# Difference between revisions of "SC²S Colloquium - September 21, 2017"

(Created page with "{| class="wikitable" |- | '''Date:''' || September 21, 2017 |- | '''Room:''' || 02.07.023 |- | '''Time:''' || 4:00 pm, s.t. |- |} == Florian Hecher: Development of a tool for...") |
|||

Line 11: | Line 11: | ||

== Florian Hecher: Development of a tool for postprocessing of simulated earthquake data == | == Florian Hecher: Development of a tool for postprocessing of simulated earthquake data == | ||

As they are really difficult to predict, simulating tsunamis is an important task in estimating their impact. This simulation should be well connected to a previous earthquake simulation. For the earthquake simulation tool SeisSol and the tsunami simulation tool sam(oa)² this connection is currently quite poor, as SeisSols output format doesn't match the input requirements of sam(oa)². This thesis aims to resolve this issue by converting SeisSols triangular mesh to sam(oa)²s rectangular grid. This conversion is done via integral over the original data for the rectangles area. The result is a program with a low runtime that provides a conversion with high accuracy and enables precise simulations of possible tsunamis. | As they are really difficult to predict, simulating tsunamis is an important task in estimating their impact. This simulation should be well connected to a previous earthquake simulation. For the earthquake simulation tool SeisSol and the tsunami simulation tool sam(oa)² this connection is currently quite poor, as SeisSols output format doesn't match the input requirements of sam(oa)². This thesis aims to resolve this issue by converting SeisSols triangular mesh to sam(oa)²s rectangular grid. This conversion is done via integral over the original data for the rectangles area. The result is a program with a low runtime that provides a conversion with high accuracy and enables precise simulations of possible tsunamis. | ||

+ | |||

+ | ==Dmitrij Boschko: Orthogonal Matrix Decomposition for Adaptive Sparse Grid Density Estimation Methods == | ||

+ | A new algorithm for adaptive sparse grid density estimation is introduced and analyzed. In an offline/online splitting context an orthogonal decomposition of the underlying system matrix and Sherman-Morrison rank-one updates are used. This allows for regularization, which can be followed by adaptivity. The current implementation's runtime is analyzed and evaluated with respect to the Cholesky decomposition based algorithm for the same task. The new algorithm reaches faster solving times, while the speed of the other online subroutines stays in the same runtime complexity, though slightly slower. | ||

[[Category:ShowComingUp]] | [[Category:ShowComingUp]] | ||

[[Category:news]] | [[Category:news]] |

## Latest revision as of 15:52, 11 September 2017

Date: |
September 21, 2017 |

Room: |
02.07.023 |

Time: |
4:00 pm, s.t. |

## Florian Hecher: Development of a tool for postprocessing of simulated earthquake data

As they are really difficult to predict, simulating tsunamis is an important task in estimating their impact. This simulation should be well connected to a previous earthquake simulation. For the earthquake simulation tool SeisSol and the tsunami simulation tool sam(oa)² this connection is currently quite poor, as SeisSols output format doesn't match the input requirements of sam(oa)². This thesis aims to resolve this issue by converting SeisSols triangular mesh to sam(oa)²s rectangular grid. This conversion is done via integral over the original data for the rectangles area. The result is a program with a low runtime that provides a conversion with high accuracy and enables precise simulations of possible tsunamis.

## Dmitrij Boschko: Orthogonal Matrix Decomposition for Adaptive Sparse Grid Density Estimation Methods

A new algorithm for adaptive sparse grid density estimation is introduced and analyzed. In an offline/online splitting context an orthogonal decomposition of the underlying system matrix and Sherman-Morrison rank-one updates are used. This allows for regularization, which can be followed by adaptivity. The current implementation's runtime is analyzed and evaluated with respect to the Cholesky decomposition based algorithm for the same task. The new algorithm reaches faster solving times, while the speed of the other online subroutines stays in the same runtime complexity, though slightly slower.