Visualization Award

 

All contributors of PASC14 (talks, posters) have been invited to participate in an award for the best visualization.

An interdisciplinary committee composed of:

  • Dr. Andreas Hirstein, NZZ am Sonntag • Ressortleiter Wissen
  • Prof. Dr. Peter Arbenz, Professor for Computational Science, Department of Computer Science, ETH Zurich
  • John Biddiscobe, Computational Scientist, CSCS 

nominated the following three best visualizations (in ranking order):

  1. Visualization of heated coaxial jet flow and its noise radiation, Michael Gloor (ETH Zurich)
  2. High resolution regional weather model, Oliver Fuhrer (MeteoSwiss)
  3. Simulation of earthquake dynamics in the Landers fault system, Alice Gabriel (University of Munchen)


The three scientists have been rewarded with Amazon vouchers (CHF 50.00).

The gallery with the visualizations that participated to the contest is displayed on the right side.

Gallery

 

Visualization of heated coaxial jet flow and its noise radiation, Michael Gloor (ETH Zurich)

Goal of the research

The research helps to better understand fundamental mechanisms that are important for the noise generation in unsteady coaxial jet flows. The gained knowledge shall help to design quieter jet engines, which will reduce the negative impact of aircraft noise on communities close to airports.

Description

The movie visualizes a coaxial jet flow as it can be found in the exhaust of modern jet engines that are used for aircraft propulsion. The visualization highlights the flow instability process and the transition from laminar to turbulent flow. Intensive mixing between the heated primary jet and the cold bypass flow lead to strong noise radiation.

 

High resolution regional weather model, Oliver Fuhrer (MeteoSwiss)

Goal of the research

Develop a very high-resolution forecasting model capable of running on high-performance architectures with hybrid node architectures.

Description

The movie shows a visualization of the high resolution regional weather model (COSMO-1) - currently being developed by MeteoSwiss - with simulated cloud coverage over a mosaic of satellite images. This model will likely be implemented on a high-performance computing architecture using a hybrid node design. The visualization begins 19 June 2013 and runs for several days (as indicated). The first day of the simulation was an extremely warm day over Switzerland with hardly any thunderstorms and a strong southerly flow (Foehn). On the second day, the southerly flow persists and a perturbation passes over Switzerland with associated precipitation. On the third day, the flow gains a westerly component with a low pressure system over the Northern Sea. As a high pressure ridge passes over Switzerland, clouds clear away temporarily over midday. Spatial resolution is 1.1km while temporal resolution is 2 minutes. The model is developed by the the Consortium for Small-scale Modeling (COSMO) formed by several national meteorological services (cosmo-model.org). Foreground (model data) shows cloud coverage (3 cloud layers in different saturations of white) as well as precipitation rate (blue layer). Background shows Blue Marble next generation mosaic of satellite images with 500m spatial resolution (August 2004).

 

Simulation of earthquake dynamics in the Landers fault system, Alice Gabriel (University of Munchen)

Goal of the research

The observed high-detail rupture evolution generates synthetic ground motions with frequencies up to 10 Hz relevant for seismic hazard assessment and civil engineering purposes. Furthermore, fault interaction, such as branching and rupture jumps, can be investigated with highest detail helping to improve the understanding of earthquakes.

Description

High-detail earthquake evolution in a pioneering petascale simulation of the 1992 Landers earthquake. The multiphysics earthquake simulation on unstructured tetrahedral meshes couples rupture dynamics on complex fault geometries and high-order accurate seismic wave propagation. We visualize particle motion on the complete fault system throughout the dynamic earthquake rupture process. The occurrence of multiple rupture fronts, enhanced by the full model complexity, can be observed clearly.

 

Microscopic dynamics of perovskites, Giovanni Pizzi (THEOS, EPF Lausanne)

Goal of the research

The ultimate goal of our research is to understand and carefully describe the microscopic dynamics of perovskites (in this example, barium titanate) at different temperatures, so as to clarify the driving mechanisms of piezoelectricity and ferroelectricity and finally solve the long-standing debate in the literature about the nature of the paraelectric-ferroelectric phase transition in these materials. Such a thorough understanding will also allow us to develop novel, more efficient materials.

Description

Molecular dynamics simulation of a supercell of the BaTiO3 perovskite system (barium titanate) at a temperature of 400 K. In the second part of the video, the dynamics of one single subcell is emphasized, to focus on the microscopic dynamics of the B cations (titanium atoms, blue atoms) with respect to the oxygen octahedral cage (red atoms). The color plot that appears is a histogram of the titanium positions with respect to the instantaneous center of the cage, integrated over the simulation time. The peaks in the histogram prove that, on average, the B cations prefer to occupy off-centered positions rather than the center of the octahedral cage.

 

Secondary Instability and Transition in Swept Hiemenz Flow, Michael John (ETH Zurich)

Goal of the research

This instability occurs on airplane wings. We want to inhibit this kind of transition to turbulent flow and thereby reduce airplane fuel consumption.

Description

Depicted is a 3D boundary-layer flow. The streamlines (black) and vortices (isosurfaces) visualize transition to turbulent flow. The flow is disturbed by two (primary) vortices. They grow, saturate and become unstable to a tiny time-dependent, oscillatory (secondary) motion. As a result, the flow is turbulent, which increases the wall friction.

 

in-silico cell proliferation using SEM++, Gerardo Tauriello (ETH Zurich)

Goal of the research

The goal is to study the emergence of phenomena initiated by single cells. For this purpose, we have developed SEM++: a discrete particle method to model biological processes from the sub-cellular to the inter-cellular level (DOI 10.1007/s40571-014-0017-4).

Description

3D simulation of a growing and dividing cell. Cells are modeled as a collection of sub cellular elements (SCE) which can move freely in space. The nucleus is a large SCE. SCE are duplicated to grow the cell. Before cell division, a second nucleus is introduced and cell division proceeds by assigning each particle to its closest nucleus. Growth progression: bottom left to top right. Blue: cell membrane, red: nuclei

 

134,000 molecules, Raghunathan Ramakrishnan (University of Basel)

Goal of the research

This research aims to forecast molecular properties of hundreds of thousands of molecules using rapid and accurate statistical models. The 134,000 molecules depicted in this figure provide excellent datasets to develop and fine tune machine learning methodologies. Further more, this study sets up a very high standard in the benchmarking of large scale first principles calculations.

Description

This heat map shows the distribution of 133,885 small organic molecules as a function of their shape and stability. The X and Y axes are normalized moments of inertia (I1/I3 and I2/I3, both dimensionless) and the color maps to the heats of atomization (H in kcal/mol) from density functional theory calculations. Also shown are the representative 1D (rod shaped), 2D (disc shaped), and 3D (near spherical) molecules at the corners along with the most stable molecule. Each single dot corresponds to one or many molecules and data required to generate this plot utilized over million CPU hours.

 

Brain cancer, Jana Lipkova (ETH Zurich)

Goal of the research

The goal of our research is to develop a personalized approach for inferring the evolution and spatial distribution of brain tumors. We use MRI scans to reconstruct the patient anatomy and to estimate patient specific tumor properties in order to assist clinical interventions.

Description

The figure shows a snapshot from a simulation of tumor growth and invasion in human brain. The data for the brain anatomy is obtained from MRI database BrainWeb.