XSEDE13 is hosted by
In cooperation with
Non-Profit Silver Sponsors
Non-Profit Bronze Sponsors
XSEDE is supported by the
National Science Foundation
Movie-cube: Spatial representation of time-based media
David Bock (NCSA), Virginia Kuhn (USC)
The XSEDE-funded Large Scale Video Analytics (LSVA) project addresses obstacles in image-retrieval and research using extreme-scale archives of video data. Deployed on the SDSC Gordon Compute Cluster, the LSVA includes the creation of visualization tools that enhance research in several ways: novel visualizations employ spatial and temporal simultaneity, revealing unique aspects of a single film sequence; comparative visualizations represent relationships among multiple films within an archive; and, finally, the integration of visualization imagery becomes an input tag and a front end process that feeds the Medici content management system and enhances word-based labels, helping to close the semantic gap that occurs when words are applied to images. The technique developed to generate spatial representations of video is demonstrated in this animation. Custom software converts a digital video sequence into a three-dimensional dataset, termed a movie-cube, by extracting and ordering each frame of the sequence along the Z axis. Once in this form, a variety of techniques commonly used in scientific visualization can be used to show multiple perspectives, allowing analysis of image-based, time-based media that is unattainable in traditional methodologies. A custom visualization system is used to render and examine the dataset as shown in the animation by first animating a slice plane along the Z axis representing time within the video sequence. As expected, such movement within the movie-cube reveals the original movie sequence. Experimenting with different orientations of the slice plane, however, reveals unique and interesting patterns showing various aspects of the time-based data within a single spatial representation. Note, for example, how such visualizations provide a clear representation of cinematic elements such as camera shots, angles and movements.
Visualization of vortical structures from DNS of spatially developing turbulent boundary layers
David Bock (NCSA), Antonino Ferrante (University of Washington)
The XSEDE-funded DNS of Spatially Developing Turbulent Boundary Layers project simulates the three-dimensional turbulent flow over a flat plate. Using XSEDE resources Kraken and Ranger, the project goals are to simulate such fluid flow using the largest Reynolds number ever reported, perform the first direct numerical simulation of fully-resolved droplet-laden isotropic turbulence, and share the generated DNS database with the scientific community. XSEDE ECSS support for this project includes Petascale scaling and deployment, development of a high-level HDF5 parallel IO layer (H5DNS), and custom visualization of simulation results. The time-dependent movement of the vortical structures identified with the λ2-method is represented in this animation extracted from a DNS using 151 billion grid points. A custom visualization system is used to render ray-traced volume renderings of the λ2 field along the turbulent boundary layer from a variety of different viewpoints. Visual cues are enhanced with lighting, shadows, and representation of the rectilinear simulation grid. Evident in the visualizations are the vortical structures as quasi-streamwise tubes and arches.
Probing the Effect of Conformational Constraints on Binding
Anne Bowen (TACC), Yue Shi (TACC)
Increasing the strength of binding between a molecule and a receptor is an important technique in the design of effective drugs. One experimental technique to increase the strength of binding (called "binding affinity") is to synthesize molecules that are already in the shape that it will take when bound to a receptor. This technique works because it decreases the binding entropy which increases the overall binding affinity. A recent experimental study of a series of receptor-molecule complexes (the Grb2 SH2 domain with peptide analogues) aimed to increase the binding affinity by introducing a bond constraint. However, the constrained molecules had less favorable binding entropies than their flexible counterparts. Yue Shi of the Ren lab at UT Austin aimed to probe the origin of this entropy paradox with molecular dynamics simulations which were run on Lonestar and Ranger at TACC. Their group used approximately 2 million CPU hours on Ranger and almost 1 million on Lonestar this past year. Their research addresses biological and medical challenges from single molecules to the genome with high performance computing and theory. In collaboration with other experimental groups, they utilize computer modeling and simulations to understand these complex biomolecular systems and to discover molecules for treating disease and improving human health. Effectively communicating the results of their computational studies to experimentalists is essential to the success of their collaborative efforts. Anne Bowen of the Texas Advanced Computing Center collaborated with Yue Shi to prepare animations and graphics to better explain the origins of the "entropy paradox" to experimentalists and the general public.
Visualizing Simulated Volcanic Eruptions
Amit Chourasia (SDSC), Darcy Ogden (UCSD), Kenneth Wohletz (LANL)
Eruptive conduits feeding volcanic jets and plumes are connected to the atmosphere through volcanic vents that, depending on their size and 3D shape, can alter the dynamics and structure of these eruptions. The host rock comprising the vent, in turn, can collapse, fracture, and erode in response to the eruptive flow field. This project uses cutting edge visualization to illustrate and analyze results from fully coupled numerical simulations of high speed, multiphase volcanic mixtures erupting through erodible, visco-plastic host rocks. The visualizations explores the influence of different host rock rheologies and eruptive conditions on the development of simulated volcanic jets.
Inside Views of a Rapidly Spinning Star
Greg Foss (TACC), Greg Abram (TACC), Karla Vega (TACC), Ben Brown (UW-Madison), Mark Miesh (NCAR)
"Inside Views of a Rapidly Spinning Star" shows a sampler of visualized variables from a star simulation- similar to our Sun in mass and composition but spinning five times faster. The movie compares the variables radial velocity, enstrophy, and velocity magnitude. The animations included in this video were created with VisIt (LLNL, DOE Advanced Simulation and Computing Initiative) and Longhorn, the visualization cluster at TACC. These particular animations were selected from dozens made over a period of two months to visually explore the dataset Ben Brown (U. of Wisconsin at Madison) generated using ASH (Anelastic Spherical Harmonic, originally developed at the University of Colorado) on TACC's HPC system Ranger (72 hours on 256 processors) in June 2012. This is one of several simulations Brown and Mark Miesch (University Corporation for Atmospheric Research) have produced as part of their ongoing investigation in stellar phenomena. This video provides an illustration of the simulation results.
Coherent structures, intermittent turbulence, and dissipation in high temperature plasmas
Homa Karimabadi (UCSD), Vadim Roytershteyn (SciberQuest, Inc.), Minping Wan (University of Delaware), William Matthaeus (University of Delaware), William Daughton (LANL), Michael Shay (University of Delaware), Burlen Loring (LBNL), Joe Borovsky (University of Michigan), Sandra Chapman (University of Warwick), Takuma Nakumura (LANL), Penny Wu (University of Delaware)
An unsolved problem in plasma turbulence is how energy is dissipated at small scales. Particle collisions are too infrequent in hot plasmas to provide the necessary dissipation. Simulations either treat the ﬂuid scales and impose an ad hoc form of dissipation (e.g., resistivity) or consider dissipation arising from resonant damping of small amplitude disturbances where damping rates are found to be comparable to that predicted from linear theory. Here, we report kinetic simulations that span the macroscopic ﬂuid scales down to the motion of electrons. We ﬁnd that turbulent cascade leads to generation of coherent structures in the form of current sheets that steepen to electron scales, triggering strong localized heating of the plasma. The dominant heating mechanism is due to parallel electric ﬁelds associated with the current sheets, leading to anisotropic electron and ion distributions which can be measured with NASA's upcoming Magnetospheric Multiscale mission. The motion of coherent structures also generates waves that are emitted into the ambient plasma in form of highly oblique compressional and shear Alfven modes. In 3D, modes propagating at other angles can also be generated. This indicates that intermittent plasma turbulence will in general consist of both coherent structures and waves. However, the current sheet heating is found to be locally several orders of magnitude more efﬁcient than wave damping and is sufﬁcient to explain the observed heating rates in the solar wind. In this work the visualization and analysis; visualization software developed by RDAV and LBNL; and the use of XSEDE visualization resource Nautilus lead to breakthroughs in the understanding of fundamental physical processes responsible for turbulent heating of the solar wind. We present the following figures and movies that were a key to the success of this collaboration.
Ultra-high Resolution Simulation of a Downburst-producing Thunderstorm
Leigh Orf (Central Michigian University), Robert Sisneros (NCSA), George Bryan (NCAR)
Downbursts are intense downdrafts that are spawned by thunderstorms. In some cases, the downburst-producing thunderstorms are weak and short-lived, but occur in environments which are ideal for the formation of strong downdrafts. Downbursts can cause tornado-strength winds at the earth's surface, resulting in significant damage to property. The simulation we present occurs in such an environment, one that is typical for the High Plains of the United States. A thunderstorm grows and produces snow and graupel within the cloud which, as it falls downward into warm air, melts to form rain which then evaporates. The melting and evaporation cool the air, which is free to accelerate downward in this type of environment. The main research objective of this work is to understand the specific wind loads that downbursts present to structures such as power transmission lines, which are vulnerable to downburst winds. Up until recently, downbursts were typically simulated using simple models that force the downdraft using either a momentum or density source without including the thunderstorm cloud itself. Only recently has it been possible to simulate the lifecycle of an entire downburst-producing thunderstorm at resolutions sufficient to properly capture the flow near the ground where the strong winds do their damage. In addition to providing data valuable to wind engineers interested in designing structures that can withstand downburst winds, thunderstorm simulations at resolutions only possible using supercomputing facilities can also be used to aid meteorologists in better understanding both the physical processes causing the downbursts as well as the surface damage patterns caused by downbursts. The imagery of this simulation highlights the origin and motion of the fastest-moving air. We chose to focus on potential temperature perturbation as a volume-rendered field in order to trace the motion of the air, exploiting the fact that in this type of thunderstorm, which produces a low-precipitation "dry downburst," the negative buoyancy that forces the downdraft is primarily thermodynamic in origin, resulting from evaporation, sublimation, and melting of hydrometeors. In so-called "wet" downbursts, which typically occur in more humid environments, negative buoyancy results primarily by the drag induced by heavy precipitation. Negative potential temperature perturbation is proportional to positive density perturbation, and is the source of the term term "density current" which describes phenomena such as downbursts. This simulation was run on the kraken XSEDE supercomputer at NICS. It utilized 18,000 cores and produced 25 TB of data in four wallclock hours. The simulation was rendered on the Blue Waters supercomputer using VisIt.
AllTaxa: A Web Interface for Exploring Species Distribution Models
Scott Simmerman (UT-Knoxville), Blaise Decotes (UT-Knoxville), Kimberly Shook (UT-Knoxville), Jian Huang (UT-Knoxville)
The All Taxa Biodiversity Inventory (ATBI) is an ongoing effort to document all living species in the Great Smoky Mountains National Park. Biologists and ecologists with the National Institute for Mathematical and Biological Synthesis (NIMBioS) planned to use MaxEnt, a Java-based presence-only modeling software, to model species distributions across the entire Park based on the data from the ATBI. We at the Center for Remote Data Analysis and Visualization (RDAV) collaborated with NIMBioS to facilitate this modeling effort using high performance computing as well as to provide further analysis and visualization capabilities. We developed an entire workflow involving parallel runs of MaxEnt on Nautilus, parallel code for pair-wise comparisons of models, custom visualization of species distribution models (SDMs) using VisIt, and statistical analysis of results using multi-dimensional scaling (MDS) with R. To disseminate our work to the biologists as well as to National Park personnel and management and the general public, we designed and deployed a web interface to allow interactive exploration of the results. Being able to see and compare distribution patterns for species across the entire Park aids the Park management in their stewardship of national resources. They can explore interrelationships of species and be more informed about patterns regarding invasive species, environmental changes, and threatened species.
These SDMs also inform the future efforts of collecting species in the Park, helping to fulfill the mission of the ATBI.
The First Star: Birth through Death
Matthew Turk (Columbia University), John Wise (Georgia Tech), Sam Skillman (University of Colorado), Mark Subbarao (Adler Planetarium)
This visualization was prepared from simulations of the birth, life, and death of the very first star in the Universe. These stars were likely very massive, between 30 and 300 times the mass of our sun, and lived brief lives before enriching their surroundings with heavy elements. All simulations were conducted using the code Enzo ( http://enzo-project.org/ ) and all visualization conducted with the analysis toolkit yt ( http://yt-project.org/ ). Both are freely available, open source, community-driven and widely deployed and developed on NSF XSEDE. These simulations and visualizations utilized the NSF XSEDE systems Kraken and Trestles.