Hi Christoph, - We need both interpolated densities and currents. I just showed you the densities as an example.
- Yes, indeed n=9. Now that you mention it, I'm confused about the parameters relwidth and n. What's the difference between the two? - I added you to the code repo on github: Run using (in the following sequence): time python dump_system.py time python solve.py time python densities_and_currents.py The parameters for the run are set in params.py. The numbers currently set will produce a 5000x5000 site model. The first script will dump the system to disk, the second will compute the scattering states, and the third will load the system and the scattering states and compute the densities and the currents. Code blocks to produce the plots are in PostProcessingTMF.ipynb (slightly disorganized - please let me know if you need help here). Cheers, Mani On Thu, Oct 28, 2021 at 7:06 PM Christoph Groth <[email protected]> wrote: > Hello again, > > I noticed two problems with my reply. > > • I was speaking about currents while you interpolate densities. (This > does not matter much.) > > • I was assuming that your output field is 5000^2 while it’s rather > (20*9)^2 because you keep the n parameter to interpolate_density at > its default value of 9. Can you verify this? > > The second point modifies my analysis quite a lot in that the kernel > size is only 9*9 and not 250*250. Now it’s something like 32000 CPU > cycles per kernel evaluation, which seems horrible. > > Could you send me a minimal script that reproduces your problem, > i.e. that interpolates some (whatever) density with the same parameters > that you use? Then I can check things directly instead of guessing. > > Thanks > Christoph >
