Joseph Tassarotti

Verifying Concurrent Randomized Algorithms Degree Type: Ph.D. in Computer Science
Advisor(s): Robert Harper
Graduated: December 2018

Abstract:

Concurrency and randomization are diffiult to use correctly when programming. Because programs that use them no longer behave deterministically, programmers must take into account the set of all possible interactions and random choices that may occur. This dissertation describes a logic for reasoning about programs using both of these effects. The logic extends a recent concurrent separation logic with ideas from denotational semantics for probabilistic and nondeterministic choice, along with principles for probabilistic relational reasoning originally developed for sequential programs. The resulting logic is used to verify probabilistic behaviors of a randomized concurrent counter algorithm and a twolevel concurrent skip list. The soundness of the logic, as well as the proofs of these examples, have been mechanized in Coq.

Thesis Committee:
Robert Harper (Chair)
Jan Hoffmann
Jeremy Avigad
Derek Dreyer (MPI-SWS)

Srinivasan Seshan, Head, Computer Science Department
Tom M. Mitchell, Interim Dean, School of Computer Science

Keywords:
Concurrency, program logics, separation logic, randomized algorithms, verification