Learning Forum

[Fatemeh Karami]

Hello,

 

I am having a problem with running codes using the inverse design method in Lumerical. I want to use this method to optimize an MMI structure (like the Y_branch example, https://optics.ansys.com/hc/en-us/articles/360042305274-Inverse-design-of-y-branch) by using 400 iterations. I used 3D FDTD solver to optimizs a structure 10 microns long. Some other parameters are: dx = 1e-09, pgtol = 1e-09, ftol = 1e-09, scaling_factor = 1e9, mesh size_y = 30nm, and mesh size_x = 50nm (changing the mesh size does not affect the problem).

 

When I run the code, it gets stuck and freezes after around 30 iterations. It also runs some iterations more than 8 times. Is this a problem with the code or with my system?

 

My system information is: cores: 64, Logical processors: 128, Physical memory: 256 GB.

 

I would appreciate it if you could help me.

 

Thank you,

Fatemeh

[Guilin Sun]

In such case, please check your license and know how many processes it allows, and check the memory DURING the simulation. Most often during the simulation it consumes more momery. If the memory is not sufficient it will page the hardware which is why it freezes since it is too slow.

[Fatemeh Karami]

Thank you for your support.

how can I check the license for processes?

 

[Guilin Sun]

PLease refer to https://optics.ansys.com/hc/en-us/articles/360052724713-List-of-licensed-features-by-product

you may need to reach your account manager for details of your license.

[Fatemeh Karami]

Hello,

Thank you for your suggestion. I followed your instructions, but the issue still persists.

Could you please provide any other recommendations? Alternatively, can I share my code privately for you to check?

Thank you for your assistance.

[Guilin Sun]

Somehow my reply was not successful so I duplicate it again:

since we are not allowed to check files on the forum, please contact your account manager for the files.

Did you check the "task manager" during the simulation and see if the memory is the bottleneck? Even though one simulation does not need a lot of memory iteration may need.

 without error message it is hard for us to find a solution.

You may first try 2D simulation, or if you have VarFDTD try it first.

 

 

 

[Fatemeh Karami]

 

Thank you for your response.

 

The problem is that it progresses well until the 14th iteration (increasing transmission from 25% to 80%), but afterward, during the “making forward solve” process, it either takes more than a day or freezes. The size of the design area is 75 microns.

 

I checked the Task Manager; please see the attached photo.

 

Thank you.