Return to the PREVIOUS PAGEObjective: The purpose of this project is to develop hybrid plasma particle-in-cell (PIC) codes on parallel computers. The codes are useful for investigating kinetic plasma processes associated with global space plasma phenomena. The specific applications include magnetic reconnection processes responsible for magnetic sub-storms in the Earth's magnetosphere. Due to limitations in computer capabilities, particle modeling of magnetic reconnection phenomena has not been conducted.
Approach: The major difficulties for developing parallel PIC codes on parallel computers are the gather-scatter scheme and the load balancing problem among processors. The gather-scatter scheme arises from the need to communicate between particle quantities (like charge and current densities) and the grid quantities (like electric and magnetic fields). An efficient gather-scatter scheme is investigated to guarantee an efficient parallel PIC code. In order to achieve optimal performance on parallel computers, program development effort is also focused on distributing the computation load evenly among processors.
The particle-in-cell codes are developed by solving Maxwell equations together with particles to represent the charge and current carriers. The particles are advanced in time using the particle equations of motion with the self-consistently calculated electric and magnetic fields. We will use the hybrid PIC approach in which electrons are considered as fluid and ions are treated as particles. Using this approach, effects which are outside of the scope of the fluid description and essential to many space plasma physics problems can be calculated.
Accomplishments: Because the project has just been initiated in the summer of 1994, no major accomplishments have been made yet. We are currently improving the efficiency of gather-scatter algorithms for parallel PIC codes. In general the developed gather-scatter algorithm will improve the performance of simulation codes on parallel computers.
Significance: The developed parallel hybrid PIC codes on massively parallel computers will allow space scientists to model large scale space phenomena efficiently within a reasonable time. The developed codes will be used to investigate important energy dissipation processes due to ion dynamics during magnetic reconnection.
Status/Plans: Our immediate plan is to improve the efficiency of basic components in the parallel PIC codes. Currently we are investigating the performance of various gather/scatter scheme for parallel PIC codes. Later we will examine various load balance schemes to optimize the usage of processors and will develop a new sorting algorithm specific for PIC codes.
Point of Contact:
Return to the PREVIOUS PAGE