8.21.2 Tensor Component to Index Translation

Tensor Component to Index Translation is the reverse direction of the correspondence between tensors and their numerical arrays: given a table of components already computed or measured in a fixed basis, it is the process of assigning the correct index letters, the correct number of them, and the correct upper-or-lower placement for each, so that the resulting index expression faithfully represents the same object and transforms correctly when the basis is changed. Unlike the forward translation, which starts from a known abstract tensor and evaluates it to get components, this direction starts from bare numbers and must recover or verify the tensorial structure — the order, and the variance of each slot — that those numbers actually possess.

The Core Problem: Numbers Alone Do Not Reveal Variance

Two Arrays Can Look Identical Yet Transform Differently

A square array of numbers could represent a (2,0) tensor, a (1,1) tensor, or a (0,2) tensor; nothing in the bare list of numbers distinguishes these cases, since all three have the same number of components in the same shape. Component-to-index translation therefore cannot proceed from the numbers alone — it requires independent knowledge of how the array transforms under a change of basis, which is the actual criterion that fixes how many indices should be upper and how many should be lower.

is the transformation expected of a (1,1) array under a Jacobian J, and an array is correctly translated as T^i_j only if it is actually observed or known to transform this way, rather than with two factors of J or two factors of J⁻¹.

Determining Variance From the Transformation Law

The practical procedure for component-to-index translation is to test, or to already know from the context in which the array arose, how each slot changes under a basis change: a slot that transforms with the Jacobian J is assigned an upper index, and a slot that transforms with the inverse Jacobian J⁻¹ is assigned a lower index. Only after this transformation behavior has been established, slot by slot, can the correct index expression be written down.

Translating From a Fixed-Basis Computation Back to General Notation

Recovering the General Formula From a Specific-Basis Result

A calculation carried out entirely in one specific basis — for instance, components computed directly in an orthonormal Cartesian frame — often loses any visible distinction between upper and lower indices, since in an orthonormal basis the two coincide numerically. Component-to-index translation, in this setting, means re-deriving or re-stating the same computation with the correct upper/lower placement restored, so that the resulting expression remains valid after a change to a non-orthonormal or curvilinear basis, where upper and lower components generally differ.

Example: Recognizing a Contraction Hidden in a Sum

Given only a computed formula such as s = Σᵢ aᵢbᵢ obtained numerically, without any accompanying index-notation derivation, component-to-index translation requires recognizing which of aᵢ and bᵢ transforms contravariantly and which transforms covariantly (or that both are components of ordinary vectors in an orthonormal basis, in which case one must be reinterpreted with the metric) before the sum can be correctly rewritten as an index contraction a^i b_i rather than left as an unlabeled numerical accumulation.

Assigning Free Versus Dummy Index Letters

Distinguishing an Index That Ranges Freely From One Already Summed

When translating a component array — or a formula that combines several component arrays — into index notation, it must also be determined which indices survive into the final result as free indices (varying independently in the output) and which are already summed away in producing the given numbers, since a sum already carried out numerically corresponds to a dummy index that should not reappear as a free index in the translated expression.

Choosing Distinct, Unambiguous Letters

Once the free and dummy structure of a given numerical computation is understood, the actual choice of index letters is a matter of convention rather than content, subject only to the requirement that no letter is reused in a way that would create an unintended additional summation; component-to-index translation therefore also involves selecting index letters that do not collide with letters already used elsewhere in a larger surrounding expression.

Diagram of Recovering Index Structure From Raw Numbers

Consequences of an Incorrect Translation

Silent Errors in Later Symbolic Manipulation

Assigning the wrong variance during component-to-index translation does not produce an obviously invalid expression; it produces an expression that looks well-formed but is false as soon as a coordinate transformation, contraction, or further tensor operation is applied to it, since the mistranslated index will not cancel or combine correctly with matching indices elsewhere. This is the principal risk that makes correct component-to-index translation important: the error is invisible until the resulting expression is tested against an actual change of basis or combined incorrectly with a genuinely tensorial quantity.

Reconciling With Independently Known Identities

A practical safeguard against mistranslation is to check the proposed index expression against any independently known identity the underlying quantity must satisfy — for example, that a proposed metric-derived quantity must be symmetric, or that a proposed curvature-derived scalar must be coordinate-independent — since an incorrect variance assignment will typically violate such a known property even when the raw numbers were computed correctly in the original fixed basis.