6.2.5 Tensor Slot Classification Area

Tensor Slot Classification Area explains how tensors structure data via slots, key for algebraic operations and mathematical modeling.

Tensor Slot Classification Area is the branch of tensor classification that examines each individual argument slot of a tensor on its own terms, recording for every slot its variance, the specific vector space it draws its values from, and whether the slot is free or bound by a contraction, rather than aggregating slots into a single count, split, or sequence. Where slot counting produces one arithmetic total and valence produces one ordered sequence of variances, slot classification produces a separate descriptive record for each slot individually, making it the finest-grained of the areas concerned with a tensor's argument structure.

What a Single Slot Is Classified By

Variance of the Slot

Each slot is classified first by variance, contravariant (upper) or covariant (lower), exactly as in type and valence classification, but here the classification is attached to one named slot at a time rather than summarized across all slots at once.

slot m of T : variance \in {↑, ↓}

Associated Vector Space

Each slot is classified second by which vector space supplies its values; for a tensor built over a single space V every slot draws from V or its dual V*, but for a more general multilinear map T : V₁ × ... × V_k → F built from several distinct spaces, slot classification records exactly which V_i each argument position belongs to.

Free Versus Bound Status

Each slot is classified third as either free, meaning it still accepts an argument and appears as a live index in expressions involving T, or bound, meaning it has already been paired with another slot through contraction and summed over, so that it no longer appears as a free index in the resulting expression.

Diagram of a Fully Classified Slot Record

Slot Classification for Multi-Space Tensors

Beyond Tensors Built From a Single Space

Slot classification is essential once a tensor is treated as a general multilinear map T : V₁ × V₂ × ... × V_k → F drawing on distinct vector spaces rather than repeated copies of one space V; in this general setting, order alone does not even specify which space each slot belongs to, only how many slots there are.

Recording Slot-to-Space Assignment

slot (m) \mapsto V_{m}

is within scope as the explicit function assigning each numbered slot to the specific vector space it draws from, a piece of information that order, type, and valence classification do not track since they only concern variance and count.

Free and Bound Slots Under Contraction

How Contraction Reclassifies Slots

Before a contraction is performed, both slots involved are classified as free; after the contraction, both are reclassified as bound, and they are removed from the list of slots that still appear as live indices in the contracted expression.

\sum_{i} T_{i}^{i}

illustrates a single contraction: before summation, the two i indices are two distinct free slots; the summation binds them together, and neither remains a free slot in the resulting scalar.

Distinguishing a Free Index From a Dummy Index

A free slot corresponds to what is often called a free index, one that ranges independently and appears on both sides of an equation involving the tensor; a bound slot corresponds to a dummy (or summation) index, one that is summed over and does not appear in the final free-index list of the result.

Distinguishing Slot Classification From the Other Areas

Slot Classification Versus Slot Counting

Slot counting reduces all slots to a single integer total; slot classification instead preserves and labels every individual slot, making it possible to answer questions such as "which slot is bound" that a bare count cannot address.

Slot Classification Versus Valence

Valence records the ordered sequence of variances across all slots but assumes every slot draws from the same underlying space (or its dual); slot classification additionally records the specific space each slot belongs to and its free-or-bound status, information valence does not carry.

Slot Classification Versus Type

Type aggregates variances into the two totals p and q; slot classification keeps every slot's variance, space, and status separately labeled, making it a strictly more detailed record than type from which type, order, and valence can each be recovered by aggregation.

Why Slot Classification Is Useful

Necessary for Multilinear Maps Over Several Spaces

Any tensor formalism that admits multilinear maps over genuinely distinct vector spaces, rather than only repeated tensor powers of one space, requires slot classification to keep track of which argument belongs to which space, a need that does not arise when every slot is known in advance to draw from the same V.

Precise Bookkeeping for Complex Contraction Patterns

When a computation involves several contractions performed in sequence, tracking each slot's free-or-bound status individually prevents a slot from being mistakenly contracted twice or a bound slot from being mistaken for a still-free one, a bookkeeping precision that coarser classification areas do not provide.