Picongpu: Temperature setting is correct in 0.4.0-rc1

[PrometheusPi]

After some strange simulation results and an initial offline discussion with @ax3l the question was, whether the temperature setting is still correct.
This question was also mentioned/raised in #2687.

I analyzed the simulations and it looks like setting the temperatures is correct.
Below you can find the results.
Particles have different masses, drift velocities and temperatures.
The transparent colors represent the simulation data (histograms from the macro-particle momentum) while the solid lines represent the expected distribution from the Maxwell-Boltzmann distribution.

equal mass (electron), T_0=50 eV, T_1 = 30 eV, + drifts

equal mass (electron), T_0=50 eV, T_1 = 10 eV, + drifts

equal mass (electron), T_0=50 eV, T_1 = 5 eV, + drifts

different masses (ions), T_0=50 eV, T_1 = 30 eV, + drifts

different masses (ions), T_0=50 eV, T_1 = 10 eV, + drifts

different masses (ions), T_0=50 eV, T_1 = 5 eV, + drifts

(the x-axis represents the velocity normalized to the speed of light, the y-axis is the normalized distribution probability)

[PrometheusPi]

This issue was just opened for future references after discussion it with @psychocoderHPC.
Since there is no error in the current implementation, I will close this issue.

[ax3l]

Excellent, thank you for testing and documenting the result. :+1:
You tested in 0.4.0-rc2, right?

[PrometheusPi]

No, not quite, I tested it with 0.4.0-rc1.

[ax3l]

That's ok as well, rc1 fixed the RNG on CPU, rc2 does contain fixes that do not affect your test: #2686

[PrometheusPi]

@ax3l Did you mean the 6f172c1e8 Rng: add std::random_device gen fix from alpaka?

[ax3l]

yep