Skip to main content
Wiki|Contact
Log In
  • About
    • Objectives
    • Workplan
    • Partners & Coordinator
    • Collaborators
  • Publications
  • Results
    • Pilot Demonstrators
    • Flagship codes
    • Repository
    • Deliverables
    • Success Stories
    • Live Exercises
    • KPIs
  • Events
    • Workshops
    • Meetings
    • Conferences
    • Training
  • Media
    • News
    • In press
    • Videos
    • Dissemination
    • Branding material
  • ShEESE
    • Equality Committee members
    • Equality and Diversity Policies
    • ShEESE Activities
  • Search

ExaHyPE

Intranet
Contact
LinkedIn
Twitter
Code name ExaHyPE
Developer(s)

Leonhard Rannabauer (TUM), Philipp Samfaß (TUM), Jean-Matthieu Gallard (TUM), Tobias Weinzierl (Durham University) and Anne Reinarz (Durham University)

Contact person: Michael Bader

Link

https://www.exahype.org/

Short description

The ExaHyPE engine numerically solves systems of hyperbolic PDEs, as stemming from conservation laws. A concrete model for seismic wave propagation problems is being developed within the ExaHyPE project. The model is based on high-order Discontinuous Galerkin (DG) discretization, local time-stepping and works on tree-structured Cartesian meshes. The activities in this CoE will focus on setting up concrete services based on the ExaHyPE engine and seismic models.

Original code level

2

Current code level

2

Pilot(s) involved

PD1, PD4, PD5

Co-design

Co-design activities: 

Optimisation of backends for small tensor and matrix operations on various CPU architectures (Intel, AMD, ARM)

Main results and References

Modeling improvements:

  • Automatic curvilinear mesh generation for domains with complicated topography

  • Implementation of UQ-based scenarios for seismic hazard assessment

Performance results

Algorithmic and HPC improvements:

  • Improved vectorization due to algorithmic modification of the ADER-DG scheme and change of memory layout (“Array of Struct of Array”)

  • Reactive load balancing via task offloading and experiments with team-based resilience

ExaHyPE is an open source engine to solve hyperbolic partial differential equations in first order formulation on dynamically adaptive spacetree grids.
The ChEESE project has received funding from the European Union’s Horizon 2020
research and innovation programme under the grant agreement Nº 823844. All rights reserved. Legal Notice.
© CHEESE-COE.EU COPYRIGHT 2018 - 2019
  • About
    • Objectives
    • Workplan
    • Partners & Coordinator
    • Collaborators
  • Publications
  • Results
    • Pilot Demonstrators
    • Flagship codes
    • Repository
    • Deliverables
    • Success Stories
    • Live Exercises
    • KPIs
  • Events
    • Workshops
    • Meetings
    • Conferences
    • Training
  • Media
    • News
    • In press
    • Videos
    • Dissemination
    • Branding material
  • ShEESE
    • Equality Committee members
    • Equality and Diversity Policies
    • ShEESE Activities
  • Search