13.15.1 Tensor Contraction Slot Count Reduction

Tensor Contraction Slot Count Reduction is a method in tensor algebra that simplifies expressions by reducing the number of slots through contraction operations.

Tensor Contraction Slot Count Reduction is the property that describes how many index slots, viewed as the labeled positions available for tensor indices, disappear from a tensor's multilinear signature each time a contraction is carried out. It refines the general notion of order reduction by focusing specifically on the bookkeeping of slots rather than on the abstract order number alone.

Definition

A tensor of type $(p, q)$ can be viewed as a multilinear map with $p$ covector slots and $q$ vector slots:

T : V^{*} \times \dots \times V^{*} \times V \times \dots \times V \to ℝ

with $p$ copies of $V^{*}$ and $q$ copies of $V$ . Contraction removes exactly one slot of each kind, so the slot count reduction per contraction is:

Δ p = - 1, Δ q = - 1

Slot Identity Before and After

Slot Labeling

Before contraction, the tensor has $p + q$ distinguishable slots, each accepting either a vector or covector argument depending on its variance. Contraction selects one upper slot and one lower slot and fuses them: instead of accepting two independent arguments, that pair of slots is saturated internally by the summation over a shared basis.

Slot Removal Mechanism

The fused pair does not become a single new slot; it is eliminated entirely from the map's domain. The remaining slots retain their original labels and variance, unaffected in kind by the removal, only shifted in position if a strict left-to-right slot ordering is maintained.

Counting Behavior Across Multiple Contractions

Single Contraction

slots after = (p + q) - 2

k Independent Contractions

slots after = (p + q) - 2 k

This holds as long as each of the $k$ contractions consumes a distinct upper slot and a distinct lower slot, so that no slot is counted toward more than one contraction.

Slot Exhaustion Limit

Slot count reduction cannot proceed past the point where either the upper or lower slot supply reaches zero. If $p < q$ , at most $p$ contractions are possible before the tensor has only lower slots remaining, with slot count:

(p + q) - 2 p = q - p

Diagram of Slot Fusion

Distinction From Order Reduction

Slot count reduction and order reduction describe the same numerical drop, but slot count reduction emphasizes structural bookkeeping: which specific argument positions vanish and which persist, and how the remaining slots retain their original variance and relative ordering, which matters when tracking how a contracted tensor continues to act as a multilinear map on the remaining arguments.