13.7.5 Tensor Partial Contraction Composition Role

Tensor Partial Contraction Composition Role explains how partial contractions are combined in tensor algebra, shaping their mathematical operations and structure.

Tensor Partial Contraction Composition Role is the function that partial contraction serves in linking together separate tensors or separate stages of a computation, acting as the mechanism by which the output of one tensor operation is fed as an index-connected input into another. It characterizes partial contraction not merely as a rank-reducing operation performed in isolation, but as the connective operation that allows multiple tensors, or successive contraction steps, to be assembled into a single coherent composite expression.

Conceptual Basis

Contraction as a Connective Operation

Beyond reducing the rank of a single tensor, partial contraction can join two or more tensors by summing over an index shared between them, effectively composing separate tensorial objects into one combined object whose structure reflects the relationships established by the contracted indices.

Analogy With Function Composition

Just as function composition combines two functions by feeding the output of one into the input of another, tensor composition through partial contraction combines two tensors by feeding a free index of one into a matching index slot of another, with the contracted index disappearing from the final expression.

Composition Without Full Collapse

Because the composition role of partial contraction leaves at least one index free, the result of the composition remains a tensor capable of participating in further operations, distinguishing this role from a terminal full contraction that removes all indices and ends the chain of composition.

Formal Description

Composing Two Tensors Through a Shared Index

Given tensors $A^{i}$ and $B_{i j}$ , partial contraction over the shared index $i$ composes them into:

C_{j} = A^{i} B_{i j}

where $j$ remains free, so $C_{j}$ is available for use in subsequent operations.

Chained Composition Across Several Tensors

The composition role extends naturally to chains of tensors linked by successive shared indices, such as:

D_{k} = A^{i} B_{j}^{i} C_{j k}

where each contracted index links one factor to the next, forming a composed sequence analogous to a chain of matrix multiplications.

Matrix Multiplication as a Special Case

The composition role of partial contraction reduces to ordinary matrix multiplication when applied to two rank-two mixed tensors, contracting the shared index between them:

P_{k}^{i} = A_{j}^{i} B_{k}^{j}

illustrating how matrix products are themselves an instance of tensor composition through contraction.

Properties of the Composition Role

Associativity of Chained Composition

When several tensors are composed in sequence through distinct contracted indices, the overall result does not depend on the order in which the individual contractions are carried out, provided each contraction acts on a disjoint pair of indices, mirroring the associativity familiar from matrix multiplication.

Preservation of Remaining Structure

Indices not involved in the composing contraction pass through unchanged, retaining their original variance and dimension, so the composed result's structure is fully determined by which indices were free on each input tensor and which were consumed in linking them together.

Dependence on Matching Variance

Composition through contraction requires that the shared index be contravariant on one tensor and covariant on the other, since this is the same variance-matching requirement that governs any valid contraction, whether applied within a single tensor or across multiple tensors.

Applications of the Composition Role

Building Composite Tensor Networks

The composition role underlies the construction of tensor networks, where many individual tensors are linked through shared contracted indices to represent a large composite object without explicitly expanding it into a single tensor of very high rank.

Sequential Transformations

In contexts where a tensor represents a linear transformation, composing several such tensors through contraction over shared indices represents the sequential application of those transformations, analogous to composing linear maps.

Modular Construction of Invariants

Complex scalar invariants are often built by first composing several tensors through partial contraction into an intermediate tensor, and only afterward applying a final full contraction, making the composition role a necessary intermediate step in the overall derivation of such invariants.