an invariant scalar built from the tensor, independent of any transformation between specific charts, is the standard way to determine whether a genuine singularity is present.

Practical Handling

Choosing Charts with Compatible Domains

Sound practice in working with tensor components across coordinate systems is to verify, before applying the transformation law, that the point of interest lies within the overlap of the two charts' domains and away from any point where the connecting Jacobian degenerates, since the law is only guaranteed valid under these conditions.

Composing Transformations Through Intermediate Charts

When two charts do not directly overlap in a useful region, or their direct Jacobian is difficult to compute, an intermediate chart with valid transformation boundaries to both may be used to compose two transformation laws in sequence, extending the practical reach of the transformation law without violating the invertibility requirement at any individual step.