Closed YCAOsz closed 4 years ago
Hi, Could you provide the whole namelist within a file that we can run ? Here, some characters are missing.
Julien
This is the python file (just change the .txt to .py would work) If I change the density function to constant(), it would run longer, but still stops at some point. Also, if I extract the envelope (Envabs), it gets blended into the field and the wakefield wasn't generated some time into the simulation. What should I modify to get a good 1d simulation of laser wakefield? Thank you (by the way is there a forum to discuss these issues of smiles?) Smilei_envelope_111519.txt
Hello, I see that your timestep may be too high for an envelope simulation. The Courant-Friedrichs-Levy condition of the envelope solver is even more restrictive than the one of the "standard" Yee solver. Your envelope values maybe are increasing exponentially, blocking the simulation at some point. Normally I use a timestep dt=0.8*dx, could you try it?
With this modification I seem to arrive to a longer distance, with a well-formed wakefield:
PS since you are not interested in the ion motion, you can comment the ion Species block, the Esirkepov method will take care of adding an implicit layer of neutralising ions (see the section "Why ions are not present" in this tutorial https://smileipic.github.io/tutorials/advanced_wakefield_electron_beam.html)
Thank you. The simulation can last to about 3cm in real unit before the envelope blows up. However, is it possible to make the simulation last even longer (to about 6cm)? Would it be better to use a gaussian laser wave? (though I'm not sure if i can put the laser at some offset in x coordinate so that I don't need to start at x_min and time the window movement.)
Following the envelope evolution, what do you mean that it blows up? If it increases in an uncontrolled way, maybe you should decrease even more the dt, like 0.7*dx
or 0.6*dx
.
If you are seeing strange effects at the borders, since the envelope has only reflective boundary conditions, probably you will need a larger window.
Besides, you are using a dx that is very small (dx = 0.153
in code units), which should be used for a standard laser simulation where you need to sample the laser high frequency oscillations. With the envelope model you can increase it, to sample only the envelope of the laser (see Figure in http://www.maisondelasimulation.fr/smilei/laser_envelope.html) and reduce your simulation time.
In your case, your laser fwhm in field (not intensity fwhm) is laser_fwhm = 129
in code units (~21 optical cycles), so you can increase your dx
even up to dx = 1
in code units. This way, your simulation will be much faster. If however you think that you will have to resolve finer envelope modulation effects (e.g. due to depletion), then you should use a dx
smaller than 1.0
@YCAOsz Did you try the recommendations from @Z10Frank ?
@YCAOsz In absence of answer I'll close the issue, please feel free to reopen it if you have news about it.
Hi. I experienced a problem using smilei for 1d laser wakefield simulation. When I used the script below for a long 1d simulation, the process only produces about 25 outputs and shows this error:
I'm not sure why this happens. Maybe the density function? (The script is shown below) Thank you