✦ For everyone, free.

Practical knowledge for real and everyday life

Home

13.8.5 Tensor Trace Contraction Basis Independence

Tensor trace contraction is basis-independent, revealing intrinsic properties through invariant contraction operations in tensor algebra.

Tensor Trace Contraction Basis Independence is the property whereby the scalar produced by contracting a mixed slot pair of a tensor remains identical regardless of which basis is used to express the tensor's components, so that the trace is a genuine invariant of the tensor rather than an artifact of any particular coordinate description. It formalizes the guarantee that computing the trace in two different bases, related by any invertible change of coordinates, always yields the same numerical value.


Conceptual Basis

Why Basis Independence Is Expected

Tensors are defined as objects whose components transform in a specific, coordinated way under a change of basis, with contravariant indices transforming via a Jacobian matrix and covariant indices transforming via its inverse. The trace pairs exactly one of each, so the two transformation factors have the potential to cancel, which is precisely what basis independence asserts happens.

Basis Independence as a Defining Feature of Contraction

The requirement that contraction pair a contravariant index with a covariant index is not an arbitrary convention but is what makes basis independence possible in the first place; pairing two indices of the same variance would not produce this cancellation and would not yield an invariant result.

Relationship to Physical and Geometric Meaning

Basis independence is what allows the trace to be interpreted as carrying intrinsic meaning about the object being contracted, since a quantity that depended on the choice of coordinates would not correspond to any property of the underlying map or field itself.


Formal Description

Transformation of the Mixed Tensor

Under a change of basis described by an invertible matrix with components Aii and its inverse Ajj, a mixed tensor transforms as:

Tji = Aii Ajj Tji

Cancellation Upon Tracing

Setting the primed indices equal and summing to compute the trace in the new basis gives:

Tii = Aii Aij Tji

and because AiiAij reduces to the identity δji, this simplifies to:

Tii = Tii

confirming that the trace computed in the new basis equals the trace computed in the original basis.

General Statement

For any invertible change of basis applied to a rank-two mixed tensor, the identity above shows that the scalar produced by the trace contraction is preserved exactly, establishing basis independence as a direct consequence of the inverse relationship between contravariant and covariant transformation rules.


Consequences

Trace as a True Invariant

Basis independence elevates the trace from a mere computational shortcut to a genuine invariant of the tensor, meaning it captures information about the underlying object itself rather than about the coordinate system chosen to describe it.

Foundation for Further Invariants

Because basis independence holds for the trace of a single mixed tensor, it extends to traces of tensors built from products or powers of an original tensor, forming the basis for constructing entire families of invariants used to characterize linear operators or tensor fields.

Consistency Check for Computations

Since the trace must agree across any two valid bases, recomputing it after an explicit change of coordinates provides a direct and reliable check on whether a tensor's transformation law has been applied correctly in a given calculation.


Limits of the Property

Restriction to Mixed Pairs

Basis independence as described here applies specifically to contractions over a mixed contravariant-covariant pair; summing over two indices of the same variance, which is not a valid contraction, does not enjoy this cancellation and does not produce a basis-independent result.

Dependence on Linearity of the Transformation

The cancellation relies on the transformation being linear, as captured by the Jacobian matrix and its inverse; the property applies to tensors precisely because their defining transformation law is linear in this sense.

Applicability Beyond Rank Two

The same cancellation mechanism extends to a single mixed pair contracted within any higher-rank tensor, with the remaining free indices continuing to transform according to their own variance while the contracted pair contributes no dependence on the change of basis.