12.15.1 Tensor Direct Sum Factor Selection

Tensor Direct Sum Factor Selection is the act of designating which particular summand Vₖ among the constituents V₁, V₂, ..., Vₙ of a direct sum V₁ ⊕ V₂ ⊕ ... ⊕ Vₙ is being referenced, extracted, or operated on at a given moment, together with the bookkeeping needed to keep that designation unambiguous as computations proceed. Because every element of a direct sum decomposes uniquely into one component per summand, any operation that concerns "the V₂-part" or "the third factor" of such an element requires first fixing precisely which index in the ordered list of summands is intended, and factor selection is the formal name for this indexing act.

The Selection Mechanism

Selection via the Index Set

A direct sum is indexed by a set I, finite or infinite, and factor selection consists of choosing one specific index k ∈ I to designate a particular summand Vₖ. The direct sum's own definition already fixes what I is and which space is assigned to each index, so factor selection does not create new structure — it identifies which piece of the existing structure is currently of interest.

Selection Realized as the Canonical Projection

Concretely, selecting factor k of an element v = (v₁, ..., vₙ) is carried out by applying the canonical projection πₖ associated with the direct sum, immediately connecting factor selection to the tensor projection operation studied more generally elsewhere: selecting a factor is precisely the special case of projection where the projection in question is one of the direct sum's own coordinate projections.

Uniqueness and Well-Definedness of Selection

Uniqueness of the Decomposition

Because every element of a direct sum has a unique representation as a tuple (v₁, ..., vₙ) with vₖ ∈ Vₖ, factor selection always produces a single, unambiguous answer: there is exactly one value πₖ(v) for each choice of v and k, never several competing candidates. This uniqueness is a direct consequence of the direct sum being a genuine direct sum rather than merely a sum of overlapping subspaces.

Contrast with Non-Direct Sums

If W₁ and W₂ overlap nontrivially inside an ambient space so that their sum W₁ + W₂ is not direct, an element of W may admit several different decompositions as w₁ + w₂, and factor selection becomes ill-defined: "the W₁-component" of w is no longer a single determined value. The direct sum operation is specifically the construction that rules this ambiguity out by design, which is why factor selection is meaningful only relative to a genuine direct sum decomposition.

Diagram of Factor Selection Across Several Summands

Factor Selection in Computation

Reducing Multi-Factor Problems to Single-Factor Problems

A common technique is to select each factor in turn, solve or analyze a problem entirely within the selected summand Vₖ using only that summand's own structure, and then reassemble the results across all selected factors using the canonical inclusions. Factor selection is what makes this decompose-and-reassemble strategy precise, since it guarantees no information is lost or duplicated between the separately handled summands.

Selection After Composition of Direct Sums

When a direct sum is itself built from further direct sums, V = (V₁ ⊕ V₂) ⊕ V₃, factor selection can proceed in stages: first selecting between the outer summands (V₁ ⊕ V₂) and V₃, and then, if the first outer summand was selected, selecting further between V₁ and V₂. This nested selection corresponds to composing the relevant canonical projections, and the final result agrees with directly selecting the appropriate leaf summand in a fully flattened three-fold direct sum, by associativity of the direct sum construction.

Selection Combined with Basis Coordinates

If each summand Vₖ additionally has its own internal basis, full identification of an element of the direct sum requires two levels of selection: first the summand index k, then the coordinate index within Vₖ's own basis. Both levels together specify a single coordinate of the direct sum relative to the concatenated basis formed from all the summands' individual bases.

Relationship to Other Tensor Operations

Selection as a Special Case of General Projection

Every factor-selection projection πₖ associated with a direct sum satisfies the linear map role and the idempotency condition required of tensor projections in general, so factor selection inherits all of the general theory of tensor projection operations without needing separate justification, while additionally enjoying the extra guarantee of mutual orthogonality among the different πₖ inherent to the direct sum's construction.

Selection Paired with Reinsertion via Inclusion

Selecting factor k and then reinserting it via the canonical inclusion ιₖ reproduces exactly the k-th summand's contribution embedded back inside the full direct sum, ιₖ(πₖ(v)), while zeroing out every other summand's contribution. Summing this reinsertion over every index k ∈ I reconstructs the original element v exactly, confirming that factor selection together with reinsertion accounts for the entirety of the direct sum's structure with nothing left over.