15.15.5 Tensor Symmetric Matrix Tensor Role

Tensor Symmetric Matrix Tensor Role is the account of how, and in what precise sense, a symmetric matrix is itself a tensor: the identification of the abstract, coordinate-free multilinear object with the concrete two-index array, and the tracking of how tensorial operations such as contraction, change of basis, and the tensor product specialize when applied to this order-two, symmetric case.

The Matrix as a Tensor Object

Coordinate-Free Definition

Abstractly, a tensor of order two on a vector space V is an element of the tensor product space V tensor V, or, when built from the dual space, an element of the space of bilinear functionals on V. A matrix is the coordinate representation of such an object once a basis for V has been chosen: the matrix entries record the coefficients of the tensor when expanded in the basis induced on the tensor product space by the chosen basis of V. The Tensor Role of a symmetric matrix is precisely this: it is the numerical shadow, relative to a basis, of a coordinate-free symmetric tensor living in the symmetric part of V tensor V.

Valence and the Distinction Between Types of Matrices

Depending on whether the two tensor factors are drawn from V or from its dual space V-star, a two-index array can represent several different tensorial objects: a bilinear form (both indices dual, i.e., covariant), a linear map from V to itself (one index of each type, mixed), or a bivector (both indices from V, i.e., contravariant). The Matrix Case of symmetric tensor theory concerns specifically the fully covariant or fully contravariant situation, where both indices transform the same way under change of basis, since only in that setting does the Component Constraint of symmetry, and its transformation law under congruence, apply without alteration.

Transformation Behavior

Congruence versus Similarity

When both indices of the tensor are covariant, a change of basis given by a matrix P transforms the components by congruence,

whereas a mixed tensor representing a linear map transforms instead by similarity,

These two transformation rules coincide only when P is orthogonal, which is why the eigenvalue theory of a symmetric matrix, viewed as a linear map, and the congruence classification of the same array, viewed as a bilinear form, become intertwined precisely in the orthonormal-basis setting used by the spectral theorem.

Preservation of Symmetry Under Congruence

The Component Constraint is preserved under congruence but not, in general, under similarity: if T is symmetric, then P-transpose T P is again symmetric for any P, while P-inverse T P need not be. This is a structural reason the covariant, congruence-transforming reading of a symmetric matrix is the one compatible with, and required by, its role as a genuine symmetric tensor.

Contraction and the Tensor Product Structure

Contraction with Vectors

Treating a symmetric matrix as a tensor allows it to be contracted with vectors to produce lower-order tensors: contracting with a single vector x produces a covector (or, after using an inner product to identify V with its dual, another vector), and contracting with two copies of x produces the associated quadratic form, recovering the Bilinear Form Relation as the special case of double contraction against equal arguments.

Embedding in the Full Tensor Product

The symmetric matrices form a linear subspace of the full tensor product space V tensor V, complementary to the space of antisymmetric (skew) tensors of order two, since every order-two tensor decomposes uniquely as a sum of its symmetric and antisymmetric parts:

The Tensor Role of the symmetric matrix is realized concretely as the first summand of this decomposition, the symmetrization of T, which is exactly the projection of a general order-two tensor onto the symmetric subspace discussed throughout Tensor Symmetric Decomposition Structure.

Consistency With Higher-Order Symmetric Tensor Theory

The Symmetrization Operator

The general symmetrization operator, which averages a tensor's components over all permutations of its indices to produce a symmetric tensor of the same order, reduces, for order two, to the averaging of a matrix with its own transpose shown above. This identifies the abstract symmetrization procedure used throughout symmetric tensor theory with the elementary operation of extracting the symmetric part of a matrix, confirming that the Matrix Case's Tensor Role is not an isolated fact about matrices but the smallest instance of a uniform, order-independent construction.

Bridge to Decomposition and Rank Theory

Because the symmetric matrix is recognized as a genuine tensor, all general tensor-theoretic notions, including tensor rank, border rank, and decomposition into pure power forms, apply to it directly, and the Tensor Role clarifies precisely why the specialized results of the Matrix Case, such as the equality of symmetric and ordinary rank and the availability of the spectral theorem, must be understood as particular consequences of general tensor theory at order two rather than as facts belonging to linear algebra alone.