University:
Boston University
Course:
CAS PY 105 | Physics 1
Academic year:

2020
Views:

433

Pages:

19
Author:

Ryland Whitehead

One Problem with PIC Particle noise from limited numbers of particles  Random walk statistics:  Example n=144 particles/cell -> sn=8.3%  Fixes:  Nature reduces this through electrostatic shielding  Use non-point particles  Use millions and millions of particles  Use super computers! s  n particles/ cell Boundary Conditions (BC)  Simulations of all types require BC  BC introduce limitations and, sometimes, error  Example: Periodic is the simplest BC     The right side connects to the left The top to the bottom Particles leaving the Left reenter on the Right and visa versa Particles leaving the top -> bottom … Boundary Conditions Cause Example in 1D Limitations Periodic boundaries quantize the simulation:  Only a full wave or  Integer multiples allowed  Simulations must not focus on waves spanning the system Other BC have other issues True in fluid simulators as well PIC Code… // Read parameters from the input file: // Main timestep loop: for (it = it0; it <= nt; it++) { // Apply the standard leapfrog method infile(argv[1]); // Initialize the dynamic variables : init_misc(); init_particles(pic, w, misc); init_fluid(fspecie, pic); init_field(Efield, rho); // Calculate the charges and currents on the grid. charges(rho, pic, fspecie, 0); // Find the electric field on the grid at t=n: efield(Efield, rho); // Output any initial diagnostics: output (argv[1], pic, fspecie, Efield, rho, misc, w, it); leapadv_subcycle(pic, fspecie, rho, Efield, w, misc); //Deal with any Boundary condition issues boundary(pic, Efield, w, misc, it); // Output data, diagnostics and restart: output (argv[1], pic, fspecie, Efield, rho, misc, w, it); } }// End of main timestep loop Charges.cc & density.cc void charges(FArrayND &rho, particle *pic, fluid *fspecie, int it) { rho = 0.; void density(FArrayND &den, int id, particle *pic, fluid *fspecie, FTYPE scaler) { gather(den, INDICIES(pic[id].x, pic[id].y, pic[id].z), scaler*pic[id].n0); for (int id=0; idCs, streaming instability develops Modified two-stream or Farley-Buneman Instability E0 direction (m) Electrojet PIC Simulation New thing learned: • Saturation though Mode coupling • Saturated wave speed • Average Tilting of Wave • Thermal Behavior ExB direction (m) Meteor Plasma waves Large Aperture Radar Detection of a Meteor ALTAIR meteor detection Altitude (km) Leonids picture from the shuttle Time (s) Particle in Cell Simulations of Meteor Plasma 3 nplasma isocontours B nplasma cross-sectional plane Particle in Cell Simulations of Meteor Plasma – with a wind B Conclusions  New 3D Simulations  Enables exploration of Meteor Evolution  Future: Spectra to connect to observations Conclusions  Simulations enable us to explore nonlinear systems  Simulations subject to systematic limitations and numerical errors  Enable us to better understand our:  devices,  Models, and  Nature.  Future Simulation Work:  Better Algorithms  More Parallel Efficiency  Vast array of applications!