252-0268-00L  Concepts of Concurrent Computation

Semester Spring Semester 2013
Lecturers B. Meyer, S. Nanz
Periodicity yearly course
Language of instruction English

Abstract Concurrent programming is one of the major challenges in software development. The "Concepts of Concurrent Computation" course explores important models of concurrency, with a special emphasis on concurrent object-oriented programming and process calculi.
Objective After completing this course, students will understand the principles and techniques of concurrent programming, supporting theories allowing formal reasoning about concurrent systems, and advances in concurrent object-oriented programming.
Content Topics include:

- Concurrent and parallel programming
- Multitasking and multiprocessing
- Shared-memory and distributed-memory multiprocessing
- Notion of process and thread
- Performance of concurrent systems

Approaches to concurrent programming
- Issues: data races, deadlock, starvation
- Synchronization algorithms
- Semaphores
- Monitors
- Java and .NET multithreading

Concurrent object-oriented programming: the SCOOP model
- Processors; handling an object
- Synchronous and asynchronous feature calls
- Design by Contract in a concurrent context
- Separate objects and entities
- Accessing separate objects; validity rules
- Synchronization: waiting, reserving, preconditions as wait conditions, Wait by Necessity
- Examples and applications

Programming approaches to concurrency
- Message-passing vs. shared-memory communication
- Language examples: Ada, Polyphonic C#, Erlang (Actors), X10, Linda, Cilk and others.
- Lock-free programming
- Software Transactional Memory

Reasoning about concurrent programs
- Properties of concurrent programs
- Temporal logic
- Process calculi: CSP, CCS
- Proofs of concurrent programs
Literature - Bertrand Meyer and Sebastian Nanz: Course textbook (draft)
- Mordechai Ben-Ari: Principles of Concurrent and Distributed Programming. Prentice Hall, 2006
- Maurice Herlihy and Nir Shavit: The Art of Multiprocessor Programming. Morgan Kaufmann, 2008
- Gregory R. Andrews: Foundations of Multithreaded, Parallel, and Distributed Programming. Addison Wesley, 1999
Prerequisites / Notice The course's lectures are of two different kinds: the Tuesday session (two hours) is a traditional lecture; the Wednesday lecture (one hour) is devoted to seminar talks. The seminar talks will be given about half of the time by well-known international experts in concurrency; the rest of the time they will be given by the student participants, based on research papers related to the topics of the course. The research papers to be presented will be assigned at the start of the course. Depending on the number of participants, some of the exercise sessions may also be used for seminar presentations.