15.11.1 Tensor Independent Symmetric Component Selection

Tensor Independent Symmetric Component Selection is the procedural act of designating, for a given totally symmetric tensor, a concrete algorithm and canonical rule that picks out exactly one index tuple from every permutation orbit, producing the specific list of index tuples whose stored values will serve as the tensor's independent degrees of freedom. While the general notion of an independent selection was introduced earlier as any non-overlapping, complete choice of orbit representatives, this treatment focuses on the selection as a procedure to be carried out, examining the algorithmic steps, the available choices of canonical rule, and the practical considerations that guide which rule is adopted in a given setting.

Framing selection as a procedure rather than only as an abstract property emphasizes that a symmetric tensor's independent representation is not unique in the abstract but requires a specific, agreed-upon convention; two different implementations of symmetric tensor storage, each valid, may use different canonical rules and therefore store their independent components in different but equally correct arrangements.

The Selection Procedure

Step One: Enumerate the Index Range

The procedure begins with the full range of index tuples available to a rank-n tensor over a d-dimensional space, each index ranging independently over 1 through d, giving d^n candidate tuples before any symmetry considerations are applied.

Step Two: Apply a Canonical Ordering Rule

A canonical ordering rule is then applied to every candidate tuple, mapping it to a single representative tuple of its orbit; the most standard choice sorts the indices within the symmetrized positions into non-decreasing order, though any rule that consistently picks one member per orbit is equally valid in principle.

Step Three: Collect Unique Representatives

Filtering the full index range down to only those tuples that are fixed points of the canonical ordering rule, meaning tuples already in canonical form, yields the final independent selection, with the count of surviving tuples matching exactly the orbit count established by the multiset formula.

Alternative Canonical Rules

Non-Decreasing Order as the Default

The non-decreasing order rule, sorting indices from smallest to largest, is the most commonly adopted canonical rule because it produces a lexicographically enumerable, easily sorted list of representatives and aligns directly with the standard monomial ordering used in the polynomial analogy for symmetric tensors.

Non-Increasing and Other Consistent Orderings

A non-increasing order rule, sorting indices from largest to smallest, works equally well as a canonical rule, since it likewise picks exactly one representative per orbit; more generally, any total order imposed on the index values, together with sorting according to that order, defines a valid canonical rule, and the choice among these alternatives is purely a matter of convention rather than correctness.

Consequences of Changing the Rule

Switching from one valid canonical rule to another changes which specific tuples appear in the independent selection but does not change the count of selected tuples, nor does it affect the correctness of the resulting reconstruction rule, provided the same canonical rule is used consistently for both selection and lookup.

Selection in the Presence of Partial Symmetry

Restricting the Procedure to Symmetrized Positions

When a tensor is symmetric only in a subset of its index positions, the selection procedure applies the canonical ordering rule exclusively to the indices occupying that subset, leaving indices in non-symmetrized positions unsorted and therefore free to vary independently across the selection.

Combined Counting for Mixed Symmetry

The resulting independent selection for such a partially symmetric tensor has a size equal to the orbit count of the symmetrized subset multiplied by the unrestricted range of the remaining positions, reflecting that selection within the symmetric block and full enumeration outside it combine multiplicatively rather than being handled by a single unified orbit count across all positions.

Practical Considerations in Choosing a Selection Procedure

Compatibility With Existing Conventions

When symmetric tensors are represented or exchanged between different computational systems, adopting a shared canonical ordering rule for selection is necessary to ensure that independent component data produced by one system can be correctly interpreted and reconstructed by another; a mismatch in canonical rule between producer and consumer would cause misinterpretation of which stored value corresponds to which orbit.

Alignment With the Polynomial and Basis Representations

Because the non-decreasing order rule aligns directly with the standard monomial basis used in the symmetric power basis construction and with the natural variable ordering of the polynomial analogy, this rule is typically preferred whenever a symmetric tensor's independent components are intended to be interpreted alongside, or converted into, an associated homogeneous polynomial representation.