Getting started

To run PEPC you first need to create a run directory. This can be anywhere, but we will assume it is placed in the PEPC install directory ($PEPC), such as the example 'tutorial'.

Sample run scripts (.sh) and parameter files (.h) can be found here:

• billiards.h - Inter-particle forces switched off; reflective boundaries for various geometries
• clamp.h - Puts electrons in thermodynamic eqm using constant temperature dynamics
• eqm.h - Plasma sphere - energy conservation test
• ions.h - Creates ion 'crystal' using artificial lennard-jones potential
• laser.h - Ponderomotive laser heating
• wire.h - Laser interaction with wire target

Example parameter (.h) file

 &pepcdata
 ! particles
 ne = 12000
 ni = 12000
 
 plasma_config = 1
 
 ! set up plasma target
 target_geometry = 1 ! sphere 
 ! target_geometry =7 ! hollow sphere 
 ! target_geometry = 3 ! wire
 ! target_geometry = 0 ! rectangular slab
 
 ! physics stuff
 Te_keV = 0.5 ! Temperatures in keV
 Ti_keV =0.
 mass_ratio = 2000. ! Ion:electron mass ratio
 coulomb = .true. ! Compute Coulomb forces
 lenjones = .false. ! Compute Lennard-Jones forces
 bond_const = 2.e-3
 r_sphere = 10. ! Sphere radius/disc or wire diameter
 
 x_plasma = .1 ! plasma thickness
 y_plasma = 2. ! plasma width (slab target)
 z_plasma = 2. ! plasma width (slab) or wire length
 xl = 2 ! graphics box dimensions
 yl =2
 zl =2
 
 ! beam
 beam_config_in = 0 ! beam off
 ! beam_config = 1 ! fixed beam, initialised at start
 ! beam_config = 2 ! user-controlled particle source
 ! beam_config = 4 ! ponderomotive laser heating
 	
 vosc = 6.0 ! laser amplitude (vosc/c)
 omega = 0.5 ! laser frequency (omega/omega_p)
 sigma = 6. ! focal spot size
 tpulse = 20. ! pulse duration
 lambda = 1.0 ! wavelength in microns
 
 ! control
 nt =10 ! number of timesteps
 dt = 0.2 ! timestep
 eps = 2. ! Coulomb potential softening parameter
 theta = 0.3 ! multipole clumping parameter
 mac=0 ! tree-walk switch
 idump = 4000 ! particle dump frequency
 iprot=1 ! protocol frequency
 itrack=10 ! density tracking frequency
 particle_bcs = 1 ! boundary conditions for particles
 scheme = 1 ! integrator scheme
 ncpu_merge = 1 ! merge factor for restart
 debug_level = 2 ! protocol debug level
 debug_tree = 0 ! tree debug level
 restart = .false. ! restart switch
 vis_on = .false. ! visualisation switch
 ivis = 2 ! vis frequency for particles
 ivis_fields = 5000 ! vis frequency for fields
 ivis_domains = 5000 / ! vis frequency for tree boxes

This parameter file is first copied to run.h by the run script or job. See the User Guide for a more comprehensive list and description of input parameters. More complex examples can be found in the demos.

To execute the code on a Linux PC with mpich:

./eqm_linux.sh

For JUROPA use:

msub juropa.job

For JUGENE use:

llsubmit eqm.bgp

Output data

The output files will be stored either in the run directory or in the subdirectories dumps/ fields/ log/ etc. The most important of these are:

• energy.dat

Kinetic and potential energies etc., expressed in keV per particle. 9 y-columns in ASCII format, containing the following:

• omega_p t - normalised time
• U_pot - electrostatic potential energy
• U_mag - magnetic energy (not yet implemented)
• U_kin-e - electron kinetic energy
• U_kin-i - ion kinetic energy
• U_beam - beam energy
• U_tot - total energy
• I_pond - laser intensity
• x_c - position of critical density

• run.out Printed diagnostics/protocol
• load_TTT.dat Shows approx load balance amoung CPUs at timestep TTT

• Particles

Particle data is output independently by each CPU to avoid memory and MPI bottlenecks for large runs, and can be found in:

dumps/parts_pNNNN.TTTTTT
dumps/info_pNNNN.TTTTTT

Currently the format of the particle dump is a 15-column ASCII file (13 reals, 2 integers) with the following content:

x, y, z, px, py, pz, q, m, Ex, Ey, Ez, pot, owner, label

The number of particles written out together with other data is contained in the associated info file. Each subdirectory pNNNN contains data for task number NNNN at the checkpoint timestamps TTTTTT, whose frequency is controlled by the input parameter idump. Data for each task can be merged for postprocessing with the script bin/merge1_dump, for example:

merge1_dump 000100

will generate will create 2 new files in the subdirectory dumps in the same format as the partial dumps containing the complete particle data at time 000100:

dumps/parts.000100
dumps/info.000100

These can either be used by a postprocessor or as an initial configuration for a new run.

• Fields

Gridded data created during the run is placed in the fields directory.