9.2.5 Tensor Basis Dependence Area

Tensor Basis Dependence Area explores how tensor properties rely on chosen bases, shaping their representation and transformation across mathematical frameworks.

Tensor Basis Dependence Area is the classification of the different quantities encountered in tensor calculation according to how strongly their numerical value depends on the choice of basis or coordinate system — ranging from individual tensor components, which are fully basis-dependent and change under any change of basis, through non-tensorial auxiliary quantities such as connection coefficients, which depend on the basis in a more complicated, non-tensorial way, to fully contracted scalar invariants, which are entirely basis-independent. It organizes quantities by this single question — how does this number change if the basis is changed — so that the correct expectations are brought to each type of quantity before it is manipulated or compared across different calculations.

The Fully Basis-Dependent Area: Individual Tensor Components

Components Change Predictably but Completely

An individual tensor component, such as T^i_j, changes under a change of basis according to the standard tensor transformation law, picking up one Jacobian factor for each upper index and one inverse-Jacobian factor for each lower index; the change is total in the sense that essentially no combination of components is guaranteed to stay fixed, other than in the special case of a fully contracted expression.

{\bar{T}}_{j}^{i} = \frac{\partial {\bar{x}}^{i}}{\partial x^{k}} \frac{\partial x^{l}}{\partial {\bar{x}}^{j}} T_{l}^{k}

Predictability Despite Full Dependence

Although individual components are fully basis-dependent, the manner of that dependence is entirely predictable and governed by a single, universal rule (the tensor transformation law), which is precisely what distinguishes this area from the next: full basis dependence combined with a simple, uniform transformation rule is a very different situation from basis dependence that follows no such simple rule at all.

The Non-Tensorially Basis-Dependent Area

Connection Coefficients as the Central Example

Christoffel symbols and other connection coefficients depend on the basis in a way that does not follow the ordinary tensor transformation law; their transformation includes an extra inhomogeneous term involving second derivatives of the coordinate change, on top of the terms a genuine tensor would have. Quantities in this area are basis-dependent in a strictly more complicated sense than ordinary tensor components, since no single Jacobian-based rule alone describes how they change.

{\bar{Γ}}_{j k}^{i} = (tensor-like terms) + (second-derivative term)

Why This Area Requires Special Handling

Because quantities in this area do not transform as tensors, they cannot be substituted freely into formulas that assume tensorial behavior — most notably, they cannot be treated as ordinary tensor components when checking whether a proposed identity has tensorial meaning — and calculations involving them typically combine several non-tensorial pieces in a specific way engineered so that the inhomogeneous terms cancel, restoring an overall expression that does transform tensorially even though its individual ingredients do not.

The Fully Basis-Independent Area: Scalar Invariants

Fully Contracted Quantities Carry No Remaining Basis Dependence

A quantity obtained by contracting every free index of a tensor expression away entirely — a trace, a fully contracted curvature scalar, a squared norm — is a scalar that takes the identical numerical value in every basis, since the Jacobian factors from each contracted pair cancel completely; quantities in this area have, by construction, no basis dependence left to track at all.

The Role of This Area as a Terminus

Basis-independent scalars function as the terminal, most information-poor but most directly comparable output of a tensor calculation: while individual components and non-tensorial auxiliary quantities carry more detailed information, only fully basis-independent scalars can be compared directly as plain numbers across calculations carried out in entirely different bases or coordinate systems without first applying any transformation.

Diagram of the Three Basis Dependence Areas

Practical Use of the Classification

Choosing Which Area a Comparison Belongs To

Before comparing a numerical result across two different bases or coordinate systems, identifying which basis dependence area the quantity in question falls into determines the correct method of comparison: components require applying the transformation law first, non-tensorial quantities require the extended, non-tensorial transformation rule (or must be combined into a tensorial expression before comparison is meaningful), and scalar invariants can be compared directly with no transformation at all.

Diagnosing Errors by Misplaced Expectations

A frequent source of error is applying the expectations of one basis dependence area to a quantity that actually belongs to another — for instance, expecting a connection coefficient to transform as a tensor, or expecting an individual component (rather than a fully contracted scalar) to be the same number in every basis; correctly locating a quantity within this classification before reasoning about its basis dependence is what prevents such misapplied expectations.