SCIBERQUEST

HIGH INTERACTIVITY VISUALIZATION SOFTWARE FOR LARGE COMPUTATIONAL DATA SETS EXISTING SCIENTIFIC VISUALIZATION TOOLS HAVE SPECIFIC LIMITATIONS FOR LARGE SCALE SCIENTIFIC DATA SETS. OF THESE FOUR LIMITATIONS CAN BE SEEN AS PARAMOUNT: (I) MEMORY MANAGEMENT, (II) REMOTE VISUALIZATION, (III) INTERACTIVITY, AND (IV) SPECIFICITY. IN PHASE I, WE PROPOSED AND SUCCESSFULLY DEVELOPED A PROTOTYPE OF A COLLECTION OF COMPUTER TOOLS AND LIBRARIES CALLED SCIVIZ THAT OVERCOME THESE LIMITATIONS AND ENABLE RESEARCHERS TO VISUALIZE LARGE SCALE DATA SETS (GREATER THAN 200 GIGABYTES) ON HPC RESOURCES REMOTELY FROM THEIR WORKSTATIONS AT INTERACTIVE RATES. A KEY ELEMENT OF OUR TECHNOLOGY IS THE STACK ORIENTED RATHER THAN A FRAMEWORK DRIVEN APPROACH WHICH ALLOWS IT TO INTEROPERATE WITH COMMON EXISTING SCIENTIFIC VISUALIZATION SOFTWARE THEREBY ELIMINATING THE NEED FOR THE USER TO SWITCH AND LEARN NEW SOFTWARE. THE RESULT IS A VERSATILE 3D VISUALIZATION CAPABILITY THAT WILL SIGNIFICANTLY DECREASE THE TIME TO KNOWLEDGE DISCOVERY FROM LARGE, COMPLEX DATA SETS.

STUDIES OF RECONNECTION IN THE SOLAR WIND HAVE REVEALED SOME SURPRISING RESULTS THAT ARE POSING SERIOUS CHALLENGES TO CURRENT THEORETICAL UNDERSTANDING OF THE RECONNECTION PROCESS. OUR PROPOSED WORK WILL FOCUS ON ANSWERING THE IMPORTANT SCIENCE QUESTION OF WHY DO RECONNECTION EVENTS IN THE SOLAR WIND EXHIBIT SUCH DRASTICALLY DIFFERENT PROPERTIES COMPARED TO PRESENT UNDERSTANDING OF RECONNECTION OR EVENTS OBSERVED IN OTHER ENVIRONMENTS, SUCH AS THE EARTH'S MAGNETOSPHERE OR IN THE SOLAR ATMOSPHERE? TO THIS END, THE OBJECTIVE OF THIS 3-YEAR PROPOSAL IS TO CONDUCT STATE-OF-THE-ART KINETIC SIMULATIONS FOR PARAMETER REGIMES RELEVANT TO THE OBSERVED SOLAR WIND RECONNECTION EVENTS. THE SIMULATION RESULTS WILL BE COMPARED AGAINST COMBINED PLASMA AND MAGNETIC FIELD DATA FROM NASA'S ACE AND WIND MISSIONS. AS SUCH, OUR PROPOSED RESEARCH IS HIGHLY RELEVANT TO THE STRATEGIC GOALS OF NASA S OVERALL MISSION, ITS STRATEGIC SUB GOAL 3B, AND ITS RESEARCH 'BJECTIVE 3B.1 AND IS APPROPRIATE FOR THE SOLAR AND HELIOSPHERIC SR&T. WE EMPHASIZE THAT THE LIMIT OF LOW MAGNETIC SHEAR IS ONE OF THE LEAST STUDIED AND POORLY UNDERSTOOD REGIMES IN COLLISIONLESS RECONNECTION. THIS LACK OF THEORY HAS MADE IT DIFFICULT TO INTERPRET THE OBSERVATIONS. THE PROPOSED SIMULATIONS WILL ADDRESS THIS ISSUE AND PROVIDE A FRAMEWORK FOR INTERPRETATION AND ANALYSIS OF DATA. OUR PRELIMINARY RESULTS HAVE ALREADY PROVIDED INSIGHT AND POSSIBLE EXPLANATIONS OF SOLAR WIND RECONNECTION EVENTS. OUR PROPOSED WORK IS STRUCTURED TO DEVELOP THESE IDEAS FURTHER AND PROVIDE DIAGNOSTICS THAT CAN TEST THEIR VIABILITY IN THE SOLAR WIND DATA. FINALLY, THE PARTICULAR SIGNIFICANCE OF THE SOLAR WIND MEASUREMENTS IS THAT THEY OPEN UP A NEW, AND IN MANY WAYS UNPARALLELED, LABORATORY FOR STUDYING THE LARGE-SCALE PROPERTIES OF RECONNECTION IN SPACE PLASMAS. FURTHER INVESTIGATION OF RECONNECTION WILL NOT ONLY YIELD UNDERSTANDING ABOUT ITS PROPERTIES IN THE SOLAR WIND, BUT WILL HAVE BROAD IMPLICATIONS REGARDING RECONNECTION AS IT OCCURS IN THE MAGNETOSPHERE, IN THE SOLAR ATMOSPHERE AND IN OTHER PLASMA ENVIRONMENTS.

WITH THE INFORMATION AGE HAS COME A DRAMATIC INCREASE IN OUR ABILITY TO GENERATE AND COLLECT DATA WHILE THE TRADITIONAL TECHNIQUES TO ANALYZE THIS TSUNAMI OF DATA HAVE PROVEN WOEFULLY INADEQUATE. THE AMOUNT OF DATA COLLECTED AND STORED ELECTRONICALLY IS DOUBLING EVERY THREE YEARS. MANY RESEARCHERS ARE SIMPLY OVERWHELMED BY THE AMOUNT OF DATA. SCIENTIFIC VISUALIZATION REMAINS THE DOMINANT FORM OF DATA EXPLORATION BUT CURRENT TOOLS ARE FOCUSED ON IMAGE CREATION AND LACK THE MUCH NEEDED EMBEDDED ANALYSIS CAPABILITIES. TO ADDRESS THIS URGENT NEED, WE PROPOSE TO DEVELOP AND INTRODUCE A FIRST-OF-ITS KIND TOOL KIT CALLED SCIVIZ2. SCIVIZ2 BRINGS TOGETHER KEY INNOVATIONS IN THREE SEPARATE FIELDS OF SCIENTIFIC VISUALIZATION, DATA MINING, AND COMPUTER VISION TO OFFER AN INTEGRATED SOLUTION FOR PHYSICS MINING OF THE MOST COMPLEX OF DATA SETS; MULTI-DIMENSIONAL, MULTI-VARIATE DATA SETS. SCIVIZ2 BRIDGES THE EXISTING GAP IN SOFTWARE TECHNOLOGY BY PROVIDING A SCALABLE KNOWLEDGE DISCOVERY TOOL THAT COMBINES THE POWER OF MACHINE LEARNING TECHNIQUES WITH SCIENTIFIC VISUALIZATION. OUR PHASE I GOAL IS TO DEVELOP A PROTOTYPE AND DEMONSTRATE THE VIABILITY OF THE SOLUTION BY APPLYING IT TO ONE OF THE MOST CHALLENGING DATA ANALYSIS PROBLEM, THAT IS THE DATA FROM PETA-SCALE PARTICLE SIMULATIONS. THE UNDERLYING PHYSICS INVOLVES COMPLEX INTERACTION OF MULTI-VARIATES AND THE OUTPUT INVOLVES BOTH GRID-BASED AND PARTICLE PHASE SPACE DATA WITH NEARLY TRILLION PARTICLES AND OVER 200 TB OF DATA FROM A SINGLE RUN. PHASE II ACTIVITY WOULD INVOLVE FULL DEVELOPMENT OF SCIVIZ2 AND ITS DISSEMINATION TO THE WIDER COMMUNITY. OUR CLOSE COLLABORATION WITH NASA'S HPC GROUP WILL BE HEAVILY LEVERAGED TO TEST AND RELEASE THE SOFTWARE FOR OPEN USE.