Dear prof. Steven Johnson and meep users:
I am reporting a problem of field blowing out. This may be a similar problem to the reports of Andika Asyuda in September or Jefferson Thomas in July. As I do not see their solutions, this problem may be a limitation of current meep applications. I am attaching a ctl file which is one of those I have tested. Its output is too big (351 KB) and I cannot attach it here. Please teach me how to attach the big file. In the attached ctl file, I attempt to calculate the transmission flux spectra of two layers of close packed gold spheres whose diameters are 28 nm. The 3-dimensional lattice is made up of 230 nm x 230 nm in X and Y, and 958 nm in Z. Along the Z-direction, 200 nm PML, 250 nm free space, 58 nm gold sphere layer, 250 nm free space, 200 nm PML are positioned. Periodic boundary conditions are placed for all X, Y, Z directions, so the gold sphere layer is like a 2-dimensional film. Close to the upper PML, an E-field source is placed which is supposed to generate a plane wave propagating along Z. Close to the lower PML, a transmission detector is placed to collect the flux. In this ctl file, not only transmission spectrum but also integrated E square can be calculated at a fixed wavelength (632.8 nm) by switching the comment lines. In this particular ctl file, the calculation needs almost 25 GB of memory, but I have calculated for similar gold particle layers in various sizes, which have demanded memories of 4 GB to 40 GB. When the gold sphere layer was small (e.g. crystalline packing with 9 spheres in a reduced lattice, 90 nm x 102 nm in X and Y), all calculations were perfect. (I was able to get good results when the gold sphere layers were increased up to 6.) The transmission flux converged to 0.2% of the peak value in less than 6 time unit. I was able to integrate E square at a single wavelength (632.8 nm) between 6 – 12 time units. But calculation when the gold spheres were placed in expanded lattice (like the attached ctl file), the transmitted flux began to increase after 4 or 5 time units and blows out. The E square integration at the single wavelength showed that field blew out after 8 time units or so. Over the last couple of months, I have calculated this kind of calculation for so many different gold layers. In general, as the number of gold particles is increased in a layer or the number of layers is increased, the problem of field blowing out becomes worse. From my experience with meep over the last several months, I cannot find any mistakes in my ctl file formulation and I cannot think of any physical origin of this field blow-out. Should we consider any mathematical bug or limitation of meep? Seong Kyu Kim, professor Department of Chemistry Sungkyunkwan University Suwon, 440-746 Korea
layer2-spec.ctl
Description: Binary data
_______________________________________________ meep-discuss mailing list meep-discuss@ab-initio.mit.edu http://ab-initio.mit.edu/cgi-bin/mailman/listinfo/meep-discuss