An index operation that sums over a pair consisting of one covariant and one contravariant index of a tensor, producing a tensor of lower rank (possibly a scalar).

Principle

Use the natural pairing between a vector space and its dual to eliminate matched index slots, reducing tensor order while preserving coordinate-covariance.

Demonstration

Given T^i_{jk} and vector v^j, contraction on j yields S^i_k = T^i_{jk} v^j. Contracting g_{ij} with g^{jk} yields δ_i^k (the Kronecker delta).

Misapplication

Summing indices of the same variance (two contravariant indices) without raising or lowering via a metric, producing an object that does not transform tensorially.

Consequence

Generates invariant scalars (traces) and lower-rank tensors used for divergences, traces, and map composition; encodes coordinate-independent evaluations.

Reversal

Tensor product: combines tensors by appending index slots without summation, increasing rank instead of reducing it.

Boundary

Requires attention to index variance and to available metric structures for raising/lowering; in coordinate-free terms, contraction corresponds to evaluation/composition of multilinear maps; not defined when dual pairing is absent or misapplied.

Semantic Tension

Versus index-free composition: contraction can be seen as composition/evaluation of linear maps, but novices may conflate it with simple elementwise multiplication or with inner products that need a metric.

Synthesis

Contraction is the canonical summation of paired dual indices that reduces tensor order by using the vector–covector pairing, producing coordinate-invariant traces and composed multilinear maps.