[View PDF]


Berk Geveci (Kitware), Jim Ahrens (LANL)


  • To understand the 3D evolution of tearing modes (causing plasma instabilities) in simulations of magnetic reconnection
  • Challenges
    • Large data size generated by high resolution simulation on 98304 cores
    • 6.4 billion cells,1.5 trillion particles, 57 TB data
    • Only remote access to the supercomputer
    • Lack of dedicated visualization resources

Two isosurfaces showing the structure of particle density (blue) and current density (red).

Two isosurfaces showing the structure of particle density (blue) and current density (red).


  • Interactive 3D visualization of simulation data
    • Particle data and Mesh data
  • Comparison with theoretical expectations
  • Rapid exploration due to limited availability of supercomputer to run large simulations
  • Scaling ParaView: general purpose data analysis and visualization tool focused on large data


  • Allowed scientists to remotely analyze and visualize their data when it is not possible to copy locally
  • Allowed scientists “to rapidly explore the grid data to understand the 3D evolution of magnetic reconnection”
  • As expected, a spectrum of tearing instabilities develops which interact, forming new current sheets and triggering secondary tearing instabilities



Simulation of magnetic reconnection by Bill Daughton (LANL) and Homa Karimabadi (UCSD)

The reason we color the particle density with current density is as follows:

1) particle density isosurface shows the flux ropes nicely.
2) we proposed and was able to verify that the flux ropes form in regions where current density intensifies. particularly illuminating is the 2D cross section that is also shown in the same 3D figure.  you see regions of high current density (red) and that is where flux ropes are formed. there is also a second type of generation mechanism of flux ropes due to flow vortex generation. there is also evidence of such flux rope generation in the reconnection exhaust (i.e., in the two blobs on each side).