12.3.1 Tensor Addition Same Type Requirement

Tensor Addition Same Type Requirement is the constraint stating that two tensors can only be added together when they share the same rank, the same dimensionality in each index, and the same variance structure across corresponding indices. This requirement ensures that the resulting sum is a well-defined tensor that transforms consistently under changes of basis, rather than an arbitrary array of numbers with no coherent transformation law.

Definition and Motivation

Formal Statement

Given two tensors $$A$$ and $$B$$, their sum $$C=A+B$$ is defined component-wise only when both tensors possess an identical index structure. For a tensor of type $$(p,q)$$, meaning it has $$p$$ contravariant indices and $$q$$ covariant indices, addition with another tensor is permitted only if that second tensor is also of type $$(p,q)$$, and both tensors are defined over the same underlying vector space and its dual.

Why the Requirement Exists

Tensors are not merely collections of numbers; they are geometric or physical objects whose components change in a prescribed way when the coordinate basis changes. If two tensors had different rank or different index placement, their components would transform differently under a change of basis, so adding their components in one basis would not correspond to adding the same "objects" in another basis. The same type requirement guarantees that the transformation law of the sum matches the transformation law of each addend, preserving the tensorial nature of the result.

Component-Level Description

Matching Indices

For tensors expressed in component form, addition proceeds index by index:

Every upper index position on the left must correspond to an upper index position of the same dimension on both operands, and likewise for every lower index position. There is no operation defined that adds, for instance, a covariant index slot to a contravariant one.

Dimension Consistency

Beyond rank and variance, the specific dimension of each index must agree. A tensor built over a three-dimensional vector space cannot be added to a tensor built over a four-dimensional vector space, even if both happen to have the same rank, because the number of components and the transformation matrices involved differ.

Consequences of Violating the Requirement

Loss of Tensorial Behavior

If addition were attempted between tensors of different type, such as a vector and a covector, the resulting array might be numerically computable in one coordinate system but would fail to transform correctly when the basis changes, since a Jacobian matrix would need to act differently on each part of the sum. The resulting object would not be a tensor at all.

Undefined Operations

When ranks or dimensions differ outright, the operation is simply undefined, since there is no natural correspondence between the components of the two arrays. There is no canonical way to align a rank-2 tensor's components with a rank-3 tensor's components for the purposes of addition.

Structural Implications

Vector Space of Tensors

The same type requirement is what allows the set of all tensors of a fixed type $$(p,q)$$ over a given vector space to form a vector space in its own right. Addition is closed within this set precisely because every tensor of that type has matching indices with every other, and the sum of any two remains of the same type.

Compatibility with Scalar Multiplication

This requirement works together with scalar multiplication to satisfy the vector space axioms: closure under addition, associativity, commutativity, existence of a zero tensor, and existence of additive inverses all depend on the fact that every tensor considered belongs to the same fixed type.

Illustration