Date and Time:

Monday 9 January 2023, 11:00 - 17:00 GMT

Tuesday 10 January 2023, 09:00 - 17:00 GMT

Wednesday 11 January 2023, 09:00 - 13:00 GMT

Location: Reed Hall, University of Exeter, Streatham Drive, Exeter EX4 4QR

Overview

In this workshop we will focus on time-parallel methods. Following exciting developments in both mathematical analysis and practical experience, time-parallel methods are undergoing a revival as a potentially powerful route to exploiting future massively parallel exascale supercomputers. Time-parallel methods address the question of what to do when one has reached the limits of strong scaling (decreasing wallclock time by increasing the number of processors working in parallel) through domain decomposition parallelisation in space. A key lesson from the recent literature is that the success of parallel-in-time algorithms critically depends on them being carefully adapted to the equation being solved. Much like regular timestepping methods, there are many parallel-in-time algorithms, and the right algorithm needs to be designed and selected according to the mathematical properties and applications requirements of the underlying system.

This workshop will combine lectures with practical demonstrations to introduce timestepping challenges and how to overcome them using time-parallel methods, such as parareal, deferred corrections and paradiag. The practical demonstrations will be based on jupyter notebooks and some experience of using python is desirable.

What to bring?

Since the course will have significant practical elements, please bring your laptop. To run code on your laptop you will need to have Python (version >3) installed, along with numpy, scipy and jupyter notebook. The code will be provided via GitHub. We can support you with all of these things if necessary.

Schedule

Monday

11.00 Introduction to the workshop: aims and practicalities, Jemma Shipton.

11.30 Exascale Computing Challenges, David Acreman

12.30 Lunch

13.30 Why is timestepping so hard? Jemma Shipton (Lecture and practical)

15.00 Coffee

15.30 Introduction to parallel timestepping approaches, Jemma Shipton (Lecture)

17.00 Finish

Tuesday

The sessions on this day will consist of a short introduction to the method followed by some practical exercises.

09.00 Deferred correction methods

10.30 Coffee

11.00 Paradiag

12.30 Lunch

13.30 Introduction to parallel profiling and benchmarking, David Acreman

15.00 Break

15.30 Troubleshooting and discussion session

17.00 Finish

Wednesday

09.00 Parareal

10.30 Coffee

11.00 How not to fool with (or be fooled by) time-parallel scaling results, Jemma Shipton

11.45 Summary and discussion

12.30 Close

Registrations are closed

Exascale Computing Challenges: Parallel-in-Time Algorithms