15.21.4 Tensor Symmetric Alternating Tensor Relation Boundary

Tensor Symmetric Alternating Tensor Relation Boundary is the precise account of how far the parallel between symmetric tensors and alternating (antisymmetric) tensors extends, given that both arise from the same permutation action on tensor power space but at opposite extremes of the partition classification, and of exactly where the decomposition theory, rank notions, and geometric pictures developed for one side stop transferring to the other.

The Shared Origin

Two Extremal Fixed-Point Sets

Both symmetric and alternating tensors are defined as fixed-point sets of the symmetric group's action on tensor power space, differing only in how the group acts: symmetric tensors are invariant under every permutation outright, while alternating tensors pick up the sign of the permutation, becoming negated under an odd permutation and left unchanged under an even one. This shared origin, discussed for the symmetric side under Tensor Symmetric Subspace Invariance, is what makes a systematic comparison between the two theories meaningful in the first place, since both are governed by the same commuting general-linear-group and symmetric-group actions.

Order-Two Coincidence

At order two, every tensor decomposes uniquely into a symmetric part and an antisymmetric part, as noted under the Tensor Role of the symmetric matrix, and the two spaces are genuinely complementary, together spanning the entire space of order-two tensors; this clean complementary relationship is the friendliest instance of the comparison between the two theories and does not, by itself, indicate how the relationship behaves at higher order.

Where the Parallel Holds

Rank-One Objects on Both Sides

Just as a symmetric tensor of rank one is a pure power form, v raised to the tensor power d, an alternating tensor of rank one is a decomposable wedge product, v_1 wedge v_2 wedge up to v_k, built from k distinct vectors; both notions of rank-one object generate, upon projectivization, a distinguished variety inside the corresponding projective space, the Veronese variety on the symmetric side and the Grassmannian (via the Plücker embedding) on the alternating side, and both varieties support a parallel theory of secant varieties governing the rank of tensors built as sums of these respective rank-one pieces.

Dimension Counting in Both Theories

The dimension of the space of alternating tensors of order k on an n-dimensional space is given by the binomial coefficient of n choose k, playing the same structural role that the binomial coefficient of n plus d minus one choose d plays for symmetric tensors under the Symmetric Power Notation, and both dimension formulas feed into analogous expected-rank computations balancing the dimension of the ambient space against the dimension of a secant variety.

Where the Parallel Breaks Down

Absence of Higher-Order Powers

An alternating tensor requires its generating vectors to be linearly independent for the wedge product to be nonzero, since repeating any vector among the k factors forces the wedge product to vanish identically; there is consequently no alternating analogue of the pure power form v raised to a power greater than one, since "raising a single vector to a wedge power" beyond the first power always gives zero. This is a sharp structural divergence from the symmetric side, where pure power forms v raised to arbitrarily high tensor powers are the central building block of the entire decomposition theory in Tensor Symmetric Decomposition Structure.

Different Notions of Rank and Genericity

Because every alternating tensor of top order (order equal to the dimension of the space) is automatically a single decomposable wedge product up to scalar, and because low-order alternating tensors have their own well-understood decomposability criteria (a two-form is decomposable exactly when its top power under the wedge product vanishes), the rank theory of alternating tensors follows a substantially different pattern than the rank theory of symmetric tensors, where the Rank Relation, Comon's Conjecture counterexamples, and the Alexander-Hirschowitz classification of exceptional cases have no direct counterpart of the same shape on the alternating side.

Diagonalization Fails to Generalize Analogously

The spectral theorem available for symmetric matrices in the Matrix Case has a different, more restrictive counterpart on the alternating side: an alternating order-two tensor, meaning an antisymmetric matrix, is instead brought to a canonical block-diagonal form under an orthogonal or symplectic change of basis, with two-by-two skew blocks rather than diagonal scalar entries, reflecting the fact that antisymmetric matrices of odd dimension are always singular and never admit a full set of one-dimensional eigenspaces in the way symmetric matrices do.

Practical Consequences of the Boundary

Methods Do Not Transfer Automatically

Apolarity theory, catalecticant matrices, and the Veronese-variety-based geometric machinery developed for symmetric tensors throughout the Geometry Role rely essentially on the existence of pure power forms and their associated homogeneous polynomials; the corresponding tools on the alternating side, built instead from the Grassmannian, Plücker coordinates, and Pfaffians, must be developed independently rather than obtained by directly substituting "antisymmetric" for "symmetric" throughout the existing theory.

Mixed Symmetry Types as the True Middle Ground

Tensors of mixed symmetry type, associated with partitions other than the two extremes and discussed under the Tensor Symmetric Tensor Representation Role, sit between the symmetric and alternating theories and inherit some features of each; understanding precisely how far the symmetric and alternating parallels extend, as surveyed here, is a necessary prerequisite for extending decomposition and rank theory further into this considerably less thoroughly developed mixed-symmetry territory.