Advanced Parallel Computing and Solvers in Engineering

Vertiefungsfach Master (3 SWS, 5 ECTS, SS 2022, Modul-Nr. MW1746, Room BC2 3.2.29 (Hochbrück), Prof. M.W. Gee)

Corona Notice: This lecture will be held in presence with additional online material.
Please sign up as usual for this module to gain access to its moodle page.

Within simulation of continuum-mechanical problems rising system sizes also challenge the capacities of computing facilities. One way to address this challenge is the utilization of parallel computing systems. Therein, work load is distributed among many processors or cores and processed simultaneously. Unfortunately, it is not possible to straightforward utilize common serial software on parallel computing platforms or multicore systems. Rather, methods, algorithms and software have to be tailored to meet the specific needs posed by parallelism and parallel hardware architectures. This course provides an overview over methods and techniques that are common in computational structural and fluid dynamics.

Contents

  • Introduction to the design of parallel algorithms
  • Selected parallel algorithms
  • Domain decomposition techniques in continuum mechanics
  • Parallel iterative methods for the solution of large linear systems of equations
  • Parallel preconditioning
  • An introduction to multigrid techniques
  • Algebraic multigrid for elliptic problems such as solids mechanics or thermal convection

Prerequisites

  • Finite Elements or similar course recommended, but not mandatory
  • Knowledge of (any) programming language is helpful but not mandatory (Exercise will be in C/C++)

Lecture & Exercise hours

  • Lecture will be Thursdays, 14:15-15:45 in BC 3.2.29
  • 1 Block Exercise, Thursday 14:15-18:00, dates will be announced
  • Lecture is either in English or German, depending on the audience. All course material is in English.
  • Recorded lecture and exercise material will be made available via links on moodle.

Preliminary Time Table

  • 28.4.22, 14:15-15:45, Introduction & lecture
  • to be announced

Exercise

  • Exercise materials will be provided through moodle.
  • Exercises will be hands-on, you will need your own laptop (any operating system, no requirements)

Exam

  • will be announced

Slides

  • Course material will be supplied in moodle. Sign up for the course in TUMonline to gain access.

 

Office hours