upic-emma-2.0

PSTAD Branch

(December 13, 2018 -- make a version for PSTAD solver.)

This is the repo for UPIC-EMMA 2.0, a UPIC-EMMA based on UPIC 2.0 by Viktor Decyk (which includes MPI/OpenMP parallelization).

New features include:

• non-square cells
• FDTD operators for benchmarking
• PSATD operator for boosted-frame simulation
• additional boundary conditions for fields and particles to allow the study of laser plasma interactions at different intensities.
• OpenPMD support

Upcoming features include:

• binary collision
• Galilean transformation
• OpenACC/OpenMP with offloading support

It also supports Doxygen documentation, to use it, type:

doxygen Doxyfile

Input parameters:

• N_threads (integer): number of OpenMP threads

• FDTD (integer): FDTD=1 -> finite difference operator of N-th order (N is even), or 0 -> use traditional difference opreator ( i k )

• cfl (real): timestep in terms of the CFL condition, if cfl=0.95 then dt is 0.95 * CFL limit

• laserpulse (logical): if laserpulse = .TRUE. then a laser is initialized

• relativistic (logical): if relativistic = .TRUE., then relativity is turned on for orbit integration

• moving_ions (logical): if moving_ions = .TRUE., then ions are mobile and have mass, if .FALSE., then ions are treated as immobile fluid

• Delta_x, Delta_y = spatial cells dimension in units of (Delta)

• indx, indy = spatial dimensions of the simulation box
  along the x and y axis. Pay attention :
  The number of gridpoints Nx and Ny
  along x and y axis must be a power of 2 due
  to the FFT algorithm. We impose so :
  Nx = 2^indx and Ny = 2^indy where
  indx = 1 + floor(ln(Lx/Delta_x)/ln(2)) and indy = 1 + floor(ln(Ly/Delta_y)/ln(2)) such that Lx = 2^indx and Ly = 2^indy For example, with Delta = 1. : Lx = 2, 4, 16 , 32, 64, 128, 256, 512, 1024,
  2048, 4096, 8192, 16384, 32768, 65536,
  131072, ... with
  indx = 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, ... respectively

• t_end = time at end of simulation, in units of
  (/ omega_p) or (/ omega_0) depending on units

Plasma properties (useful only if plasma = 1) :

• atomic_weight = plasma atomic weight A (g/mol) #

• atomic_number = plasma atomic number Z () #

• ionization_state = plasma ionization state Z* () #

• ax/ay = smoothed particle size in x/y direction #

• den_me = Number of electrons per macro electron #

• Npic = Number of macro electrons per spatial cell #

• param = vTe / c with c the velocity of light in vacuum #

• ptx, pty, ptz = root mean square electron momentum (relativistic=1) or velocity (relativistic = 0) components in the x, components in the x, y and z directions resp.
  in units of ( me ) Delta omega_p
  Be careful, since Te = (Te,x + Te,y + Te,z) / 3, vTe = sqrt( (vtx^2 + vty^2 + vtz^2) / 3 )
  must be respected

• px0, py0, pz0 = mean plasma momentum (relativistic=1) or velocity
  (relativistic = 0) components in the x, y and z
  directions resp.

Laser pulse properties (useful only if laser = 1) :

• propdir = 1 (x) or 2 (y) = main propagation direction # (MPI parallelization is done along the y-direction) # #

• polardir = polarisation parameter such that -1. <= polardir < 1. : # _ polardir = 0. corresponds to a linear polarization # along : _ z (propdir = 1) # _ x (propdir = 2) # _ polardir = - 1. correspond to a linear polarization # along : _ y (propdir = 1) # _ z (propdir = 2) # _ polardir = +/- 0.5 correspond to a circular polarization # _ polardir = else correspond to elliptic polarizations # polardir > 0 -> clockwise # polardir < 0 -> counterclockwise #

• shape = 1 (Gaussian wave shape) or 0 (Plane wave) #

• theta = angle of incidence relative to the 'propdir'-axis # assuming that shape = 1 (for a plane wave, rotate the # plasma!!!) # Be careful : theta must be between -45. and 45. so # #

• tlaunch = time at which the laser pulse starts to be launched # in units of omega^-1 #

• FWHMt = time duration Full Width at Half Maximum of the laser # pulse envelop in units of omega^-1 # #

• xfocal/yfocal = focal point coordinates along x and y in units of # (Delta) for Gaussian waves #

• FWHMs = laser pulse spot size Full Width at Half Maximum at the # focal point in units of (Delta) for Gaussian waves # (the laser pulse waist at focal point : # W0 = FWHMx / sqrt(4ln2) ) # #

• omega0 = Laser pulse angular frequency in units of (omega) # -> omega0 = 1. is imposed if units = LPI # #

• a0 = Maximum amplitude of the laser pulse potential vector # in units of (me Delta omega / e) # -> E0 = omega0 * a0 is the maximum amplitude of the # laser pulse electric/magnetic field # #

Boundary conditions

• BCx = 1 (periodic) or 0 (absorbing) # condition along the x-axis for both particles and fields # #
• BCy = 1 (periodic) or 0 (absorbing) # condition along the y-axis for both particles and fields # #
• For absorbing cond., the Perfectly Matched Layer (PML) technique # is used for the E-M fields (Berenger, 1996) : # _ PML_scheme = 0 : Yee scheme # PML_scheme = 1 : Implicit scheme # _ L_PML is the PML layer thickness in units # of Delta # _ n_cond_PML is the integer n such that the PML# electrical and magnetic conductivity read : # sigma(x) = sigma_m * (|x-x_I|/L_PML)^n # where x_I is the position of the interface # between the PML and the effective sim. box. # 3 < n_cond_PML < 4 is optimal # #

Diagnostics :

• Delta_t_diag = results are saved in text files in results/ # every Delta_t_diag in unit of (omega_p or omega0)^-1 #

density profile :

• density_x/y, write down expressions for density along x/y # directions. The names of coordinate of x is x1 # and y is x2. many math functions and branching # function are supported. # #

Revision History:

• 1/8/2018: Initial README (F. S. Tsung)
• 11/19/2020: Adding documentation for input deck