1.10.2 Index Tensor Notation

Index Tensor Notation is a concise mathematical notation used to represent tensors, employing indices to denote components and relationships in multi-dimensional spaces.

Index Tensor Notation is the specific notational system in which a tensor is represented by a symbol carrying one or more indices, upper for contravariant components and lower for covariant components, and in which every algebraic operation on tensors, addition, contraction, tensor product, symmetrization, is carried out as an explicit manipulation of those indices according to a fixed set of rules. It is also called Ricci calculus, after its systematic development, and it remains the primary working notation for explicit tensor computation, since it makes visible, at every step, exactly which components are being combined and how the resulting object's type is determined.

Index tensor notation differs from direct or coordinate-free notation in that it always refers to the components of a tensor relative to a specific, though arbitrary, basis, rather than to the tensor as an abstract object. This component-level commitment is what gives index notation its computational power: every manipulation reduces to arithmetic operations on labeled numbers, governed by a small number of consistent rules.

The Anatomy of an Indexed Symbol

Base Symbol and Index Slots

An indexed tensor symbol consists of a base letter, such as T, together with a specific arrangement of upper and lower index slots, such as T^i_jk, which has one upper slot and two lower slots. The number of slots, and their arrangement, is fixed once the tensor's type, (p, q) in this case (1, 2), is fixed, and every valid expression involving this symbol must respect that arrangement.

Index Labels Versus Index Values

An index label, such as i in T^i_jk, is a placeholder that ranges over the numerical values 1 through n, where n is the dimension of the underlying vector space. Writing T^i_jk refers to the entire family of components, while writing T^1_{23}, for example, refers to one specific numerical component, obtained by substituting particular values for the index labels.

T_{23}^{1} is one specific component of T_{jk}^{i}

Free and Dummy Indices

Free Indices

A free index appears exactly once in a given term, is not summed, and indicates that the term, taken as a whole, still depends on that index's value. The set and position of free indices in an expression determine the type of the resulting tensor: an expression with one free upper index and one free lower index yields a (1, 1) tensor.

Dummy Indices and the Summation Convention

A dummy index appears exactly twice in a single term, once as an upper index and once as a lower index, and by the Einstein summation convention, is automatically summed over all its values, contributing no index to the final result. Dummy indices are also called summed or contracted indices, and their specific letter is arbitrary: relabeling a dummy index throughout a term does not change its meaning.

A_{j}^{i} B^{j} = A_{k}^{i} B^{k}

The Rule Against Triple Repetition

A well-formed index expression never repeats the same index label three or more times within a single term, since the summation convention is defined only for a matched upper-lower pair; an index appearing three or more times signals an error, most often a mistaken reuse of a label that should have been given a distinct name.

Core Operations in Index Notation

Addition

Two tensors of the same type can be added only if their free indices match exactly in name, position, and variance, and the result is computed by adding corresponding components.

C_{j}^{i} = A_{j}^{i} + B_{j}^{i}

The Outer (Tensor) Product

The tensor product of two indexed tensors is written by simple juxtaposition, placing all the indices of both tensors side by side with no shared labels; the result carries every index from both original tensors, and its rank is the sum of the two original ranks.

C_{jl}^{i k} = A_{j}^{i} B_{l}^{k}

Contraction

Contraction is performed by giving one upper index and one lower index of a tensor, or of a product of tensors, the same label, which by the summation convention triggers an automatic sum over that shared label, reducing the total number of free indices by two.

s = T_{i}^{i} (full contraction of a rank-2 tensor, yielding a scalar, the trace)

Raising, Lowering, and the Metric

Converting Between Index Types

The metric tensor g_ij and its inverse g^ij are used to convert a contravariant index into a covariant one, or the reverse, through contraction. Lowering an index on v^i produces v_i = g_ij v^j; raising an index on ω_i produces ω^i = g^ij ω_j. This operation changes the notational type of a specific index without changing the underlying geometric content of the tensor, since the metric encodes exactly the information needed to identify a vector space with its dual.

g^{ik} g_{kj} = δ_{j}^{i}

Symmetrization and Antisymmetrization Notation

Round Brackets for Symmetric Parts

Enclosing a set of indices in round brackets, as in T_(ij), denotes the symmetric part of the tensor with respect to those indices, computed by averaging the tensor's components over every permutation of the enclosed indices.

T_{(i j)} = \frac{1}{2} (T_{ij} + T_{ji})

Square Brackets for Antisymmetric Parts

Enclosing a set of indices in square brackets, as in T_[ij], denotes the antisymmetric part, computed by averaging over every permutation with a sign attached according to the parity of the permutation.

T_{[i j]} = \frac{1}{2} (T_{ij} - T_{ji})

Decomposition Into Parts

Any two-index tensor can be written as the sum of its symmetric and antisymmetric parts, T_ij = T_(ij) + T_[ij], and this decomposition generalizes, with more elaborate bracket combinations, to tensors with more than two indices of the same variance, allowing index notation to express refined symmetry structure compactly.

Common Pitfalls in Index Manipulation

Mismatched Free Indices

An expression in which the free indices on one side of an equation do not match those on the other, in name, count, and upper-lower placement, is not a valid tensor equation, regardless of whether the numerical values happen to agree in a specific coordinate system; this mismatch typically signals an error in how an operation was applied.

Reusing a Dummy Index Label Across Terms

When an expression contains a sum of several terms, each term's dummy indices must be treated as local to that term; reusing a dummy label across unrelated terms in a way that creates ambiguity about which indices are meant to be paired is a frequent source of confusion, and careful relabeling avoids the issue.

Confusing Contraction with Simple Multiplication

Placing two indexed quantities side by side without a shared upper-lower index pair does not trigger summation; the summation convention applies only when an index is genuinely repeated in the required upper-lower configuration, so index notation requires precise attention to which indices are actually shared between factors.