Dear Steven G. Johnson, Thank you very much for your answer, I will try to set a non-uniform resolution. Currently I am using the instance c4.8xlarge of the amazon cluster. Is it possible to upgrade to a higher instance? It is the highest instance I see in the AWS marketplace but I wonder if there Is some way of getting more power.
Best regards, Juan. From: Steven G. Johnson<mailto:stevenj....@gmail.com> Sent: Thursday, November 26, 2020 3:44 PM To: Juan Ramón D<mailto:deop...@hotmail.com> Cc: meep-discuss@ab-initio.mit.edu<mailto:meep-discuss@ab-initio.mit.edu> Subject: Re: [Meep-discuss] RAM issues You could try using a coordinate transformation to simulate a non-uniform resolution, to increase resolution in your metal layer: https://meep.readthedocs.io/en/latest/FAQ/#does-meep-support-grids-with-non-uniform-discretization Plasmonic effects are pretty challenging for FDTD because of the resolution requirements, in general — I often switch to finite-element or boundary-element methods in that case. On Nov 26, 2020, at 7:22 AM, Juan Ramón D <deop...@hotmail.com<mailto:deop...@hotmail.com>> wrote: Dear meep users, I am trying to simulate the reflectance spectrum of a Stacked complementary “Plasmo – Photonic cristal” as described in the following paper: https://pubs.acs.org/doi/10.1021/nl504755d. I am using a cell of size 0.4 x 0.7 x 12 (microns). I have managed to get acceptable results with a resolution of 200 points /micron. However, the results are not fully converged (there is a thin 42nm gold layer which might be causing trouble). When I try to use a higher resolution, e.g. 250 points/micron, my code stops after defining the geometry: mpirun noticed that process rank 8 with PID ---- on node ------ exited on signal 9 (Killed). I think this is a problem of RAM because it only happens for big cells / high resolutions I have used a computer with 32Gb RAM and the amazon cluster, instance c4.4xlarge with the same result. I want to try with 64Gb amazon cluster instance, but anyway it won’t be posible to increase too much the resolution, because the RAM memory depends on res^3. Is there any way I can reduce the RAM usage for such a simulation, to get a higher resolution? Could it be because of a different reason? I can only think of reducing the cell (PML layers) and using non-dispersive models for the materials, but both methods compromise the results of my simulation. Thank you very much. Best regards, Juan. _______________________________________________ meep-discuss mailing list meep-discuss@ab-initio.mit.edu<mailto:meep-discuss@ab-initio.mit.edu> http://ab-initio.mit.edu/cgi-bin/mailman/listinfo/meep-discuss
_______________________________________________ meep-discuss mailing list meep-discuss@ab-initio.mit.edu http://ab-initio.mit.edu/cgi-bin/mailman/listinfo/meep-discuss
