6.23 Tensor Component Count Relation

Tensor Component Count Relation is the formula giving the total number of scalar components needed to specify a tensor once bases have been fixed, stating that a tensor of type (p, q) built over a vector space V of dimension d has exactly d^(p+q) independent components, one for every possible assignment of a basis index to each of its p + q slots. This relation converts the abstract order-type classification of a tensor into a concrete count of numbers, and it is the quantity that determines how much storage, computation, or bookkeeping a given tensor actually requires in practice.

Deriving the Count

Each Slot Ranges Independently Over the Basis

For a vector space V of dimension d with basis {e₁, ..., e_d}, each index slot of a tensor's components T^{i₁...i_p}_{j₁...j_q} ranges independently over the d values 1 through d, regardless of whether that slot is contravariant or covariant. Since there are p + q slots in total and each ranges over d values independently, the number of distinct index combinations, and hence the number of components, is

the product of d taken p + q times, one factor for each slot.

Variance Does Not Affect the Count

Notably, the component count relation depends only on the total order p + q, not on the split between p and q; a type (2,0) tensor, a type (1,1) tensor, and a type (0,2) tensor over the same d-dimensional space all have exactly d² components. Variance affects how those components transform under a change of basis, but not how many of them there are.

Worked Examples

Low-Order Cases

A scalar, order 0, has d⁰ = 1 component regardless of d, matching the fact that a scalar is a single number. A vector or covector, order 1, has d¹ = d components, matching ordinary coordinate representations. A second-order tensor, order 2, has d² components, matching the familiar d × d matrix.

A Concrete Numerical Instance

A type (1,2) tensor over three-dimensional space, such as the Riemann curvature tensor restricted to a single point in a 3-dimensional manifold (order 4 in that specific case, but illustrating the same rule), requires 27 numbers to specify completely in a given basis — a count that grows rapidly with order even for modest dimension.

Diagram of Component Growth

Practical Consequences of the Relation

Storage and Computational Cost

Because the component count grows as d^(p+q), the memory required to store a tensor's coordinate representation, and the computational cost of operations that touch every component, scale exponentially with order for fixed dimension. This is the formal reason high-order tensors become expensive to handle numerically, and it motivates specialized techniques such as tensor decompositions that avoid ever materializing the full component array.

Independence Before Symmetry Constraints

The count d^(p+q) gives the number of components before any symmetry or antisymmetry constraints are imposed on the tensor; if a tensor is known in advance to be symmetric or antisymmetric in some of its indices, the number of truly independent components is smaller than d^(p+q), and the component count relation should be understood as an upper bound in such cases, with the exact reduced count depending on the specific symmetry pattern involved.