COLLOQUIUM DI DIPARTIMENTO
High Performance Computing
Problem solving at realistic complexities: why high performance computing
matters
Luca Heltai, Director of the Master in High Performance Computing, SISSA
MathLab, Trieste.
04/05/2016 ore 15:30
dove: Largo San Leonardo Murialdo, 1 - aula F
The practical numerical solution of real-world problems has many levels of
complexities. A single method, or a single algorithm, is often not
sufficient to address realistic problems from applied sciences and
engineering. For example, it is common for a finite element code to use
fully adaptive, unstructured meshes; nonlinear, globalized solvers;
algebraic multigrid; block preconditioners; or non-linear material
properties.
Real-life programs usually need to run on thousands of processors, often
with different architectures, and they generate (or manipulate) gigabytes
(often petabytes) of data at each run. Codes at this level of complexity
can no longer be written from scratch by a single person, and not even by a
consistent group of well prepared PhD students. However, over the past
decade, many high quality libraries and tools have been developed that make
writing advanced computational software simpler.
In this talk I will briefly introduce the deal.II finite element library (
http://www.dealii.org), of which I am one of the developers, and I will
show some examples of its application in real-life problems. I will comment
on the lessons learned from developing this massively parallel library for
the solution of complex problems, and I will present the Master in High
Performance Computing (MHPC) that has been developed jointly by SISSA and
ICTP to respond to the strong need of professionals, capable of solving
efficiently large computational problems both in Academia and Industry.
