TrackParticles has different fusion neutron weights at low energies about smilei HOT 10 CLOSED

Please provide some more details on how your diagnostics where implemented and post-processed. As the two types of diagnostics process the same particles, they should match.

from smilei.

ParticleBinning diagnostic:

def angles:
return np.arctan2(particles.py, particles.px)
DiagParticleBinning(
name = 'neutron angle',
deposited_quantity = "weight",
every = every,
species = ["neutron"],
axes = [
[angles , -1*math.pi, math.pi, 360],
["ekin", 0., 50, 100, "edge_inclusive"]
]
)

And then to get spectrum I extracted the data and integrated across the appropriate angular bins. With TrackParticles, I output all variables and create a similar histogram with scipy.stats.binned_statistic_dd, something like:

stats_data = stats.binned_statistic_dd([Theta, En], values = wn, statistic = 'sum', bins=[theta_bins, energy_bins])

where Theta and energy were calculated from the tracked Px and Py components, and wn is the macroparticle weights.

from smilei.

My guess is that your method with tracked particles divides by different bin sizes than that done in happi for particle binning. As you are using a custom axis, it cannot know the geometry of your binning. It's just a guess

from smilei.

Bin sizes are the same for both tracking and particlebinning, so all energies would be affected equally when dividing by bin size, not just those <2.5 MeV. I cannot see anything strange with the weights, their values are broadly consistent above and below 2.5MeV, just that there are a significant number of tracked neutrons in that low energy range, perhaps that isn't being counted by the binning diag? Or their weights are calculated differently between the two diagnostics.

from smilei.

I have no clue what is going on here. The particle binning really just does a histogram and divides by the bin size which is indeed constant in your case.

Just in case you may want to check the raw data in the hdf5 file.

from smilei.

Just an idea that crossed my mind. You wrote "Theta and energy were calculated from the tracked Px and Py components". But there should be Pz as well for calculating the energy. Did you take it into account?

Another detail: what is the geometry, and what are boundary conditions?

from smilei.

Geometry is 2D, so I didn't output Pz component. Boundaries are silver-muller.

from smilei.

2d means x and y, but there is definitely a contribution from px.

I meant boundary conditions for particles

from smilei.

Ah ok, I can rerun the simulation outputting Pz components as well and test it out. Boundaries were open.

from smilei.

Hi, including the Pz component has resolved the issue, thanks!

from smilei.