8.2.5 Tensor Contraction Index Area

Tensor Contraction Index Area refers to the process of summing over paired indices in tensors, reducing their rank and simplifying complex multidimensional expressions.

Tensor Contraction Index Area is the conceptual domain of tensor index notation devoted to contraction: the operation of pairing one superscript index with one subscript index on a tensor, or across a product of tensors, and summing over that pair to produce a tensor of reduced rank. It encompasses the notational mechanics by which contraction is expressed — through the repetition of a single letter, once up and once down — together with the algebraic consequences of performing it.

What Contraction Does

Rank Reduction

Contraction takes a tensor of type (p, q) and produces a tensor of type (p − 1, q − 1), by summing over one chosen superscript slot and one chosen subscript slot simultaneously. The two contracted slots are removed from the result entirely; the remaining p − 1 superscript and q − 1 subscript slots persist as free indices labeling the reduced tensor.

T_{j}^{i} \mapsto T_{i}^{i} = scalar

A rank-two mixed tensor contracted over its single superscript-subscript pair produces a scalar, commonly called the trace of the tensor.

The Repeated-Letter Mechanism

Contraction is signaled in the notation purely by writing the same letter in the chosen superscript and subscript slots; no separate contraction symbol is needed, since the Einstein summation convention automatically interprets a one-up-one-down repeated letter as an instruction to sum over it, which is precisely the operation contraction performs.

Contraction Within a Single Tensor Versus Between Factors

Self-Contraction (Trace)

When the contracted superscript and subscript both belong to the same tensor object, the operation reduces that one tensor's own rank internally, producing what is often called a trace when the result is a scalar, or a partial trace when higher-rank indices remain free.

Contraction Between Two Factors

When the contracted superscript belongs to one tensor and the paired subscript belongs to a second tensor multiplied alongside it, contraction serves as the mechanism for combining the two tensors into a single new tensor, as in matrix-vector multiplication w^{i} = T^{i}_{j} v^{j}, where the dummy index j links a subscript slot of T to the superscript slot of v.

w^{i} = T_{j}^{i} v^{j}

Diagram of the Contraction Index Area

Multiple and Successive Contractions

Contracting More Than One Pair at Once

A tensor with several available superscript-subscript pairs may have more than one pair contracted within the same term, each pair using its own distinct repeated letter, producing a single combined reduction in rank equal to twice the number of pairs contracted.

T_{i j}^{i j} = scalar

Sequencing Successive Contractions

When more contractions are needed than can be performed in a single term, they are carried out in a sequence of steps, each reducing the rank further, with each step's specific contracted pair identified anew by its own repeated letter, chosen fresh so as not to clash with letters still in use from earlier or later steps in the sequence.

Invariance of Contraction

A Basis-Independent Operation

Although contraction is performed using components relative to a chosen basis, the resulting tensor is independent of that basis choice: contracting the same pair of slots in any other basis produces the components, in that basis, of the very same reduced tensor, since contraction is definable directly from the tensor's universal property without reference to any particular basis.

Why This Matters for the Notation Area

This basis-independence is what justifies treating contraction as a legitimate tensor operation rather than a mere numerical manipulation of an array of numbers; the contraction index area therefore sits squarely within tensor algebra proper, rather than being a computational shortcut applicable only in a specific coordinate system.

Common Named Uses of Contraction

Trace and Determinant-Related Quantities

The trace of a linear operator, expressed as T^{i}_{i}, is the most familiar named instance of contraction, and it recurs throughout tensor calculus as an invariant scalar built from a mixed tensor.

Inner Products and Norms

Contracting a vector with itself after lowering one of its indices via the metric, g_{ij} v^{i} v^{j}, produces the squared norm of the vector; this combination of raising or lowering with contraction is one of the most frequently used constructions built from the contraction index area throughout physical and geometric applications of tensor algebra.