Verification of concurrent lock-free algorithms

Concurrent algorithms are intrinsically difficult to get correct. Formal verification techniques present a way of establishing trust in the correctness by means of a mathematical proof. Currently, two approaches are being investigated in this area: model checking, a fully automatic proof of correctness which is however limited in the application domain, and interactive proving, which is not fully automatic but can treat a larger class of applications.

In the talk recent results on the verification of lock-free algorithms using the second approach are presented. Lock-free algorithms implement a concurrent access to standard data structures like stacks, queues or sets without locking the data structure. This lock-freedom allows for a high degree of concurrency, however, at the price of an increased complexity in verification.