Computing inverses

We compare methods of computing inverses of matrices over division rings. The most direct way is via Cohn's theory of full matrices, which was improved by Malcolmson, Schofield, and Westreich. But it is simpler to work with finite dimensional representations and generic matrices. In this talk, mostly expository, we describe the relevant algebraic techniques, including a description of full matrices, Sylvester rank functions, coproducts, and generic division algebras and its underlying theory of polynomial identities.