7.11.3 Tensor Higher Order Index Tuple

Tensor Higher Order Index Tuple is the ordered collection of index values that together specify a single entry within a higher-rank tensor's component array, treating the entire set of indices attached to a tensor as one composite address rather than as isolated, unrelated labels.

Defining the Tuple

An Ordered List of Index Values

For a tensor of rank (k), the index tuple is the ordered list ((i_1, i_2, \ldots, i_k)), where each (i_m) ranges independently over the basis directions from one to the dimension (n) of the space, and the order of the entries in the tuple matches the order of the tensor's argument slots.

The Tuple as a Complete Address

A single index tuple uniquely determines exactly one scalar entry of the tensor's component array, so the tuple functions as a complete coordinate address within the multi-dimensional array, analogous to how a pair of coordinates addresses a single cell in a matrix.

Counting and Enumerating Tuples

Total Number of Distinct Tuples

Since each of the (k) positions in the tuple independently takes one of (n) possible values, the total number of distinct index tuples equals the dimension raised to the power of the rank, matching exactly the total number of entries in the component array.

Lexicographic Ordering of Tuples

Index tuples are often enumerated in lexicographic order, incrementing the last position first and carrying over to earlier positions once the last position exhausts its range, providing a systematic way to list every entry of the component array exactly once.

Variance Carried by Each Tuple Position

Position-Dependent Transformation Role

Each position within the index tuple is associated with a fixed variance type, either contravariant or covariant, inherited from the corresponding argument slot of the tensor, so the tuple as a whole encodes not just numeric values but also which transformation rule applies to each of its entries.

Consistency Requirement Under Basis Change

When the underlying basis changes, every entry of the index tuple transforms according to its own assigned rule, and the complete transformation of a single component is the product of the individual factors contributed by each position of the tuple.

Tuples and Symmetry

Equivalence Classes Under Permutation

For a tensor known to be symmetric in a subset of its indices, distinct index tuples that differ only by a permutation of those particular positions address entries that are guaranteed to hold equal values, reducing the number of independent entries below the raw count of all possible tuples.

Antisymmetric Tuples and Vanishing Entries

For a totally antisymmetric tensor, any index tuple containing a repeated value in two antisymmetric positions necessarily addresses an entry equal to zero, since swapping the repeated indices must flip the sign of the entry while leaving it numerically unchanged, forcing it to vanish.

Diagrammatic Illustration

An index tuple pictured as a coordinate address pointing into a multi-dimensional array, analogous to a street address locating a single building within a grid of city blocks.

Role in Computation and Notation

Foundation for Summation Conventions

The Einstein summation convention operates on index tuples implicitly, treating any index letter that repeats once as an upper position and once as a lower position across a product of tensors as instructing a sum over that particular slot of the combined tuple.

Basis for Iteration in Algorithms

When implementing tensor operations computationally, iterating over every possible index tuple in a systematic order provides the most direct way to visit every entry of a tensor's component array exactly once, forming the backbone of loops used for entrywise operations, contractions, and tensor products.