1.11.3 Summation Convention

Summation Convention is the general category of agreed-upon rules that determine, for a given piece of tensor algebra, exactly when a repeated index triggers an implicit sum, what range that sum runs over, and how exceptions to the default rule are signaled, of which the Einstein convention is the single most widely adopted specific instance. Where the Einstein convention itself fixes one particular rule, repetition of an index once upper and once lower implies summation, summation convention as a broader topic covers the surrounding decisions that any such rule requires: the bounds of summation, how genuine exceptions are marked, and how alternative repeated-index rules differ from the Einstein default in contexts where it is not adopted wholesale.

A summation convention, in general, is any agreement that allows a sum to be written without an explicit summation symbol by inferring, from some structural feature of an expression, most commonly index repetition, that a sum is intended. Adopting such a convention trades a small amount of initial unfamiliarity for a substantial and lasting gain in the compactness and readability of long tensor computations.

The Scope of the Convention

What Counts as "Repeated"

The convention requires precisely specifying what counts as a repeated index for summation purposes: within the standard Einstein rule, only a pairing of one upper occurrence and one lower occurrence of the same label, within a single multiplicative term, counts; two upper occurrences, two lower occurrences, or occurrences split across separately added terms do not trigger summation under the default rule, and any of these situations occurring unintentionally typically signals an expression error.

The Range of Summation

The convention must also fix the range over which a triggered sum runs, ordinarily every value from the chosen lower bound of the index range to its upper bound, matching the dimension of the underlying vector space. In contexts distinguishing a restricted subset of coordinates, such as spatial indices within a spacetime index range, the summation convention may specify a narrower range for indices drawn from a particular alphabet, with the full range reserved for indices drawn from another.

Suppressing the Convention When Needed

The "No Sum" Notation

Occasions arise where an index is genuinely repeated in the upper-lower pattern that would normally trigger summation, but a sum is not intended, most often when discussing one specific component rather than the general pattern. Convention handles this by an explicit disclaimer, commonly the parenthetical annotation "no sum," placed beside the expression, overriding the default rule for that specific instance.

Explicit Summation Symbols as a Fallback

When an intended sum does not match the automatic trigger pattern, such as a sum over an index that appears three times, or a sum that should run over only part of the usual range, convention falls back to writing an explicit summation symbol with its bounds stated, temporarily suspending the implicit convention for that particular expression so the intended, non-standard sum can be communicated unambiguously.

Variants of the Convention

DeWitt and Extended Summation Conventions

Beyond the standard Einstein rule, extended summation conventions have been proposed for contexts involving continuous or functional indices, such as the DeWitt convention used in field-theoretic settings, in which a repeated "index" may range over a continuum rather than a finite discrete set, and the implicit sum is understood to denote an integral rather than a finite sum. This extension preserves the compact notational spirit of the original convention while adapting its meaning to a setting where the underlying index set is no longer finite.

Contexts That Do Not Adopt the Convention

Some treatments, particularly introductory ones or those working exclusively in low dimension, choose not to adopt any implicit summation convention at all, writing summation symbols explicitly throughout to avoid requiring readers to track free versus dummy indices. This choice trades notational compactness for explicitness, and recognizing that a given source has made this choice is itself part of correctly interpreting its notation.

Interaction With Other Conventions

Dependence on Index Placement Convention

The summation convention's trigger condition, one upper occurrence paired with one lower occurrence, depends entirely on a consistently applied index placement convention; without a fixed and correctly followed rule for which indices are written as superscripts and which as subscripts, the summation convention has no reliable signal to detect, since it is placement that distinguishes a genuine dummy pair from an unrelated coincidence of matching labels.

Interaction With Index Range Convention

The bounds over which an implicitly triggered sum runs are inherited directly from whatever index range convention, one-based or zero-based, restricted or full, is in force for the specific index label involved, so the summation convention cannot be applied correctly in isolation from the surrounding range conventions already established for the document or field in question.

Why the Convention Is Treated as Foundational

A Force Multiplier for Compact Notation

The summation convention is what allows the rest of tensor algebra's index notation to remain compact even as expressions grow to involve many simultaneous contractions; without it, every tensor equation of any complexity would require a proliferation of explicit summation symbols that would substantially obscure the underlying structure the notation is meant to reveal.

A Frequent Source of Misreading When Assumed Silently

Because the convention operates invisibly, contributing no explicit symbol to the page, it is also a frequent source of misreading when a document assumes it without stating so, or when a reader accustomed to explicit summation encounters index-notation expressions for the first time; explicitly confirming which summation convention, if any, a given source has adopted is accordingly one of the first steps in reading unfamiliar tensor material accurately.