universal property of the tensor product

A universal property describes an object not by how it is built but by what it does best: the tensor product is the most efficient device for converting bilinear maps into linear ones, and nothing about its internal construction matters once you know this job description. Any two objects satisfying the description are canonically isomorphic, so the tensor product is determined up to unique isomorphism.

Statement: there is a bilinear map ι: M × N -> M ⊗ N, ι(m, n) = m ⊗ n, with the property that for every R-module P and every bilinear map B: M × N -> P there exists a UNIQUE R-linear map B': M ⊗ N -> P with B = B' ∘ ι. In words, every bilinear map factors uniquely as a linear map composed with the universal one. This makes ⊗ the carrier of the functor that represents bilinear maps: Bilin(M × N, P) ≅ Hom_R(M ⊗ N, P).

The practical upshot is that you define a map out of M ⊗ N by specifying it on pure tensors via a bilinear rule and checking only bilinearity — the universal property guarantees a well-defined linear extension. This is exactly how the wedge, symmetric, and contraction maps are defined cleanly, and why proofs about tensor products almost never touch the quotient-of-free-module construction.