1.10 Tensor Algebra Notation Foundations

Tensor Algebra Notation Foundations is the collection of symbolic conventions, index placement rules, and summation shorthand used to write, manipulate, and communicate tensor expressions unambiguously. Because a tensor is fundamentally characterized by how its components transform under a change of basis, and because tensor expressions routinely involve many indices summed and combined in specific patterns, a disciplined notation is not a stylistic convenience but a functional necessity: without it, the type of a tensor, the operations being performed on it, and the correctness of an expression become difficult or impossible to verify at a glance.

At the center of this notation lies the distinction between upper indices, denoting contravariant components, and lower indices, denoting covariant components, together with the Einstein summation convention, which removes the need to write explicit summation symbols by treating any index repeated once as an upper index and once as a lower index within a single term as automatically summed. These conventions, along with a small set of auxiliary symbols and rules, form the shared language in which essentially all tensor algebra is expressed.

Index Placement

Upper Indices and Contravariance

An upper index, written as a superscript such as v^i, denotes a contravariant component. Contravariant components transform with the Jacobian of the coordinate change itself, meaning they scale inversely to how the basis vectors scale, so that the underlying geometric object they describe remains invariant. Ordinary vector components, such as the components of a displacement, are conventionally written with upper indices.

Lower Indices and Covariance

A lower index, written as a subscript such as ω_i, denotes a covariant component. Covariant components transform with the inverse Jacobian, matching how basis covectors, or gradients of coordinate functions, naturally scale. Components of linear functionals, gradients, and differential one-forms are conventionally written with lower indices.

Mixed-Index Tensors

A general tensor may carry any number of upper and lower indices simultaneously, such as T^i_jk, which has one contravariant index and two covariant indices. The pattern and count of upper versus lower indices is called the type of the tensor, conventionally written (p, q), with p the number of upper indices and q the number of lower indices, and this type must be tracked precisely, since operations that are valid for one type are frequently invalid for another.

The Einstein Summation Convention

The Repeated Index Rule

The Einstein summation convention states that whenever an index appears exactly twice in a single term, once as an upper index and once as a lower index, a summation over all values of that index is implied, without writing an explicit summation symbol. This convention transforms expressions that would otherwise require an explicit sum into compact products of indexed quantities.

Free Indices Versus Dummy Indices

An index that appears only once in a term, not paired with a matching upper-lower partner, is called a free index, and it survives in the result of the expression, indicating one dimension of the output tensor's type. An index that appears twice, once up and once down, is called a dummy index, and it is entirely summed away, contributing no index to the result. In an expression such as T^i_j v^j, the index j is a dummy index that is summed over, while i is a free index, so the result is an object with a single upper index, matching a vector.

Consistency of Free Indices

In any valid tensor equation, the set of free indices appearing on the left-hand side must exactly match the set of free indices appearing on the right-hand side, in both name and position, upper or lower. This consistency requirement is one of the fastest checks available for verifying that a tensor expression is well formed, since a mismatch in free indices signals either a typographical error or a genuine inconsistency in the underlying expression.

The Kronecker Delta and the Metric

The Kronecker Delta as the Identity

The Kronecker delta, δ^i_j, equals one when i equals j and zero otherwise, and it functions as the tensor representation of the identity map: contracting any tensor's index against the Kronecker delta leaves that tensor unchanged, aside from relabeling the contracted index.

Raising and Lowering Indices with the Metric

The metric tensor, g_ij, and its inverse, g^ij, provide the standard notational mechanism for converting a contravariant index into a covariant one, or vice versa, an operation called raising or lowering an index. Contracting a vector's upper index with the metric's two lower indices produces the corresponding covector, and contracting a covector's lower index with the inverse metric produces the corresponding vector.

Notational Variants

Abstract Index Notation

Abstract index notation uses letters such as i, j, k as formal labels attached to a basis-independent tensor object, indicating its valence and the slots into which vectors or covectors may be inserted, without referring to any particular numerical coordinate system. This differs from component notation, in which the same letters range over numerical values from 1 to n and refer to a specific chosen basis.

Coordinate-Free Notation

Coordinate-free notation, such as writing a tensor product simply as u ⊗ v or a contraction as tr(T), suppresses indices entirely, emphasizing the tensor as an intrinsic object rather than an array of numbers. This notation is often preferred for stating general theorems, while indexed notation is often preferred for carrying out explicit computations, and fluency in translating between the two is a core skill built on top of these notational foundations.

Symmetrization and Antisymmetrization Brackets

Parentheses enclosing a group of indices, such as T_(ij), denote the symmetric part of the tensor over those indices, obtained by averaging over all permutations of the enclosed indices. Square brackets, such as T_[ij], denote the antisymmetric part, obtained by averaging over permutations with alternating sign. These bracket conventions provide a compact notation for the symmetrization and antisymmetrization operations that recur throughout tensor algebra.

Why Consistent Notation Matters

Verifying Correctness by Index Bookkeeping

Because tensor equations must balance in both the number and placement of free indices, careful notation allows errors to be caught mechanically: an equation with a mismatched free index, a repeated index that appears in the same position twice, or an index summed against another index of the same variance rather than the opposite variance, signals a definite error before any numerical computation is even attempted.

A Shared Language Across Applications

Because the same index and summation conventions apply uniformly regardless of the specific vector space, dimension, or application involved, tensor notation functions as a shared language across the many fields in which tensors appear, allowing expressions written in one context to be read and interpreted correctly in another without additional translation.