3.2.5 Tensor Covector Tensor Slot Role

Tensor Covector Tensor Slot Role is the description of the specific function a covector performs when it occupies one lower-index position, or slot, of a general (p, q)-tensor, namely acting as the component of the tensor that reads a single designated vector argument and contributes a scalar factor to the tensor's overall multilinear output, independently of what happens in the tensor's other slots. Understanding the slot role of a covector is what makes it possible to interpret a complicated higher-rank tensor as built up, slot by slot, from elementary covector and vector actions, rather than as an opaque multi-indexed array.

A Single Slot Within a Simple Tensor

Isolating One Factor

In a simple (p, q)-tensor T = v_1 ⊗ ⋯ ⊗ v_p ⊗ ω_1 ⊗ ⋯ ⊗ ω_q, each ω_k occupies one lower-index slot, and its role, when T is evaluated as a multilinear functional on p covector arguments and q vector arguments, is to consume exactly the k-th vector argument u_k and contribute the scalar factor ω_k(u_k) to the overall product:

Each covector factor's slot role is entirely local: ω_k does not interact with u_j for j ≠ k, nor with the upper-index vectors v_i or their covector arguments η^i, so the slot structure fully separates the contributions of each factor.

Slot Role in General, Non-Simple Tensors

Extension by Linearity

For a general (p, q)-tensor T = Σ_α (simple term)_α, expressed as a sum of simple tensors, the value T(η^1, ..., η^p, u_1, ..., u_q) is the sum of the values contributed by each simple term, so the k-th lower-index slot role, reading u_k, is still well defined for T as a whole even though T itself does not factor into a single product; the slot role attaches to the position within the tensor's index structure, not to any single covector factor of a particular decomposition.

Coordinate Description of a Slot

Relative to a basis, the k-th lower-index slot of T's coordinate array T^{i_1...i_p}_{j_1...j_q} is the position occupied by the index j_k, and reading that slot with a specific vector u_k = Σ_m u_k^m e_m amounts to contracting the j_k index of T against the coordinates u_k^m of u_k, Σ_{j_k} T^{...}_{...j_k...} u_k^{j_k}, exactly reproducing the vector-covector pairing at the level of a single index.

Independence and Order of Slots

Slots Can Be Read in Any Order

Because evaluating a multilinear functional on a fixed tuple of arguments produces a single, order-independent numerical answer once the tuple itself is fixed, the covector slots of a tensor can be read, that is, contracted against their corresponding vector arguments, in any order without changing the final scalar result; only the assignment of each argument to its correct slot matters, not the sequence of contraction operations performed to reach the answer.

Partial Evaluation Leaves a Lower-Rank Tensor

Reading only some of the lower-index slots of a (p, q)-tensor, supplying vectors for u_1, ..., u_r with r < q but leaving the remaining q - r lower slots and all p upper slots unfilled, produces a (p, q - r)-tensor, since the unfilled slots retain their original multilinear-functional role while the filled slots have been converted into fixed scalar contributions absorbed into the result. This partial-evaluation behavior is what makes the slot role of individual covector factors compositional: covector slots can be consumed one at a time, at any point, without requiring the full tensor to be evaluated at once.

Distinguishing Slot Role From Variance Type

Slot Role Versus Index Position

Variance type, covariant or contravariant, determines what kind of argument a slot accepts, a vector for a lower index or a covector for an upper index, while slot role additionally specifies which particular argument, among possibly several of the same variance type, a given factor is responsible for reading. A (0, 3)-tensor ω_1 ⊗ ω_2 ⊗ ω_3 has three lower-index slots, all sharing the same covariant variance type, but each occupies a distinct slot role, reading u_1, u_2, and u_3 respectively, and permuting which covector occupies which slot generally changes the tensor, unless the tensor happens to lie in the symmetric or alternating subspace where slot permutation has a controlled effect on the result.