6.3.3 Tensor Order Component Index Count

Tensor Order Component Index Count is the same integer k as the tensor order slot count and argument count, considered this time from the concrete perspective of addressing a single stored number in a tensor's component array: it is the number of separate integer indices, (i₁, i₂, ..., i_k), that must be supplied together to identify exactly one component once a basis has been fixed for every argument. Where the slot count comes from abstract index notation and the argument count comes from the multilinear-map definition, the component index count is the practical fact a programmer or explicit calculation must respect when reading or writing one entry of the array.

Indexing a Single Component

The Index Tuple Required to Address a Component

A single component of an order-k tensor is written T_{i₁...i_k} (mixing upper and lower positions as appropriate) and is addressed by supplying a full tuple of k integers, one per index position, each ranging from 1 up to the dimension of the vector space supplying that slot:

Supplying fewer than k integers leaves the component under-specified; supplying more than k is meaningless, since there is no slot left to receive the extra index.

Nested-Loop Enumeration of All Components

Within this perspective, enumerating every component of an order-k tensor requires exactly k nested loops, one per index position, each looping over the dimension of its corresponding space; the total number of loop iterations, and hence the total number of components, is the product of the k dimensions.

Diagram of Nested Loops for k = 3

Why This Count Is the Same k as Slot Count and Argument Count

One Free Index Symbol per Loop Variable

Each free index symbol appearing in T_{i₁...i_k} corresponds to exactly one loop variable needed to enumerate the array, so the component index count read off an explicit array traversal always equals the slot count read off the abstract index notation for the same tensor.

One Loop Variable per Multilinear Argument

Each loop variable also corresponds to exactly one argument position of the underlying multilinear map, since fixing all k loop variables to specific integer values amounts to evaluating the map at one specific choice of basis vector in each of its k argument slots; this is why the component index count also agrees with the argument count.

Practical Consequences of the Component Index Count

Determining the Shape of a Stored Array

The component index count k fixes how many dimensions the stored array must have; a data structure representing an order-3 tensor must be a 3-dimensional array, never a flattened 1-dimensional array pretending to be 3-dimensional without the corresponding index arithmetic to recover each of the three original indices.

Flattening a Multi-Index Into a Single Offset

is the formula, dependent on knowing k exactly, used to convert a full index tuple (i₁, ..., i_k) into a single linear memory offset when the component array is stored in contiguous, flattened form, a computation performed identically regardless of whether the tensor's order arises as a slot count or an argument count.

Cost of Reading One Component Scales With k Only in Loop Depth

Reading a single component, once its full index tuple is known, is a constant-time operation regardless of k; it is only the enumeration of all components, or the search for a component matching a partial index specification, whose cost scales with k through the depth of nested iteration required.

Relation to the Broader Order Classification

The Same Integer, Three Complementary Views

Slot count, argument count, and component index count are three names for computing the identical integer k from three complementary viewpoints — index notation, multilinear-map arity, and array-addressing dimensionality — with the choice of which name to use depending only on which viewpoint is most convenient for the discussion at hand, not on any actual difference in the underlying quantity.

A Concrete Anchor for Order Classification

Because the component index count is the version of k most directly tied to explicit computation with real numbers stored in memory or on paper, it serves as the concrete anchor connecting the abstract naming scheme of order classification to the everyday practice of writing out, reading, and computing with actual tensor components.