In topology an Akbulut cork is a structure is frequently used to show that in four dimensions, the smooth h-cobordism theorem fails. It was named after Selman Akbulut.
A compact contractible Stein 4-manifold C with involution F on its boundary is called an Akbulut cork, if F extends to a self-homeomorphism but cannot extend to a self-diffeomorphism inside. A cork (C,F) is called a cork of a smooth 4-manifold X, if removing C from X and re-gluing it via F changes the smooth structure of X (this operation is called "cork twisting"). Any exotic copy X' of a closed simply connected 4-manifold X differs from X by a single cork twist [1] [2] [3] [4] [5].
The basic idea of the Akbulut cork is that when attempting to use the h-corbodism theorem in four dimensions, the cork is the sub-cobordism that contains all the exotic properties of the spaces connected with the cobordism, and when removed the two spaces become trivially h-cobordant and smooth. This shows that in four dimensions, although the theorem does not tell us that two manifolds are diffeomorphic (only homeomorphic), they are "not far" from being diffeomorphic.[6]
To illustrate this (without proof), consider a smooth h-cobordism W5 between two 4-manifolds M and N. Then within W there is a sub-cobordism K5 between A4 ⊂ M and B4 ⊂ N and there is a diffeomorphism
which is the content of the h-cobordism theorem for n ≥ 5 (here int X refers to the interior of a manifold X). In addition, A and B are diffeomorphic with a diffeomorphism that is an involution on the boundary ∂A = ∂B.[7] Therefore it can be seen that the h-corbordism K connects A with its "inverted" image B. This submanifold A is the Akbulut cork.