Based on the structure and the refractive index contrast you will have you are correct to post your question in the OptiFDTD forum group as FDTD is the appropriate application for this type of device. We have a 30 day evaluation that you can download and see how it works for your application.
The evaluations are available under the Downloads menu at the top of the page.
You will need to provide more information in order for me to help you with this as I was able to load up and simulate your design. Can you please provide me with the following information:
– What version of the software are you using?
– Screenshots of any windows open while simulating, particularly ones showing progress.
The AWG cannot be modeled in one single simulation due to distance and curvature of the waveguides. Each segment must be modeled with the phase and amplitude from the previous section used in the input plane information. There should be another sample AWG_Star.bpd which is the output star coupler from your example generated from that script. Please review the notes in that example to understand what is being done.
You have not attached an updated design file. I already provided feedback on the one you attached earlier. Do keep in mind that without having a maintenance license, I am assuming you are using the 32-bit, my support is necessarily limited to helping with issues in the product. We are not able to provide design assistance.
Please provide specifics on the issue you have, i.e. how is your results not the desired results, and I can see what I can do.
One thing to note is that with the Fourier transforms on a the Gaussian pulse, the further you go away from the central wavelength the less accurate your results are. Your source is 0.55 um and you are running Fourier transforms from 0.3 to 1.1 um. With the 64-bit product I only find the value going over one if I move away from the central wavelength.
Based on the information you have made available it would appear that you are dealing with one of two situations:
1. You are running into a bug that is specific to the 32-bit free product which is out-dated and not currently maintained.
2. You are looking at values that are to far from your central wavelength based on your pulse width.
The reason you are not getting access to the full 4 Gb is possibly a limitation within your specific operating system and without heavy diagnostics I would not be able to tell you.
Based on Windows 10 documentation a 64-bit product can access anywhere between 128 Gb and 6 Tb depending on the version of windows you are using. Most Windows 10 systems our users are running are either Home (128 Gb) or Pro (2 Tb).
First let me just clarify your statement that you have 10 Gb of free memory. Are you referring to RAM or space on your hard drive? I know you are referring to memory (RAM) but I just want to ensure we are all on the same page.
Second you are in fact using a 32 bit product where 32-bit systems are in fact limited to 4 Gb of memory allocation regardless of how much you have available on your machine. On some systems this maximum limit can be even lower depending on how the operating system handles memory allocation. It would appear you are hitting the limit of a 32-bit system. Might I suggest you try the 30-day evaluation of the 64-bit product?
I strongly recommend that you watch the introductory webinars and look through some of the examples that come with your installation. There are a number of issues and unfortunately I do not have the resources to do one on one training.
Answers to your questions:
1. You do not need a profile for each layer the one profile is sufficient, it should be the cross section of your profile. I am not certain what you mean by stubs but your design has one waveguide and the design in the paper is a cross pattern. Use the linear waveguides and place them where you see the features in the paper.
2. In the design you attached you have a z-directed input plane and not an x. Additionally you have it set to Gaussian not modal for the transverse profile. Do keep in mind that the input plane must cross the waveguide in order to calculate a mode.
Something for you to be aware of:
1. Your resolution (set in the 3D simulations item within the project browser) is far too high. The auto setting is based on your source wavelength to ensure it is stable. However, you have feature sizes (in your case layer thicknesses) as low as 0.01. Your resolution must be such that the structure is properly discretized.