My code run for several minutes and stopped

[kundi-physics]

[Status-Mirror]

Hi @kundi-physics,

Could you copy and paste an input deck to the issue? If you surround your deck with ```

``` Like this ```

GitHub will format it correctly

Like this

Cheers, Stuart

[kundi-physics]

begin:constant

  # laser parameter
  la = 0.8 * micron
  omega = 2 * pi * c / la
  a0 = 120
  dl=10*la/c
  w0=3*la
  xr=pi*w0*w0/la
  xf0=50*la  # focus distance from left
  t0=3*dl

  #coordinate
  r = sqrt(y^2+z^2)

  # target parameter
  ro = 39.5*la

  den0 = 0.03 * critical(omega)
  x0 = 50*la
  ll = 200*la
  x1 = x0 + ll
  sigma_x1 = 5*la    # jet length corresponding to ... laser wavelength
  profile_0 = gauss(x, x0, sigma_x1)
  sigma_x2 = 5*la   # jet length corresponding to ... laser wavelength
  profile_1 = gauss(x, x1, sigma_x2)

end:constant

begin:control
  nx = 3600
  ny = 1600
  nz = 1600

  # final time of simulation
  t_end = (450 * la+ll)/c

  # size of domain
  x_min = 0
  x_max = 180 * la
  y_min = -40 * la
  y_max = -y_min
  z_min = -40  * la
  z_max = -z_min
  particle_tstart = t0
end:control

begin:boundaries
  bc_x_min = simple_laser
  bc_x_max = simple_laser
  bc_y_min = open
  bc_y_max = open
  bc_z_min = open
  bc_z_max = open
end:boundaries

begin:window
  move_window = T
  window_start_time = 165 * la/c + t0
  bc_x_min_after_move = simple_outflow
  bc_x_max_after_move = simple_outflow
  window_v_x = (1.0-den0/critical(omega)) * c
end:window

begin:species
  name = proton
  charge = 1.0
  mass = 1836.15
  npart_per_cell = 4
  density = if((x lt x0), den0*profile_0, den0)
  density = if(x gt x1, den0*profile_1, density(proton))
  density = if(r gt ro, 0.0 , density(proton))
  density = if(x gt 300*la, 0.0, density(proton))
end:species

begin:species
  name = chlorine
  charge = 17.0
  mass = 67384.7
  npart_per_cell = 1
  density = density(proton)
end:species

begin:species
  name = electron
  charge = -1.0
  mass = 1.0
  npart_per_cell = 4
  density = 18*density(proton)
end:species

begin:laser
  boundary = x_min
  intensity_w_cm2 = a0*a0*1.37e18 /la*micron/la*micron
  lambda = la
  phase= 0 + pi*r_yz*r_yz*(xf0)/(xr*xr+(xf0)*(xf0))/la-atan((xf0)/xr)
  pol_angle = 0    # p-pol = 0.0    s-pol = pi/2
  profile = gauss(r_yz,0,w0*sqrt(1+(xf0)*(xf0)/xr/xr))/sqrt(1+(xf0)*(xf0)/xr/xr)
  t_profile = gauss(time,t0,dl)
  t_start=0
  t_end=2*t0
end:laser

begin:subset
  name = check
  gamma_min = 1.005
  include_species:proton
end:subset

begin:subset
  name = check1
  gamma_min = 1.005
  include_species:electron
end:subset

begin:output
  name = o1
  file_prefix = a
  # number of timesteps between output dumps
  dt_snapshot = 10 * la/c
  time_start = t0 + ll/c

  # Properties at particle positions
  particles = check+check1
  px = check+check1
  py = check+check1
  pz = check+check1
  particle_weight = check+check1
  spinx = check
  spiny = check
  spinz = check
 end:output

begin:output
  name = o2
  file_prefix = b
  # number of timesteps between output dumps
  dt_snapshot = 10 * la/c
  time_start = t0 + ll/c

 # Properties on grid
  grid = always
  ex = always + single
  ey = always + single
  ez = always + single
  bx = always + single
  by = always + single
  bz = always + single
  charge_density = always + no_sum + species + single

end:output

[Status-Mirror]

Hi @kundi-physics,

This simulation is too large for me to run for testing. I was able to run the deck by removing the particles and changing the domain to

  nx = 360
  ny = 180
  nz = 180

so I don't think there's a problem with your deck itself. I notice you have the key particle_tstart = t0 - does your deck "stop" when particles start moving? A large 3D simulation like this is incredibly expensive to run when the particles move, so I'm wondering if your code seemed to stop when the particle-pushing calculations began.

My advice would be to retry in a 2D simulation first, before moving to 3D. I think 64 cores may not be enough for the simulation you present here.

Hope this helps, Stuart

[kundi-physics]

Dear @Status-Mirror, I have tried a 2D simulation before this. 64 cores are used in a 2D simulation, and it's exactly not enough for a 3D simulation. So I tried to use 192 cores in my new 3D simulation, but it seems to be the same as before. Maybe 192 cores are still not enough for my 3D simulation, I will try to use more cores.

Next screenshots are "output" and "slurm" respectively in my 3D simulation which used 192 cores.

Thank you for your solution! kundi

[TomGoffrey]

You are using nz = 1600. Assuming you ran the 2D simulation with the same values of nx and ny your simulation will be 1600 times larger in 3D, so you will need a significant increase in the number of cores you are using.

[kundi-physics]

@TomGoffrey Yes, you are right. When I increase the number of cores from 192 to 480, the code run. Thank you for your suggestion!

[TomGoffrey]

Good to hear. I'll close the issue now, but should you run into further problems please feel free to re-open or start a new issue.