XSEDE14 is hosted by

XSEDE, Extreme Science and Engineering Discovery Environment, xsede.org.

In cooperation with

SigApp Logo

XSEDE is supported by the 
National Science Foundation

In partnership with

 

Platinum Sponors

Gold Sponsors

Silver Sponsors

Bronze Sponsors

Non-profit Silver Sponsor

Non-profit Bronze Sponsors

 

 

 

Patron Sponsors

 

In-kind Sponsors

Media Sponsor

Visualization Showcase

 

Using Genetic Algorithm to solve Fisk University's Athletic Department Travel Routes

Jayson Ambrose

At Fisk University, all teams under the athletic department can only travel to cities or states within a certain radius to keep the loss of funds to a minimum. But there is not a primary strategy on which route to pursue in relation to the circuit of cities that the teams have to travel. The framework behind the team's travelling system reveals that the teams travel in their own directions to several cities and states to promote the university's athletic department, which it does, but it also causes the department expenses to increase due to the inconsistency of distances and lack of controlled travel range. In this thesis, we re-introduce the concept of genetic algorithm through the Travelling Salesman Problem, using its evolution of genes procedure to generate the best routes for each team to travel the minimum amount of miles. We chose the technique of genetic algorithm because it is efficient when dealing with complex problems. It can reduce the initial distance by approximately 44.5% after processing several times to generate the best distance. We implemented a body of codes that continuously executes and stores the total distances of each route created to display the results in descending order where the smallest number of distance is the best route. We expect this new approach to be beneficial to the universities athletic department as well as others who may encounter this problem because it finds the shortest route to travel which reduces the rate of money loss drastically.

 

Macro-Scale Modeling of Deflagration to Detonation Transitions in Large Arrays of Explosives

Jacqueline Beckvermit, Andrew Bezdjian, Todd Harman, Qingyu Meng, Alan Humphrey, John Schmidt, Martin Berzins and Charles Wight

The physical mechanism and constraints required for Deflagration to Detonation Transitions (DDT) in a large number of explosive devices is investigated. Motivation for this work comes from a 2005 transportation accident in Spanish Fork Canyon, where a truck carrying 36,000 pounds of seismic boosters overturned, caught fire and detonated. The explosion created a crater 30 feet deep by 70 feet wide in the middle of the highway.
In the Uintah computational framework we have developed a robust reaction model capable for simulating the different modes of combustion at large scales, by leveraging the massively parallel capabilities of the Uintah Computational Framework and large computational resources including XSEDE's Stampede machine. Uintah has shown good scalability characteristics up to 512k cores. Current efforts will give insight into the physical mechanism of DDT for large-scale explosions and help determine safe packaging configurations to reduce the probability of DDT in transportation accidents. Visualization of these simulations utilizing VisIt's parallel capabilities—while running with over 300 processors on Maverick and multiple other computational platforms—has presented new possible mechanisms behind the detonation. With these visualizing capabilities, and outputting data frequently, not only do we know when and where a detonation has occurred, we can see the mechanics behind the mechanism transitioning to detonation.

 

Visualization of large eddy simulations of extended wind farms

David Bock

Link: http://lantern.ncsa.uiuc.edu/~dbock/Vis/XSEDE/Bock_WindFarm.mov

Description
The XSEDE-funded Large Eddy Simulations of Extended Wind Farm project studies the interaction between large wind farms with multiple wind-turbines and the flow in the atmospheric boundary layer flow. Because the wind-turbines are placed relatively close together their operation and power production is influenced by the wake created by upstream wind-turbines. The simulations, in agreement with field experiment data, show that power production of wind turbines placed downstream of other turbines can decrease up to 50% with respect to the power produced by free standing wind-turbines. Since this decrease is quite substantial, one of the science goals is to use the simulations in order to understand these effects better. The simulations can also be used to test flow structures and how they depend on the design parameters of the wind farm. The XSEDE project goals are to provide assistance implementing two dimensional domain decomposition and parallel FFT as well as 3D visualization representations to assist in simulations and analysis of large wind farms.

A custom software system is used to render a variety of different visualization representations of stream-wise and wind velocity across time within the wind farm. Scenes include color-mapped slice planes and ray-traced volume renderings visualizing stream-wise velocity and particle systems visualizing wind direction and velocity. Visual and perceptual cues are enhanced with lighting, shadows, camera movement, ambient occlusion and representation of the underlying rectilinear simulation grid. Visual cues also include wind turbine models and turbine rotation matching simulation parameters.

Contributors
Visualization (and presenter)
David Bock, National Center for Supercomputing Applications

Principal Investigators
PI: Charles Meneveau, Johns Hopkins University

Project participants
Richard Stevens, Jason Graham, Claire Verhulst, Dennice Gayme, Charles Meneveau (PI), Johns Hopkins University
Darren Adams, David Bock, National Center for Supercomputing Applications

Resources used:
Trestles

 

Visualization of 3D Home Ranges for a California Condor Breeding Pair

Amit Chourasia, Jeff Tracey, James Sheppard, Glenn Lockwood, Mahidhar Tatineni and Robert Sinkovits

The critically endangered California condor (Gymnogyps californianus) is the one of the largest vultures in the world, and its wingspan of approximately three meters is the widest of any North American bird. California condors nest in cliffs and feed exclusively on carrion. As a result of shooting, lead poisoning from carcasses, poisoning, pesticides, and other anthropogenic impacts, condor populations declined to near-extinction in the 1980s. In 1987, the 22 remaining wild condors were captured and brought into a captive breeding program at the San Diego Safari Park, the Los Angeles Zoo, and the World Center for Birds of Prey. The success of these intensive captive breeding efforts led to more than 100 eggs being laid by captive condors by 1994. Following captive propagation, condors were released in California, Arizona, and Sierra San Pedro de Martir National Park in Baja, Mexico. Today, there are around 430 condors with over 200 free-ranging in the wild. Condors reintroduced to their former range in Baja Mexico by San Diego Zoo Global are telemetered with solar-powered global positioning system (GPS) receivers so that they can be tracked at high-resolution (< 20m) for long periods to date (some for over eight years). These data enable condor managers and scientists to monitor the spatial behaviors and habitat use of the wild birds and enhances the conservation management of this iconic species. The restoration of California condors is a great conservation success story. However, condor populations still face serious challenges to their recovery, including lead poisoning and the emerging threat of impacts from renewable energy developments.

In a collaborative effort, the San Diego Zoo Institute for Conservation Research, the U.S. Geological Survey Western Ecological Research Center, and the San Diego Supercomputer Center developed and implemented 3D movement-based kernel density estimator (3D MKDE) [1] to help understand condor spatial behaviors, habitat use, and the risks they encounter in their environment. In this novel method, a trivariate normal kernel with time-dependent variance is integrated over time along the animal movement path, interpolated between observed locations, to produce the 3D MKDE.

Here, we visualize the animation of 3D MKDEs within a two-day moving time window for a pair of condors. The pair was introduced into Baja, Mexico in 2003, and have been tracked discontinuously via GPS since 2007. Condor locations were captured every hour from 5:00 a.m. to 8:00 p.m., when the birds were active. In late 2010, the pair was recaptured and quarantined to be treated for lead poisoning, and were re-released in early 2011. In 2011, the pair reproduced for the first time. In this animation, we show the 99%, 95%, 75% and 50% isosurfaces of the 3D MKDEs and the interpolated move path for each condor (female – orange, male – blue). Frames with no isosurface indicate missing data or lack of sufficient data for MKDE computation. We can see that these condors often traveled together in addition to making independent flights.Most of their activity is centered on the mountain range near their nesting site and where food was provisioned by conservation managers, but occasionally these condors made longer exploratory movements. During the breeding seasons, the home ranges of this pair intersected closely as the birds often flew together during courtship, mating and chick rearing activities. This is the first time that a 3D movement-based utilization distribution for wildlife has been animated over time.


Visualization Movies:
http://visservices.sdsc.edu/projects/zoo/movies/condor_pair_home_range_data_only.mov
http://visservices.sdsc.edu/projects/zoo/movies/condor_pair_home_range_mosiac_data_only.mov

Reference:
1. J. Tracy, J. Sheppard, G. Lockwood, A. Chourasia, M. Tatineni, R. Fisher, R. Sinkovits. Efficient 3D Movement-Based Kernel Density Estimator and Application to Wildlife Ecology. To be presented at XSEDE 14 Conference, Atlanta, GA

 

Plasma Dynamics and Confinement

Greg Foss, Anne Bowen, Wendell Horton and Lee Leonard

The rotation of the earth creates a Coriolis force that makes what would be a converging or diverging motions in the atmosphere into a vortex. The structures of these waves and vortices is of interest because they significantly influence global atmospheric circulation. In intense forms, they create the storms and seed the atmosphere for tornadoes and hurricanes. The physics of the vortex structure continues to heights of 100 kilometers and beyond where solar radiation ionizes the nitrogen and oxygen molecules in the air, making the gas plasma. These plasma structures can scatter the RF signals used in global navigation systems. The study of the generation and dynamics of planetary waves that are induced by this force in the ionospheric plasma has accordingly been a subject of a great deal of theoretical and experimental investigations in recent years. Dr. Wendell Horton's research group at the University of Texas at Austin, Institute for Fusion Studies, investigates the dynamics of plasma vortices using the  XSEDE funded Stampede petascale supercomputer. Data from their simulations are being visualized by the Texas Advanced Computing Center's vis team, allowing the researchers to see the full three-dimensional structure and dynamics of these vortices for the first time. The submitted movie shows the progression of visualizations that enabled the science team to analyze the overall dynamics of the system, and explore some of the visualization techniques available.

 

Streamwise Vorticity in a Supercell Thunderstorm Simulation

Greg Foss, Amy Mcgovern, Brittany Dahl, Greg Abram and Sean Cunningham

The animation and accompanying images compare variables from a high resolution simulated thunderstorm, a storm that spawned a tornado. These graphics illustrate how vorticity in the environment aligns with the wind feeding into a storm, enhancing storm rotation.
The visualization offers insight into the inner workings of the storm and can reveal more complex patterns and relationships between atmospheric variables that may herald the eventual development of a tornado.

The plot shows wind velocity and vorticity, a measure of instantaneous rotation. Wind is shown as grey streamlines with red representing higher speed values; gold streamlines are vortex lines modeled from vorticity. The scene illustrates rotation in the updraft, a region of wind blowing upward, and these strong rotating updrafts are a hallmark characteristic of a supercell. The main feature of the illustration is a vortex ring at the top which forms as the strong updraft punches into the stable stratosphere and the air subsequently curls downward. A translucent volume rendering of cloud water combined with ice provides reference to the surrounding storm structure. The visualization was created with VisIt software (LLNL) on Longhorn at the Texas Advanced Computing Center, University of Texas at Austin. Simulation data was produced with Kraken, NICS.

Animation and images available at:
https://www.dropbox.com/home/XSEDE14%20Visualization%20Showcase%20submissions

 

Spherical Visualization: Mapping the 2014 Winter Olympics

Karla Vega, Tassie Gniady, Patrick Beard and Eric Wernert

Students and staff at the Advanced Visualization Laboratory at Indiana University created this visualization to explore data analysis and visualization techniques of major social global events. For this visualization, the effort was focused on the Sochi 2014 Winter Olympics. The data was obtained through several online resources, as well as data scrapped from social networks. The major purpose of this project was to gain a better understanding of how to interact with these types of data, engage the digital humanities community, and to create visualizations for spherical displays, in particular for the National Oceanic and Atmospheric Administration's Science On a Sphere (SOS) system. The outcome was a series of software tools developed to handle the techniques that are shown in this animation, as well as a series of visualizations that were distributed to the SOS community.

 

Visual Computing of Large Dental Imaging Data for Investigative Caries Studies

Hui Zhang, Chauncey Frend, Guangchen Ruan, Michael Boyles, Eric Wernert and Masatoshi Ando

This visualization presents 3-dimensional complete tooth (enamel, dentin and pulp) and volumetric caries lesion data corresponding to multiple threshold levels from a research study aimed at quantifying caries lesion activity in human enamel.

Caries lesion activity assessment has been a routine diagnostic procedure in caries management, traditionally employing subjective measurements incorporating visual and tactile inspections. Recently, advances in 2D/3D image processing and analysis methods and microfocus x-ray computerized tomography (m-CT) hardware, along with increased power of high performance computing (HPC), have created a synergic effect that is revolutionizing the dental computing field. This visualization produces 3-dimensional complete tooth (enamel, dentin and pulp) models from m-CT data of 9 human tooth specimens, and renders the loss of mineral content at multiple threshold levels in each tooth specimen. Our visualization has seen early use in helping answer domain-specific questions such as quantitative assessment of dynamic carious lesion activities, visual representation of tooth mineral distributions, and the mapping and correlation of caries phenomena exhibited in volume and surface imaging data.

Throughout this project and in preparation for this visualization, a combination of XSEDE and non-XSEDE resources were utilized. The majority of the computational time was spent extracting contours of various dental structures and identifying the cross-section of demineralized enamel. To ensure fast turnaround time of this task, large dental imaging data are processed and interested structures (e.g., geometries that represent dental crown, pulp and dentin surface) are extracted offline with MapReduce tasks deployed at Stampede cluster machine. The geometric models were imported into either ParaView where data and render servers (also hosted on Stampede) facilitated both interactive visual exploration (i.e. qualitative analysis) and quantitative analysis or Maya for the creation of non-interactive visualization and animation.

 

Information will be posted in the coming weeks, please check back regularly and follow XSEDE on Twitter (@XSEDEscience) and on Facebook (Facebook.com/XSEDEscience).