7.23.1 Tensor Component Object Boundary

Tensor Component Object Boundary is the conceptual line separating a tensor as an abstract, basis-independent object from the array of numbers — its components — that represent that object relative to a chosen basis or coordinate system. It marks where the invariant identity of the tensor ends and the basis-dependent description used to compute with it begins, and it is the boundary that explains why components change under a change of basis while the object they represent does not.

The Object Side and the Component Side

The Tensor as an Object

On one side of the boundary sits the tensor itself: an element of a tensor product space, or equivalently a multilinear map, defined without reference to any basis. This object has an existence and an algebra of its own — addition, scalar multiplication, tensor product, contraction — all of which can be defined purely in terms of the object's universal property, with no numbers involved.

The Components as a Representation

On the other side sits the tuple of numbers T^{i₁...iₚ}_{j₁...j_q} obtained by evaluating the tensor on the basis vectors and dual basis covectors of a chosen basis. This tuple is a representation of the object, in the same sense that a matrix is a representation of a linear map: useful and necessary for computation, but not identical to the object it represents.

where the left side is the basis-free object and the right side is its expansion in terms of components T^{i}_{j} and the basis vectors eᵢ together with dual basis covectors e^{j}.

What Crosses the Boundary and What Does Not

Components Change, the Object Does Not

Under a change of basis, the numbers in the component tuple change according to the tensor transformation law, while the object T on the left side of the expansion above remains the same object throughout; the transformation law for components exists precisely because it must compensate for the change in basis so that the same fixed object is recovered on the left.

Statements That Live Purely on One Side

A statement made about a tensor object — "T is symmetric," "T is the zero tensor," "T is the sum of two other tensors" — is a basis-independent statement and holds regardless of which side of the boundary it is checked from. A statement made about specific numerical components — "the component T^{1}_{2} equals 5" — is basis-dependent and generally true in one basis and false in another, since it refers to a quantity that lives strictly on the component side of the boundary.

Diagram of the Boundary

Why the Boundary Is Not a Contradiction

Reconciling Invariance with Numerical Change

The apparent tension between "the tensor is invariant" and "its components change" is resolved once the boundary is made explicit: invariance is a property asserted about the object side, while change under basis transformation is a property of the component side, and the transformation law is exactly the translation rule that keeps the two sides consistent with one another as the basis varies.

The Boundary Is Crossed by Choosing a Basis

Passing from the object side to the component side requires an explicit choice — a basis of the underlying vector space and its dual basis. Without making that choice, there are no components to speak of; the object exists prior to and independently of any such choice, which is exactly what places it on the far side of the boundary from its components.

Practical Implications

Computation Necessarily Crosses the Boundary

Any concrete numerical computation involving a tensor — matrix multiplication, contraction carried out by summing over indices, numerical evaluation of a physical quantity — necessarily takes place on the component side, since numbers are what computations manipulate. The object side supplies the meaning and the invariance guarantees; the component side supplies the actual arithmetic.

Correct Practice Keeps Track of Which Side a Claim Belongs To

A common source of error in working with tensors is asserting, on the basis of a computation performed in one particular basis, a conclusion intended to hold on the object side without checking that the conclusion is actually basis-independent. Sound tensor reasoning keeps explicit track of the object-component boundary, verifying that any claim carried back across it from a component computation is one that does not depend on the basis in which the computation happened to be performed.